08:57:15 Good morning everyone.
08:57:37 Everybody else missed a small talk.
08:57:42 Yes, I do. I definitely do, it's affecting my brain just the background noise the chatter The, the coffee cups Clinton, whatever.
08:57:54 The darkness breakout rooms and allies for 10 or not right or just, you know, just kind of hear the background noise.
08:58:05 Yeah, I think it's always assume you don't care about non critical things right about not physics or anything like that it's just less.
08:58:16 Right, right.
08:58:20 The dark silence of zoom and blue jeans and it's kind of weird.
08:58:27 I don't think I'll ever get used to it.
08:58:35 But I guess the numbers in the USA, improving.
08:58:39 I think so. Lexi nation is really having an impact.
08:58:47 Yeah.
08:58:49 Looks like you'll be able to go to Germany and the summer young.
08:58:55 Well, I'm not so sure yet. Unfortunately I hope so. But in Germany, the numbers are less good.
08:59:01 Yeah, but they said that they would open it up to people that have been vaccinated and you've been vaccinated. I've been vaccinated and my parents will be by end of May.
08:59:13 Yeah.
08:59:15 So at least I can visit my parents but I think the university will still be closed. And I mean I can go to PSI at least Switzerland. The problem is that all states around Germany are opening up, despite high numbers, including Switzerland.
08:59:34 So it's not clear what had happened.
08:59:36 I think there will be a false faith in Europe.
08:59:45 I don't know why everybody thinks that the researchers, don't know what they are doing, and do something against better judgment, but okay.
08:59:59 That's interesting. I didn't realize the closed caption was on so I'm kept, it's pretty, pretty works pretty good.
09:00:07 Yeah, but it can produce some funny results, especially for non native speakers.
09:00:12 I got to get that, like, back, back, back, it gets a lot of things wrong with saying, Canadian accent, I'm sure.
09:00:21 Yeah, rhythms just not refined yet.
09:00:27 And wait till you see how they translate j loud.
09:00:33 I have to say, Jay.
09:00:44 I have to, I have to say Google Hangouts is much better business, I mean everything else is worse, but the transcription from Google Hangouts is almost perfect even like hot words, it gets pretty good.
09:00:54 Up to try that feature. We houses phone system at Biennale right where you can just say that oh please say the name of the person you want to speak with that every once in a while I say my own name as they connect me to some random person.
09:01:10 Yeah, I don't think it's going to rock for me.
09:01:14 Oh, almost
09:01:17 no rock for you.
09:01:23 Okay.
09:01:24 Well, should we get started, it's one minute past, we've got 20 people on 21 people on.
09:01:34 So let me start.
09:01:38 Welcome everyone to the eighth workshop on streaming rido.
09:01:47 First off, I'd like to just come on give me.
09:01:53 me advance the screen.
09:01:56 There we go.
09:01:57 So, the workshop is divided into four sessions I'll go over them a little bit more detail but basically today we're having experimental updates, and for past, present and future experiments.
09:02:16 Thursday morning. We have a fairly packed session on hardware and vendors.
09:02:25 And then Thursday afternoon.
09:02:28 There's a session on computing and data rates Friday morning is data acquisition and Friday afternoon as a discussion on the future. And I'd like to start off by thanking all the speakers who have agreed to come out and give us talks on what they're doing
09:02:49 what they, their plans are for the
09:02:53 future streaming readout system and, and in particular, I'd like to thank the convener is to have organized all the sessions Yon Chris gin and Martin all done an excellent job, and I think it'll be a really good workshop, so just let me
09:03:16 escape from that and go over to the,
09:03:27 the timetable.
09:03:29 So, like I said, we will have
09:03:38 presentations on detectors that are in existence now or have been and what they've done to realize streaming readout and also on future detectors in Wednesday morning and afternoon.
09:03:56 Then Thursday morning we have a quite an intense session with various hardware vendors like Kane and people interested in developing new a sick chips and other a six.
09:04:18 Thursday afternoon, we'll be talking about computing and data rates.
09:04:24 Again, another fairly intense session.
09:04:28 And then Friday morning talking about data acquisition systems.
09:04:34 So I think it's a good, good morning. Friday afternoon we're going to have a basically a discussion session, regarding how we see, or how various people, Thomas over, and roll event view how things are going to proceed.
09:04:59 Clearly for an e commerce send me answers, and loss of joy. Okay.
09:05:11 Sorry, what was that.
09:05:14 So, Thomas will not be there but he certainly answers to my questions, and sorry I'm intermediate and, and I'm able also join us.
09:05:24 We.
09:05:26 Okay. Okay, good. So I think this is very important because the IC has got a very tight schedule.
09:05:37 And it will be will be very interested in how they plan to realize this what funding profile they anticipate or hope for. And also on what detector is
09:05:54 the either a first detector, or and wager a second detector or two detectives at the beginning I think there's a lot of interesting questions and comments that we can sort of raised on Friday afternoon, but to go back and start this session, we have.
09:06:16 Professor john Xuan from Iowa State University who is going to talk about one of the hopeful detectors for the EIZXJ. And what his plans are for Ajay.
09:06:44 So, john if you're ready you can share your screen and, or if you want I can display, I think, I think it should work from here. Can you see my screen.
09:06:48 Yes. That looks great. Terrific.
09:06:51 All right so so thank you very much for giving me the opportunity to come and talk to you about the HA experiment today.
09:06:58 I should apologize right off for appearing in a tie, I know that real physicist don't take anybody seriously if they're speaking wearing a tie. But I'm also doing the DC nuclear physics physics day today so I have congressional staffer meetings after
09:07:11 this, and they don't take you seriously if you're wearing a T shirt, so it's sort of the best of both worlds.
09:07:18 So what I want to do today is tell you a little bit about the HA experiment. I'm not going to be able to talk a lot about streaming readout as it as it directly pertains to add che because obviously we're not that evolved yet, but I think it's very clear
09:07:31 that streaming readout is sort of a necessity for the EIC experiments.
09:07:38 So let me start with that a one on one and tell you a little bit about what what Ajay is the acronym stands for EIC comprehensive promo dynamics experiment.
09:07:49 and we're often asked how this should be pronounced.
09:07:53 You have a choice. If you prefer the Latin you can call it fk, which is means behold, if you prefer the Italian you can use fJ, which means exception.
09:08:05 We think either of them is nice. Either way, ha is 64 institutes and counting actually as of this morning we are 65 Institute, which are collaborating to design and he detector, which will offer full kinematic coverage and an optimized far forward detective
09:08:24 system. Now for reasons I'll discuss a lot in this talk. We're investigating a design which incorporates an existing 1.5 Tesla magnet.
09:08:32 And one of the foundational principles behind ha is given the tight timescales involved, we want to reduce cost and risk especially schedule risk to allow ha to be ready for first EIC detector operations and India we speak that's this critical decision
09:08:51 for a trade also wants to be very flexible and driven by the physics. So the detector design, which were involved in the design process right now is driven by the physics performance, and we're also investigating the costs and benefits associated with
09:09:08 citing the experiment at either of the two Rick interaction regions either interaction point six with a 25 million radian crossing angle or interaction point eight, which is currently envisioned to have a 35 minute radium crossing angle.
09:09:22 I'll talk a bit about the call for proposals in a moment but we plan to submit a proposal to be the project at detector and address the full complement of the gic science program.
09:09:35 In the context of this, we always want to make it clear that the people on Ajay are fully supportive of the community desire for two detectors, at the gic, both at IPA, and ipv6, and that's necessary to maximize the scientific output of the gic.
09:09:55 And so the best result of this proposal process is two or more outstanding proposals, which helped make the case that the community is not only interested in, but can support and can design and build to detectors for the gic.
09:10:12 And as part of that process, one of the foundational principles is that jazz that we're open to everyone in the community to participate. And in fact, we're open to people and institutions who wish to contribute to the other proposals.
09:10:25 As you'll see the design starts from the CIC reference detector. And so for many detector groups, it makes sense to investigate how their particular technology might perform in different detector designs
09:10:41 were organized we don't call ourselves a collaboration instead we call ourselves a consortium.
09:10:46 The reason being, we have decided that given the tight timescale for proposals. We have formed around a steering committee, which is or head from MIT, Tanya horn for corn from see you and myself are the institutions that have joined at a form and institutional
09:11:03 board, to whom we are responsible.
09:11:06 And we are organized around a set of teams.
09:11:10 And these teams are the basis of what we need in order to have a structure that can provide the work that's needed to do for the proposal.
09:11:21 So we have a computing team detector team, a physics benchmark team, and editorial team, which is focused specifically on preparing not just the proposal, but all of the supporting documentation that will go with the proposal.
09:11:37 And being a modern organization we felt it would be ridiculous not to build diversity, equity and inclusion in at the very foundation of what it is and so we have a diversity, equity inclusion team as well.
09:11:51 The loud is assigned us or the project is assigned us a point of contact that's wrong.
09:12:08 And I've listed some resources here if you're interested in more.
09:11:59 We have a website where we try to centralize all of our information, there are a set of mailing lists, in particular the public list will get you notices about important happenings and meetings, and then our Indigo agenda page is open and you can browse
09:12:16 past meetings if you're interested in seeing what's happening. We also try to record all of our meetings, so that collaborators in different time zones who aren't necessarily able to attend can at least view the recordings
09:12:31 of ha is growing, I just wanted to show you how hJ has changed in a very short period of time, as a February 26 at a was comprised of 46 institutions, 33% of which were Non Us institutions.
09:12:47 And as have just a few days ago.
09:12:50 We have grown since, since that time by about 40%.
09:13:02 Adding very significant new institutions and non us institutions, which brings significant new capabilities to the consortium, and we're currently about 40%.
09:13:05 Non Us institutions.
09:13:10 So before I go on and talk more about hJ specifically, let me talk a bit about the CIC call for proposals to put things in context.
09:13:20 So, a joint call for proposals was issued by Brookhaven National Lab and Jefferson Lab.
09:13:25 For detectors at the CIC with a deadline of December 1 2021.
09:13:31 That is a frighteningly short period of time in which to develop.
09:13:38 Excuse me, in which to develop call a proposal for a comprehensive collider experiment, let alone and experiment at the electron eye on collider which as we're learning is completely is really completely different from the kind of collider experiments
09:13:51 have been designed before the call envisioned two possible responses. One is a response with a proposal for what is called detector one detector one is within the scope of the project which means the budget, and the management lies within the gic project.
09:14:10 And it's envisioned that US federal funds will support most but not all of the acquisition of detector one detector two is currently not in the project scope and routes to make detector to to fund sector two and make a second interaction region possible
09:14:23 are being explored proposals were also given the flexibility to consider an alternate citing scenario. Currently detector one is planned to be at Rick interaction point six, and the machine design is based around that.
09:14:38 However, they were given the option of considering an alternative citing scenario, but also the corresponding responsibility to address the implications that that would have to the gic project.
09:14:51 The science between the two detectors is envisioned to be complimentary.
09:14:56 And of course, that is, and we really think that's an absolute necessity for the ice science program detector one is envisioned as being based off of the reference detector described by the end users group in the recently published yellow report.
09:15:11 And it must address the icy white paper and as report science case.
09:15:18 It has to achieve that by meeting the performance requirements described in these documents.
09:15:25 And most importantly, and this is critical for what I'm going to discuss with that a completion of detector construction must be achieved by what's called critical decision for a the start of the ice accelerator operations.
09:15:36 The early.
09:15:40 The early date for critical decision for a is July, 2030.
09:15:45 That might sound like a long time off.
09:15:46 But many of us have been involved in the construction of large detectors before.
09:15:53 And on the timescale that's usually required to build a large detector. That's frighteningly short detector two is a complimentary detector, which can build on the capabilities in the end is in the white paper in the ns report, but also extend that into
09:16:10 new areas, and it is envisioned to be completed two years later, so it has a longer time scale.
09:16:19 Okay, with that foundation. Let me describe the foundation of it.
09:16:25 As I said he wants to develop a detector capable of delivering the full EIC science mission, that is that is absolute statement number one. There is no compromise on that, that has to be done.
09:16:37 Having said that, we believe that an important first step in doing that is appropriate utilization and or upgrades of existing detectors and infrastructure.
09:16:48 And if this goes back to that timescale issue I mentioned by using existing infrastructure, which is perfectly appropriate to the science mission.
09:16:59 You can reduce technical and schedule risks, and you can then reinvest that savings in the detectors.
09:17:07 In fact, you can buy more and better detectors. If you save money on things that don't directly detect the particles that you're interested in.
09:17:17 So what kind of infrastructure, am I talking about, well, as you're aware as many of you are aware, there is a lot of existing infrastructure at Brookhaven National Lab, that's being built around the Phoenix experiment.
09:17:32 That's Phoenix there's a heavy I and experiment which is scheduled to start running in 2020 late 20 2020 2023.
09:17:40 It represents a $27 million investment and that's just the major item of equipment cost by the DMV there are significant costs that are also invested in infrastructure associated with the interaction point equally as equally as large as at 27 $9.
09:17:56 So that's a substantial investment. And part of that investment involves an existing 1.5 Tesla solenoid, as well as the magnet flux return and cradle for the experiment.
09:18:08 That's just steel, but it's expensive steel in terms of the design acquisition, a finite element analysis everything that's required engineering time that's required to put that together.
09:18:22 I mentioned also that there's some significant infrastructure and this is one of the reasons we're interested in looking at IPA IPA already for example will contain the cryogenic connection to Rick for the bar magnet, as well as significant updates and
09:18:37 racks mechanical safety systems electrical systems, etc. in order to support the US Phoenix, there's the potential to reuse and refurbish existing us Phoenix detectors as appropriate.
09:18:48 And so all of this makes up real physical infrastructure.
09:18:52 Another point that we've been making is there is a different kind of infrastructure, a personnel, infrastructure, if you will.
09:19:01 And that has to do with the considerable recent do a project experience that the consortium has, in terms of not only s Phoenix but also projects at the LHC.
09:19:14 And so that human capital or that human infrastructure is extremely important these detector projects are extremely complex anybody who's been involved with the do a critical decision process, knows how complicated that can be, and having people with
09:19:30 the experience and skill to manage that process. It reduces risk, you're not going to run into problems you have a lower risk of running into problems and concerns, it will slow you down.
09:19:41 Now let me talk a bit more about someone right.
09:19:43 The 1.5 Tesla Bab our solenoid was built in 1997, and sold out.
09:19:50 It was built by Pasquali fabric Qatari, who is you may also know as the designer of the CMS for Tesla superconducting magnet.
09:19:58 And in fact he's a recent recipient in 2020 of an I triple E award for contributions to the field of applied superconductivity.
09:20:06 This was the first magnet that Pasquali designed for superconducting magnetic designed. It's a very conservative design in fact he refers to it as built like a tank.
09:20:16 It's 3.7 meters long as a 1.4 meter border radius. It's designed for 1.5 testicle central field at five kilograms. So there's about 20 mega joules of stored energy.
09:20:27 It was transported out of storage at slack to Brookhaven in 2015 and has undergone successful low and high field tests, and it will be installed in the Phoenix apparatus.
09:20:40 Towards the end of this year, when we are actually building up now the outer the flux return and the cradle with the magnet installed.
09:20:49 Now, you might be concerned that a magnet, Donald would be considered for an experiment that might start running in 2030, the Jefferson Lab magnet engineers did an extensive risk analysis based off of analyzing the before the performance and events.
09:21:07 When it ran and in that bar, as well as the low field and high field test experience at Brookhaven and concluded that the magnet is actually in pretty good shape.
09:21:17 And with some refurbishment should prove adequate.
09:21:22 In terms of its ability to perform for an experiment at the CIC.
09:21:29 That's one form of infrastructure, I just wanted to take a moment to talk now about another form of infrastructure, infrastructure, which is something again people don't think about and that has to do with the infrastructure associated with the, with
09:21:44 the splice plates that actually hold the flux return together. Right, so the flux return is 32 sectors of 1.3 metric tons of steel. We instrument that to make the outer calorie monitor, and it's held together by a set of splice plates, which look like
09:22:03 these, which are these structures here, which are pinned and bolted into the sector is to hold them in place.
09:22:10 As you might imagine.
09:22:12 These splice plates are under enormous stress.
09:22:16 That means they need to be heat treated so that for example, the steel doesn't yield in these stress concentrations around the puck and pin locations that heat treating is a process that's very tricky, because you need to heat treat the material all the
09:22:29 way through and it's very thick.
09:22:33 Now at this point you're asking yourself I dialed in for a streaming read reader, read out workshop, why is this guy talking about steal. The reason I'm talking about steel is because this is the kind of hidden investment that I think it's important to
09:22:44 take advantage of.
09:22:46 We went through multiple rounds of prototyping this material before we were able to heat treated and machine it without cracking and demonstrate that the finished product actually achieved the necessary tensile strength.
09:23:00 That was eight plus months of engineering time of physicists time of interacting with potential vendors, and all of that reuse saves that engineering time so that you can use it elsewhere.
09:23:14 All right, let me talk to you a bit about what he might look like.
09:23:18 And let me start by first with this schematic diagram of the HA detector, while and etc reference detector so this diagram comes right from the EIC yellow report, and it's just a schematic diagram of what the reference detector would look like in terms
09:23:34 of the acceptance that it needs to have in different regions.
09:23:50 Out in the outer region is a chronic Kalorama tree in the central region that had drawn it Kellerman is backed up by electromagnetic color imagery particle ID and tracking that tracking could be split between say a TPC and silicon, or it could be also
09:23:54 silicon in the forward region, you need tracking in the forward hadron going region you need particle ID, you need electromagnetic color imagery and electromagnetic color imagery is especially important in the electron going region for a D is machine,
09:24:11 you must detect the electron with high fidelity.
09:24:14 I've listed the institutions in FA that are, that have expressed interest in developing these detector technologies and building these detectors, just to show you the breadth of our organization.
09:24:26 And also things that are not shown, are people interested in polarized Beeman polar imagery, electronics, and can trigger.
09:24:33 Now one of the interesting things about the ice detectors, of course, is it's not just the central detector, the detector has to be very carefully integrated with the experimental been lines and the reason for that is because a large part of the physics
09:24:48 program for example for exclusive processes, require that you actually measure of say the scattered proton.
09:24:57 And that requires that the beam line actually be instrumented and considered as part of the detector.
09:25:03 So, you need require far forward detectors in the hadron going region a forward spectrometer off momentum detectors Roman pots, as well as a zero degree calorie meter, and we have a number of Institute's that are interested in the physics and the detector
09:25:17 technologies associated with this. And in the backward region, since you're luminosity monitoring where you detect scattered electrons.
09:25:28 So that's a schematic of in terms of the acceptance. What might look like well that's what we're trying to work on now.
09:25:37 As part of this proposal process we are evolving into a fully realized design by evaluating and selecting different tip detector technologies that are appropriate to the physics mission.
09:25:49 This is a mock up of what ha might look like. Based off the outer flux return the magnet, and the pedestal, that we would reuse from s Phoenix.
09:26:04 And along with different detectors in different regions that perform the physics mission. So you have the former hadron arm you have the central region, and you have the electron going region.
09:26:16 In terms of detector technologies in the central region you require some sort of very good central tracking. And so options RT people select TPC plus silicon, or all silicon, or I will say that the community really seems very excited about all silicon
09:26:31 detector options.
09:26:33 Are you back that up with an electromagnetic Calla riveter simulating glass right now seems to be a favorite technology, you do need very good resolution in the central arm, as well as hadron p ID, and one of the challenges that actually is going to face
09:26:47 is because we start with a fixed aperture bore 1.4 meters in radius.
09:26:53 We don't have as much radio space as you might have is you design a new magnet.
09:26:59 And so our spaces of the other premium and so we're looking at things like a dark or maybe ACL gods or some combination to provide the particle ID in the central barrel, and then that's backed up by hydroponic color imagery provided by the instrument
09:27:12 in flux returned from us Phoenix.
09:27:15 In the hadron end cap tracking technologies are things like maps mpg DS STC spin back chambers hadron BPI D provided by D retro Tov eh separation Puffin era Joel possibly a TRD electromagnetic calor imagery, there's number of options for electromagnetic
09:27:35 color imagery and finally hedonic Kalorama tree where that drone at Keller imagery is most likely going to be instrumented with the flux are integrated with the flux return.
09:27:43 So iron simulator is sort of a leading option.
09:27:46 And in the electron and cap, again you need tracking, you need hadron particle ID here and enrich whoops, here and enrich is a favorite electromagnetic color imagery needs to be very good in this region and so something like lead tongue state in the integration
09:28:02 and simulating glass and the outer regions as a possibility.
09:28:05 And the hydraulic Keller emitter options here are a little bit, a little bit easier you could consider instrument thing the magnet steel the the plug door magnet steel or again maybe an iron simulator option, or maybe you have a separate flux return and
09:28:20 you use lead simulator, as a hydroponic catcher.
09:28:26 In terms of physics studies, of course, he must be able to demonstrate that it can provide the physics in the full National Academy of Sciences report.
09:28:35 And we plan to do that by doing physics simulations, and I'll talk about those in a bit, that address the full range of the yellow report physics, things like the origin of the nuclei on spin where we look at the alll measured in polarized scattering
09:28:54 confined motion of particles where you look at semi inclusive deep and elastic scattering and the stress spin structure of the jet that originates from those events, 3d imaging of quarks and gluons these are exclusive properties that allow you to look
09:29:09 in detail at the structure of the proton.
09:29:14 The origin of nuclear on mass, things like pie on k on structure.
09:29:19 High glue on densities and nuclei, where you're looking for saturation effects and nuclear PDF effects in the nucleus. And so we're planning a set of simulations that are targeted towards demonstrating the capability of Ajay to perform, sort of this basic
09:29:35 physics.
09:29:38 At the same time, we're looking at studies which show unique Ajay strengths. As I mentioned, ha is really has a group of people who are very interested in the far forward.
09:29:47 So they're interested in things like light ion tagging let's say in Ed collisions, where one can tag either the proton or the neutron so that you know what the electron interacted with.
09:29:59 That's very useful for example for JC production and to practice scattering.
09:30:04 You can look at pi on k on structure, again, a an exclusive process where you need to have very good accepted acceptance in the forward region to fractal jets, especially the, well actually, the entirety of the acceptance will turn out to be a very good
09:30:21 Jeff detector and things like nuclear modifications and in medium resin evolution, where the ability to reconstruct D amp D star. In fact, we may have some favorable aspects based off the lower field, and I'll talk about that in a moment.
09:30:38 The field, of course, is an issue.
09:30:42 We think it makes perfect sense to say we need to, it makes sense to start by examining the possibilities with a key piece of infrastructure that exists, but we're not blind to the limitations of one of those limitations, is the fact that in a solenoid
09:30:59 field, as you go to higher pseudo rapidity, the shape of the field is simply not conducive to good momentum resolution at high pseudo rapidity, the field components just don't give you the bend of that you're looking for.
09:31:13 One option of course is to just crank up the magnitude of the field. In order to recover this momentum resolution.
09:31:20 That's certainly possible, and in this region.
09:31:23 Justin unmodified 1.5 Tesla field does certainly worse than a 3.0 test the field.
09:31:30 The track of momentum resolution, that's required by the yellow report physics and the jets and heavy flavor section is really quite tight and you want very good momentum resolution in this 2.5 to 3.5 reason that's important for Jetson heavy flavor it's
09:31:52 One of the things that we're investigating and this was studied in an earlier incarnation are options for shaping the field with a passive field piston shaper around the beam pipe.
09:32:02 Now with the crossing angle the beam pipe is much more complicated.
09:32:05 Nevertheless, we think that we can improve this substantially with some passive field shaping and we're taking a look at doing that.
09:32:14 However, with every with every issue. There's always an opportunity, and the opportunity arises because the lower field can also be very useful in tagging and reconstruction of low momentum particles.
09:32:30 for example, for D amp D star marathons, where you need to catch a very low momentum, Python, in order to reconstruct the D. and the minimum PT cut off is much lower and the 1.5 Tesla field than it is in a three tests.
09:32:46 That's a real balance because you have to balance.
09:32:50 Enter you have to balance momentum resolution against increased acceptance and winching Fang recently did a very nice study. It's a fast simulation study, based on the yellow report specifications, but she took a look at the reconstruction of heavy flavor
09:33:06 to case.
09:33:09 Using a one point for Tesla field and a three tests the field. And one example here is the decay of a lambda C.
09:33:14 Right.
09:33:15 And you look at this and you say, ah, three Tesla, clearly, right, much better momentum resolution and that's absolutely true.
09:33:23 On the other hand, the one point for Tesla has better acceptance and so your yield is actually higher.
09:33:32 But that yield is spread out over a wider, a wider range in mass and so you take a hit in terms of subtracting the background. And so what she did is she did a very nice study as a function of different as a function of the pseudo rapidity and PT coverage
09:33:52 and also as a function of different momentum, different particle ID assumptions. And this now this is the lambda see this is for DTK pie. And what you see here is the statistical error projected with the one she used to one point for Tesla field but the
09:34:10 magnet and the three Tesla. And what you see is the difference is maybe not as Stark as you might have thought it would be.
09:34:14 In fact, the 1.4 Tesla and the one three Tesla actually provide comparable performance in terms of the statistical air and that's the interplay between the increased acceptance and the improve resolution.
09:34:28 I mentioned simulations and so I wanted to show you where we are with simulations actually leverage is the full fun for all framework that's already used extensively by us Phoenix.
09:34:38 And so we already have a complete implementation of a prototype ha detector.
09:34:46 Within full jG for simulation framework.
09:34:49 And this is just a demonstration of that detector. Now, that detector is not what we would consider a baseline or a reference yet so for example it includes a TPC and we want to be able to examine silicon tracking options, but nevertheless it demonstrates
09:35:02 the capability. And this is a single EP event 18 on to 75. It incorporates the 25 million radian crossing angle already, so this would be an ipv6 configuration.
09:35:14 And this is just an event display where you can clearly see the scattered electron and the jet originating from the deep inelastic scattering event. Here's a forward on view where you can see that jet more clearly.
09:35:26 And you can also the beam see the beam remnants propagating through the beam pipe with the 25 milligrams and crossing angle that detector is aligned along the electron beam which is one along this portion of the beam pipe here.
09:35:38 So we already have an extensive simulation framework in place to be able to do both the detector performance studies and the physics performance studies that are necessary for the proposal.
09:35:50 And our first simulation campaign is actually starting here in a week or so.
09:35:56 So, finally let me just conclude by just reminding you the consortium is now 65 institutions, planning, design the ice project detector based around the bar solenoid.
09:36:08 Our focus is to really start by focusing on the end game, which is providing an ice detector by CD for a ready for the start of machine operations.
09:36:20 And that means we think the right place to start is with a careful examination of how we can reduce risk and cost at the outset, reducing risk primarily scheduled risk is important, and reduced cost, doesn't mean we necessarily intend to be less expensive,
09:36:38 but we have more resources available to spend on the detectors and detector technologies, which means we may be able to be more aggressive in certain detector technologies and by better performance.
09:36:52 The physics program will span the entirety of that outlined in the study in the yellow report, and the debt detector design processes be done in tandem with a wide range of full physics simulations.
09:37:03 We also do have DAC, and computing groups. And as part of a simulation, they will be sketching out the data acquisition requirements recce, and that will then will evolve more this discussion of streaming readout for.
09:37:19 So I listed on our website there if you're interested, we keep that updated with ongoing information, and I'm happy to take any questions.
09:37:29 Sorry much for that. Very interesting talk.
09:37:33 I'll open it up for questions. We've got still quite a bit of time from according to the schedule. So, either raise your hand or just speak out I don't know how chaotic I will be if you just unmute yourself and ask your questions.
09:38:01 JOHN. This is Chris Cuevas I'm, I guess I'm on the edge a block diagram there too and we've never met a course in person.
09:38:12 You seem very concerned about the schedule and wanted to ask you just a quick, a quick one on contingency How does do we view.
09:38:21 You know extensions or, or, you know, for both detectors or just, I mean how strict is this I mean are they going to look at these things, just not aware of that we had we went to the same thing with 12 Jeff upgrade and things are pretty you know they
09:38:36 they had to reschedule a few things but well you should so I would say that, so I'll give you my answer. Right, but that's a good question for Jim, in the sense that
09:38:49 it is a good thing that the project itself actually includes a detector.
09:39:08 Right.
09:38:58 It can also be a constraining thing, because for the project to hit its milestones, the detector must also hit its milestones.
09:39:08 Right. It makes sense. So it's not so it's not, it would be it would be considered a failure for a $2.5 billion project. If the accelerator ready.
09:39:20 But the detector missed CDA for a deadline. And that's important.
09:39:26 Right, that's extremely important for how it looks to upper levels of doe how it looks to Congress, how it sets the tone for future large projects in physics.
09:39:38 So that's a constraint, I mean I realized I hit on schedule a lot but that's a constraint that you have to think about, right, because the project, and the project detector are tied at the hip.
09:39:49 That, that's something you just have to keep in mind.
09:39:53 Okay, thank you.
09:39:57 Any other questions.
09:40:03 Okay. Perhaps I can ask one, um, you know I'm fJ and EIC is, you know, past my cell by date. So one of the things, However, I was active with the hero experiment, namely Zeus, basically for its whole lifetime.
09:40:28 And one of the problems we had was, well, there were two problems. One is when heroes started running at the very beginning, it didn't meet its luminosity goal.
09:40:40 It was, we had very little luminosity in 1992.
09:40:45 And that was viewed quite.
09:40:48 I've spoken to other people about this for EIC and they say that it'd be an absolute disaster. Do you have any comment on that. Do you, do we need to have physics, good quality luminosity right from the beginning or, then we wait a year or two.
09:41:10 So that depends, that depends on what you call good quality luminosity. Right.
09:41:17 If I, if I go back to it my own experience with Rick, as well as So Rick Rick was fun, because I still remember sitting in the accounting house, and the machine was running, and everybody's trying to figure out if we've seen a collision.
09:41:30 There are lots of particles in the detector.
09:41:33 Right. And it's all been gas and Halo right and you're trying to figure out, did we actually find a collision in this mess.
09:41:40 So turning on a new collider is is complicated.
09:41:44 In addition, the NIC as I understand from the machine people is going to need a significant amount of time just to clean itself up just running right just to bake out the detector components bake it out with beam before it can start to provide clean beam.
09:42:02 It's going to be a complicated, it's probably one of the most complicated machines ever built. It's going to be complicated for the accelerator people to commission the machine and bringing bringing collisions.
09:42:14 And so one of the, one of the things that we're thinking about, is because we have a number of reusable detector components, you could imagine that as a service to the machine.
09:42:26 Let's say we don't begin by installing our large silicon expensive silicon tracker, that could potentially be damaged.
09:42:34 But we install the Phoenix DPC.
09:42:38 And we use that to help the machine commission.
09:42:44 Then once the machine is beginning to come up to operation. During the shutdown, we install the central tracking array. So there are a number of thoughts on with within hJ on how we can assist the detector, or excuse me, the accelerator in meeting its
09:43:04 goals.
09:43:05 Right. And I think when I talked to the machine people, particularly in terms of establishing collisions. You really have to have some interaction with the detector that's there.
09:43:15 And maybe you don't want that to be, you know, you're super expensive find fancy detector in the very beginning.
09:43:23 Does that answer your question, I kind of talked about a little bit but but i think i think that's a great question.
09:43:29 There are lots of issues associated with how you bring this machine on that people are just now beginning to sort of struggle with.
09:43:41 Ya know i i think i agree that having a low luminosity started out, you know, helps you debug things to tune your electronics, etc.
09:43:53 So I think that's, as long as it doesn't go on too long. That's right. That's right. I'm worried about that transition from is this luminosity.
09:44:04 There's a lot of junk in my detector too. Oh okay this is low luminosity. And then ramping the luminosity Oh, yeah.
09:44:14 The other thing that we had when we installed Zeus and towards the last later years, when Hera upgraded its luminosity was trying to upgrade the detector, and it really hadn't had an upgrade designed into the base model.
09:44:37 We had to worry about.
09:44:40 Back scattering from synchrotron radiation and things like this.
09:44:45 It required an extensive detector modification to, to allow it to handle the higher luminosity that hero eventually got to do you see that, that the FAA detector wouldn't be able to be upgraded, you mentioned, not putting in the silicon vertex detector
09:45:10 off the start you know that's certainly one step, but we're. Is there room to augment the detector for higher luminosity.
09:45:22 Yeah, just to be clear, I don't think we would envision, and I don't think I would envision doing much of the HA science program with the TPC but but you could imagine commissioning with that we have been discussing potential upgrade paths.
09:45:37 And the upgrade path we've been discussing a really based off of technologies. and those those discussions are in their infancy, you can imagine, for example, that, you know, for the proposal right now or maybe for the detector we need to go with a technology
09:45:52 that we know can work, but boy this other technology boy in, you know, five years the r amp D on that might just be fantastic. So we've talked about doing exactly as you described building in potential upgrade paths, based off of increase improving the
09:46:12 performance of components, through technology choices.
09:46:18 But yes, that that is part of our thinking and we hope to provide.
09:46:22 If not within the proposal itself, but as a supplementary document, some information about upgrade paths potential upgrade paths.
09:46:32 Okay.
09:46:33 Excellent. Um, any other questions by anyone in the audience, we've got about 36 people here. So,
09:46:49 nothing.
09:46:52 Okay then. Thanks again john excellent talk, thank everybody for agreeing to come.
09:47:00 Okay, so our next speaker is Fernando Barbosa who's going to talk about an icy readout. If you want to go ahead.
09:47:14 Fernando and share your screen.
09:47:17 Yes. Good morning, screen.
09:47:32 Is it okay you know, uh, yeah. that looks great.
09:47:36 Okay, great.
09:47:37 So, um, my presentation covers, pretty much the IC without the baseline, that we have been working for perhaps the last year. I am the cam stands for the.
09:47:53 It's for the account manager for the electronics effort.
09:47:58 I always include people that have contributed to these over the past year again for the yellow report and various reviews.
09:48:11 Jean Chi Chan Chan.
09:48:26 A moved on to a different job, and we have a new person working with us on why, also from the NFL and we will hear more from gene and David Abbott later during the during the workshop.
09:48:31 So, let me move on to the next slide. So the outline is, I apologize in advance that I may go fast through a few of these slides, which might not be too much interest to the audience but in any case, I'll cover what we envision for the streaming without
09:48:48 the architecture.
09:48:51 And the number of retail channels that we have so far.
09:48:56 I'll cover a little bit about these technology nodes which is a thought that comes comes up often infrastructures that we are the infrastructure and standards that we have been working on a little bit of on the deck, and on timing in the in again, or
09:49:17 mentioned a little bit about the schedule. So the ice. Yeah, the streaming without of course is accepted and proposed by the community in general.
09:49:26 The streaming without of course is accepted and proposed by the community in general. And there are already successful in other sectors such as star and Phoenix.
09:49:34 In the future experiments of course we MLNJ lab will use steam without.
09:49:40 And there is a current collaborations between the labs to reinforce consistent development of hardware and software.
09:49:50 The aim of course is to archival the raw data, of course after nominal reduction techniques like sim suppression.
09:49:57 Of course these allows us to have many more physics laws.
09:50:12 The synchrotron radiation, so far seems to be manageable and it's lower than that Elysee, according to some calculations. And of course you can expect the rate optimization after an operational period of the machine as normally happens.
09:50:30 So here is just a diagram of the architecture.
09:50:35 Most likely common to everybody that has developed electronics from the detectors throughout that acquisition. In, basically boils down to the partitioning of the electronics.
09:50:41 So what you'll see on the left side here.
09:50:51 By the way, at the top, you'll see the rates, the data rates, the bandwidth, but this is total for the detector.
09:51:01 So we can expect to have about, 100 terabytes of complete detector. After some data reduction.
09:51:10 At the server level should expect about 10 terabytes. But the goal is eventually to get to storage to have, you're not running up to per second.
09:51:21 So we have a generic detector on the side on the left hand side, and then we have protein boards that carry the things that are pertinent to each of the detectors.
09:51:33 Some of these front end boards we all transport the data via fiber, you actually to several or link exchange something similar to the Felix board or others will require an intermediate step, and also serves as an aggregator, which, which is the front
09:51:54 end processor, which is primarily primarily dominated by FPGA rather than easy on the bottom side we see the standard infrastructure that will need for high voltage low voltage cooling systems, which we expect to be implemented, both at the detector in
09:52:14 that chassis is our project level.
09:52:18 So, these, these these, we expect these to be quotes or commercial off the shelf units and implementation.
09:52:30 So, our efforts will be primarily in developing the ASICs in the boards and of course everything that was in between. and the front end processes, the servers and storage of course there's also commercial off the shelf.
09:52:45 In turn, in terms of the development. So we have the front end electronics board, their TVs.
09:52:56 The ASICs that will need to be developed or apply to these particular detectors.
09:53:06 And then the online identification on the back side, primarily with CPUs GPUs and so on and switches commercial off the shelf. I should say in advance that one of the our consensus that we came to is that the any implementation of algorithms for let's
09:53:26 say for ml or he I should be implemented at the server level, rather than that, the FPGA level of force more flexibility that way.
09:53:40 We envision that the high speed to analyze data will be carried balls off the front end boards so from out of the box, or on the were that easy to reside in the front end processes.
09:53:54 It makes the implementation in whole the construction so much more.
09:54:00 Much easier, doing that, that way.
09:54:06 One item that should be of course innovation.
09:54:08 from the start is collaboration.
09:54:10 If collaboration is not usually part of the development, it's, I wouldn't say enough they're soft but as we have been through several developments we came to understand that code Griffin needs to be more thoughts through from the beginning.
09:54:30 Another item is global timing synchronization and I'll mention a little bit later in the presentation.
09:54:38 That is of critical importance.
09:54:41 The radiation levels so far seem to be much lower than that then let's see what says, of course, I'm going to inform of course that without technology and the partitioning of the various components.
09:54:58 So in terms of the FTP FTP development. We, given that is such a large experiment, we need to coalesce the specifications which we are developing.
09:55:11 At this moment, to minimize the number of units of different part numbers, if you will, AZ except PGP PCB performance or a survey of the power
09:55:27 that as that can be found throughout the community, different experiments, is that for 64 or 32 channel, AZ. The the power is usually have the order of one what nominally plus or minus.
09:55:51 In, so we need to address the cooling system from the start as well. We also need to address serviceability as much as possible.
09:55:57 During the for maintenance purpose for instance, which is also critical during running of the experiment.
09:56:06 In terms of development nominally it's about four years.
09:56:13 We had some discussions with different companies and we know from experience that it may take four plus or minus one year to develop the amazing. Of course we don't envision developing one from scratch, but looking at existing designs with possible upgrades
09:56:32 we have talking about the VM M, which also includes the LP Govt. That's all power. You can be transferred the receivers.
09:56:42 And the sample and the tiger and any others that might be of interest to the community.
09:56:49 standardization of the data format, every time stamps and center.
09:56:55 It's at the system systems online. It's very important.
09:56:58 Of course it's now it's too early to develop sets standardization, but it's not.
09:57:05 It's something that needs to be addressed. From the start, so we have a unified development.
09:57:16 For all these electronics firmware of course we know that takes quite a lot of expertise and a lot of development time, and they think even firmware is something that doesn't go away.
09:57:35 As we know, firmware development and resources have to be updated, not only for the development, but also for the branding of the experiment as we adjust our algorithms and so on.
09:57:44 The power that means for for the FTP might be higher, but we have more flexibility because they reside outside the, the detective volume in another part that it takes considerable time and resources testing and integration tests testing along the development
09:58:03 time. So as you design a native you have to test the AZ, and then you have to. Once you are satisfied test easy with the detector and put it all together make sure everything works well before the actually procurements and installation, it's very, very
09:58:22 important. And, of course, after installation that's is a major part of the to verify and qualify the installation, or the success of the installation.
09:58:37 Here, present the number of without channels. These based on the reference detector.
09:58:43 And it's the numbers that we came up and so that the blocks that you see on the cells that you see an orange or not, well defined that at this time.
09:58:54 But they are considered to be generally low numbers. So it's not a major impact at the bottom is just a summary table of all these numbers.
09:59:06 We all see that if you divide and segment detectors and qualify them in different technologies, you'll see that there is the silicon detector that take most of these channels that are apparently a lot of effort going on to in terms of raising these number
09:59:27 by different means for the silicon detector and the silicon detector is usually grab by eight weeks that are looking into the or attached directly to the editor.
09:59:38 I like the LP, for instance, that might require a new developer or an improvement for the IC, but this number is quite much higher than the remaining numbers for the other detectors.
09:59:55 So, we have 200 million roughly for the silicone detectors and roughly about 1 million for the other detectors, I least a total of 486, we may be able to decrease that two three, so we will be too and v3.
10:00:14 Just variants, number of viewers that just called for this purpose, between v3 by immersion, if you can generalize.
10:00:23 Make a generic AZ that service was some detectors might require timing in charge. Others might just require timing for instance, but we may be able to coalesce all the specifications into their.
10:00:38 In terms of it. I just put these just to give an idea because this question comes up often this is these are graphs that are planning for planning purposes from the SMC.
10:00:57 And you can see on the right hand side that most of the development or fabrication or work that is going on at the SMC are there the HPC in the G five developments smartphone g five, that's where most of the businesses, so the remaining part is relatively
10:01:12 low in terms of fabrication levels.
10:01:17 So in our case, we look at our community. We have employed and still employed designs that are ranged from 250 nanometers to 65 nanometers. Most of the new ones are towards the 65.
10:01:34 If you look at Moses, for instance, as an example, most of the most popular know that they offer that they offer he's the only 180 nanometers.
10:01:45 However, they still offer 350 nanometers I don't know who might be using that.
10:01:52 At this time, but they still offer the summary from from these what we can infer is that if you choose a robust technology.
10:02:02 Technology know that this time, it should still be available within the next five to 10 years shouldn't be an issue.
10:02:10 We are also working on infrastructure and standards for power supplies we expect to use standard commercial off the shelf like from a meaner and cam.
10:02:29 We are working on a grounding and shielding approach, unified approach to the wall detector. And of course the other standards are cabling.
10:02:33 That has to, they have to conform to National Electrical Code EMI and the FBI, basically you have to have class B as a driving requirements for the electronics and in the hall to avoid any of like noise and our limit the amount of radiated noise from
10:02:57 other devices in the hall. Of course, this implies that there has to be an active management of devices that are loaded into the experimental area to minimize these, these effects.
10:03:10 And, of course, everything else.
10:03:12 The power supplies. We also begin to be floating, and that of course goes with economic scheme that we are developing in terms of that they are can computing, we expect the, the winter is your 500 kilohertz.
10:03:29 Bunch crossings have been milliseconds, and nominally the peaking time, the most aggressive one we have seen so far for Jim detective for instance of the order of 40 milliseconds.
10:03:41 So whatever is he needs to be developed as to conform to that. That's the goal.
10:03:48 In terms of data archival.
10:03:52 In terms of data archival, as a nation, one of the previous slides we are the goal is to have only 10 gigabits per second, which includes the background noise estimates, and this is half of the expect for you to assess Phoenix, but it's still, it's still
10:04:06 a reasonable number.
10:04:09 So, if you take for instance, J live experiments who acts in class well.
10:04:17 The, the eventual infrastructure needs of the order of 600 milliseconds put it in.
10:04:22 So it shouldn't be sufficient with the current technology to meet the 500 kilohertz indirection rate.
10:04:31 And assuming there is an assumption of 20 weeks of running.
10:04:39 70% facility efficiency that translates into 140 terabytes per your fraud there are
10:04:46 other tools, of course, our AI and machine learning is and Commissioner Maria would be preferable or flexibility to have these at the server or the data exchange links.
10:05:00 You know, rather than at the FPGA and the frog in.
10:05:04 There is also our DMA.
10:05:07 And of course we we plan to have all these CPUs and so on, as much as possible to be commercial off the shelf.
10:05:16 There might be a development of a new Felix card.
10:05:28 But it's still just for accounting purposes I think that we need to address that. I also mentioned a little bit of timing.
10:05:32 In, and the synchronization of blocks and so on and stability especially throughout the experiment. I give an example of the blue xsJ lab where we get the, we basically have two clocks, one is RF clock derived from the accelerator.
10:05:50 In one is the experiment clock is a local clock is the clock that these then deliver to all the chances crates and so on for the acquisition promises.
10:06:04 And we monitor the difference between these two. These are not temperature control throughout the lands of course, they might be organized or them to controlled at the tech accelerator level for instance, but not swap.
10:06:21 So, to, to compensate for that. We have read backchannel feedback channels from the chances that resided experimental halls.
10:06:32 Experimental areas where we read back in the provide the appropriate compensation face compensation adjustments throughout the throughout the experiment.
10:06:44 So we get the leader that has been measured at the experimental area of your family and the femtoseconds know the experiment clock, we measure also be after the distribution and so on.
10:06:59 So multiple fibers.
10:07:21 At the back level, it's less than we measure it to be less than 42 seconds. And of course we are looking at the relative change between the two, and adjust accordingly.
10:07:15 We've determined the bench determination, which in the case of jail I have is for nanoseconds via scintillating clever solution city lighting detectives which are getting between 120 seconds.
10:07:30 Now, however for it and I should say, to get for bigger seconds. Um, it's it's not that difficult. We, we, even if you have a good design from the start.
10:07:43 reasonably conservative in terms of the hardware that you choose, and we have used multiple fibers for this purpose with low voltage Packer for instance, we get without much effort to get these levels of Jeter.
10:08:00 So it's short term.
10:08:02 But in the terms of the EIC to little bit more challenging, because there are some detective they call for 22nd resolution. In that case we need to have better than one Pico second clock jitter.
10:08:19 So, at least at the driving, if the goal for the design, we should address these as sub, sub people second.
10:08:28 Jeter throughout the experiment for good timing.
10:08:33 I'll move on. Next to the scheduling.
10:08:36 Here is a schedule that has been available for some time.
10:08:42 We are now doing in these areas of.
10:08:48 We haven't received CD one. So you are in the area of r&d and conceptual design.
10:08:55 And we have the designer then we'll start sometime.
10:09:00 Hopefully, after CD one for the infrastructure the accelerator in the tech sector efforts, construction will start after CD three. Now I should mention that this schedule is for a reference detector to be implemented ipv6 that interaction point where
10:09:23 it's the six o'clock position in on the recurring.
10:09:29 There is also of course the IPA, that's the eight o'clock, for later but it's not official. So, right now, we are working on a reference detector for IPC, all these information, its input into a management system called PC, which level.
10:09:51 And it's a scheduling tool that is used throughout industry and especially throughout the week.
10:10:00 And that becomes very very important once we start CD three.
10:10:06 Once we have all the schedules in place before CD three, that becomes their official and basically just let you know that every month we have reviews and of the scheduling.
10:10:21 And so it's scheduling costs, of course, and throughout the process, we, we want to keep that as close as possible so it's very important that we have good estimates of these developments for construction.
10:10:38 And this is the critical
10:10:42 that management sees as the critical path in red.
10:10:52 So for the timeline CD one design with the start of the design CD one was originally planned for April.
10:10:58 So, but it. Rumor has it that this actually end of June. That will get CD one in 41 of course what we propose these reference detector that is installed and ready to run on the first day Vizier feasible solution CD three is the process started construction,
10:11:20 and of course, a CD and CD one overlap a little bit.
10:11:25 And, but roughly we are talking about construction off about seven years. That takes us to operations in 2030.
10:11:34 So in summary, we are formalizing the detector readouts with patients and that's a very critical step in our role is to have these specifications.
10:11:46 Ready for dissemination for, so people can look at what is needed, not only for the front end processors. But also, especially for the easy. That is critical in terms of timelines is very critical.
10:12:05 So, we are working on that.
10:12:09 And also, we need to address the issue of their formatting and so on would be nice to come up with a format that would be common to all the detectors.
10:12:22 And of course the gic means is growing, and we tend to you for your help.
10:12:28 Thank you.
10:12:35 Thank you, Fernando, so questions.
10:12:44 I think it's easier if you just unmute and speak.
10:12:55 Okay then maybe I'll ask a simple question on one of your earlier slides you showed sort of the structure and requirements for the electronics from both the front.
10:13:12 Yeah, I think, slide six maybe it was.
10:13:17 I didn't
10:13:20 let me know which one.
10:13:25 Now go.
10:13:25 Go forward.
10:13:28 Forward.
10:13:28 Yeah there. Okay, I'm sorry no go back one, sorry.
10:13:36 Yeah, so I'm wondering you know one of the things that you, you, you mentioned, and which I find is important is you want to have calibration and monitoring.
10:13:50 How do you envisage this do you have chest pulses on the front end electronics are you going to have radioactive sources or lasers that will provide some signal that you can then monitor throughout the lifetime of the experiment over varying temperatures
10:14:12 and things like that.
10:14:16 Um, well we haven't put too much into the details of these, but I expect lasers in signals that provide the calibration precise timing will be very critical.
10:14:30 For instance, at Jefferson Lab at qx.
10:14:46 We do the timing collaborations are acquired prior to every data run. So we are talking about every few hours, we will collaborations So pretty much is automated at this stage was a little bit and that's why I mentioned earlier.
10:14:54 Earlier in the development, this is not looks more like an afterthought. Because it's not as important as the other development, but we spent quite a long time doing, making sure the calibration was correct.
10:15:10 And I think this is normal, but we right now it's, it's, well automated.
10:15:17 In this place, pretty straightforward. We understand the detector well, and it's part of understanding detector and all the electronics and so on.
10:15:25 That takes some time.
10:15:27 But at this time we don't have a real sales as to what additional resource you need for this collaboration.
10:15:41 But I think, in principle, from the electronic standpoint, we need to make sure that whatever hardware is developed includes all these infrastructure all these capabilities.
10:15:56 Okay. Yeah, that's great. Okay y'all come you have a question.
10:16:01 Yeah.
10:16:02 In one of your slides you saying that the radiation levels are expected to be much lower than the LHC intuitively I would have thought you know since that LLC, they don't have to deal with synchrotron radiation that that might be an issue at the gic.
10:16:18 Can you be a little bit more specific on what is expected exactly at the various locations for in the front end boards and the front end processes would be located.
10:16:31 Well, I don't have that with me in.
10:16:36 gene Anelka and worked on these other people, of course, I'd say, in some of the detectors.
10:16:46 For instance on. If I make these parts here will require that the frozen process of for instance bees, a little bit of a distance from the front end processes.
10:17:06 And that's why we prefer to have optical links. Of course, there is the question of the majority of their of the, the, the receivers stopped.
10:17:18 But don't I apologize, I don't have the information that we are still working actually on getting the as part of these gathering of specifications, what might be the requirement for the electronics and various parts, and guess begin my best guess is by
10:17:36 end of June we should have a firm understanding as to the requirements are pleased for these, I know there has been a lot of simulations I think may, you may hear from gene later during the corporation I'm not sure.
10:17:51 I haven't talked to him about these issue. Lately, but they might be a from him. Jen has his hand up so I.
10:18:01 He has a comment.
10:18:03 Go ahead. Tree by Ken from SoCal so I totally agree with Fernando still ongoing processor to determine the synchrotron radiation, and I still think we probably will carry forward, quite a large on some of the simple radiation.
10:18:19 Since many aspects of the machine and experimental can easily change the background.
10:18:25 And luckily many of the electronics will be away from ripping line white radiation bag on that will be strong as from the sink on both hands, and by the by the total rate is still quite uncertain I'm going to tackle it data, like how come tomorrow.
10:18:56 nonetheless I hope many things are far away from being mine so can be shielded from this particular background. However, I do understand the why I chose cloud Christian compound for silicon tracker there's no way to escape it. And especially is close to being paid, and we cannot put too much, too much material around
10:19:03 too much, too much material around it's supposed to sing relief and lens. So for that aspect of the current explanation wiring from quite moderate synchronous background, kids to two hours to a rented out as well be induced amount of radiation, so thankful
10:19:21 I think as a moment just to 12 certainty to uncertain to determine that.
10:19:24 And on. I also want to come to that. The next iteration of our designs Academy, which shifted IP, and as a time we're going to reevaluate to the sitcom background the rated again, and hopefully that can bring more clear answer than what I can provide
10:19:41 enough.
10:19:44 Okay, thank you.
10:19:47 Any other questions.
10:19:50 Okay, we're have a quick question. Oh, Yeah, go ahead.
10:19:55 I was just wondering if the ambition. Any configurable logic, right on the a sex in the front end.
10:20:04 Sorry, could you repeat the first part of your question.
10:20:07 I didn't say sex I said a sex.
10:20:09 So I was wondering if you could imagine, or if it's possible to put FPGA or configurable logic box right in the a six.
10:20:20 Yeah, in principle, ladies.
10:20:23 It's possible to do that.
10:20:25 But I am not quite sure if that would entail quite a big development to develop that particular logic for that.
10:20:41 I, in general that there is, there is no measure. Confused logic from these easy.
10:20:47 It's just for configuration of the various parameters of of the AC, like for instance gain or shaping times for instance.
10:20:57 But that's perhaps limited just to do that.
10:21:01 But if there is a better solution of course of course we are open to that, I mean that's that's that's acceptable for it. Did I answer your question.
10:21:15 Yeah, thanks.
10:21:17 You're.
10:21:22 Okay. Um, if there's no more urgent questions, then our schedule says coffee break. And what I would suggest is we're a little behind. So, but I think people need their coffee and need to be able to address.
10:21:39 Other things that they're normally doing.
10:41:56 Okay, let me share screen.
10:42:04 Okay.
10:42:08 And,
10:42:20 Some of them don't so I'm Martin pusher.
10:42:24 I work for, as Phoenix and as for a very small fraction of my time for the IC in Brookhaven National Lab. So I always like to show the satellite satellite picture from Long Island with the was was then known as the IO from Ireland, Rick as you could see
10:42:42 it from space you know like so then this is actually where, where it is over here, I'm not sure if you can see my mouse right here, and this is a little blow up off this rectangle I always like to see this, you know like, so you can actually be seen out
10:42:53 from space. Okay, so what I'll talk about today is a general overview of CS Phoenix data acquisition, and I will also point out but the benefits are for the SEO efforts.
10:43:06 Keep in mind there are two more actual s Phoenix detector related presentations here TPC sample and MVDXS both on Thursday morning, and also have one or two slides about our data acquisition I'll see that can vary fits into the general theme of streaming
10:43:23 readout and and CIC as well.
10:43:27 Okay, so first let me say why is Phoenix important for the gic. Well, okay, this is a streaming readout. So we are paving the way for actually implementing streaming without in our community of course there are other experiments and it should be and you
10:43:52 like and so on that already do this but this is really what you know like us doing it here. And then we heard this morning from genre etcher right is meant to go to Phoenix IPS locations and you know like whatever we do, there are benefits to future,
10:44:01 he as he checked to there. But then more importantly, our data acquisition system will see that it has really been a pillar of EIC themed data taking for r&d and test beams and so on and so on since about 2013.
10:44:18 That's at least when I started this, so this is for ERD one nearly six and several LD artists and other groups, which are basically working in the orbit of the gic, so I just put a little collage together here have a test been complaints all of all of
10:44:33 this year is except for this one is that the Fermilab testing that is every year it's kind of the Super Bowl of data taking when we actually make our trek to the FTBF at Fermilab and so this is the first that this 2013 one.
10:44:49 our manager of TPC that be measured here. We had a large consortium that stillness is called slice up that has all kinds of yes he think detectors in here.
10:44:57 This is about a better readout for color meters, this is a big gas detector LDRD that we had in 2019, you know, different kilometer prototypes and then of course you know like for the same TPC and gas detectors in general the zigzag readout that has resulted
10:45:15 in several papers and so on, so on so these are all stuff where the feedback is used every year. And I have estimated about 25 active installations of RCNZIC over and I'm aware of 30 about 30 or so elsewhere, I may not be aware of every single one that
10:45:35 does using it here, and the usual entry for a group to actually go on board, this is really the ease of use for some standard devices all the DRS SRS can what have you.
10:45:46 And then you don't like all the support for fully automated measurement campaigns that you need and this is normally how people get on board with the data.
10:45:55 So coming back to as Phoenix so this is really the artists view, and the non essential things have been great out here but you have from the outside. He had wanted color meters, but you heard about the morning from john right when he was going on about
10:46:13 his dog bones I mean this is kind of the heaviest detector going all around here. Electromagnetic color meter.
10:46:20 Time projection shape or a minimum buys detector floating here on the outside right so gifts are trigger intermediate tracker identity and the micro vertex detector, MMVTX was this is just like a typical layout for a collider detector with so much magnetic
10:46:34 field here. And so it's actually getting real so what I just showed you a minute ago. So, these are the hydroponic calorie meter modules here there are 32 of them together, they have like 390, tons together.
10:46:51 This is a cradle that is not yet a Brookhaven but it's it's put together it's and already completed at the vendor side and being shipped here I think in june july timeframe.
10:47:02 These are the sectors here from the electromagnetic carry meters is one module or 64 is the artists view. And also I when I mentioned 390 tons and intellectual you see me in the picture here I'm adding point to a chance to the weight by leaning on it
10:47:18 here so this is really what what we have.
10:47:21 There is another thing from the TPC what we call the wagon wheel here and the inner field cage on the, on the inside, so all the different pieces are coming together and then of course you know like our darling, our former former Obama did you already
10:47:35 heard about this just again me walking through it. So you get a sense of scale you know like how this actually looks like so we're very proud that we have all these things really coming together and looks very nice and you have something to show.
10:47:49 Okay, so these are the detectors that are taking part in the streaming readout here so it's clearly the TPC and I NTT and then our metric, micro vertex detector.
10:48:03 And so, you know, standard TPC few cage drift, to the outside and so on and so on. I will concentrate a little bit on the TPC but not too much but the other ones are just equally important and I'm doing streaming without.
10:48:17 Okay, so I have some good news about the data volume so when you have been at various conferences and also predecessor of this meeting, we will always going on about our data rate and.
10:48:28 So, there are some good news and our updated data rate, and so I've taken some numbers from the beam news proposal. And the point is here that we are now.
10:48:40 Estimating significantly lower data volumes. Due to the introduction of a beam crossing angle. And so the nice thing is it's not only less fewer conditions and less data volume.
10:48:54 But what we are really sacrificing or getting rid of other ones which are really not useful for us. And this is a little cartoon here.
10:49:01 Nothing is to scale. So this is the inner tracking acceptance here. And if you have some beans crossing the bunch of links is larger than the acceptance.
10:49:10 So you have all these collisions, all over the place, and several and a lot of them outside of the inner tracking accepting so these guys are still loading the TPC was charged but we are not really making any money off these guys here, largely useless
10:49:24 without the inner tracking. So now, if you put an angle and then just keep in mind this angle is wildly accelerated right so you know it's just not up to scale.
10:49:33 And then of course you know like you. The, the overlap of the two beams is just getting much smaller and you're getting more of the collisions in the acceptance and this is reason reason why we have been going down and it reduces the keep the charge density
10:49:51 that is the largest contributor and the data volume. And these are the updated numbers of what you have heard before as hundred 35 gigabits per second evolution, then this is broken down by the three years of our running the with different systems gold
10:50:01 gold in one one and three with largest volumes and then we have PPMP on the nuclei here and you see instead of hundred 35, we are saying 7349 97 was better luminosity and and so on.
10:50:19 At the moment so that looks much better. And as I said, we are not sacrificing any physics signal.
10:50:25 Keep in mind so that this is taken from our being used proposal of what you do normally there and said okay so even if we are only getting this much then we still get this right i mean like these are relatively conservative numbers.
10:50:40 And clearly, right with strife was all our my to do much better here. So, but, you know, like this is what we have written down and we want to do a lot better here, but all the planning has still taking place with your original hundred 35 gigabits so
10:50:52 can actually get you know like is 180 we can sort of like dump 230 petabytes a year. If we need to storage system so significantly higher uptime that we actually count on and want to achieve our okay and so this is actually very nice.
10:51:09 So this is the overview of the data acquisition system and then you like it skin all like what you expect and many similar things that you've seen the two green groups of detectors This is basically the color meters and also the minimum bias detectors.
10:51:23 They have a triggered readout.
10:51:25 And people have asked a different conference before so why don't you ever seen streaming and the answer is money, so we just don't have some money to do everything so we have to reuse our old electronics and also still be triggered without here.
10:51:37 And then we have the other as I've mentioned you know like MVTX in it at TPC in streaming read on it also breaks down along those lines here with the different things so these are front and modules, it's just whatever you call them right and then they
10:51:50 are the data collection modules and these guys are referred to as sub event buffers here. We haven't the order of 20 of those and then in the streaming end of business and these things have the name has stuck FTP fondant electronics.
10:52:03 We use the Phoenix cards in the event buffer and data compressor machines that's called immediacy. And then we have a big hundred gigabits of high end network switch.
10:52:17 And then we have these buffer boxes. And then the empty buffer boxes they interface, these are resist here, these green arrows here this is like a 2.5 mile one of fiber, they go to, what used to be known as Atlas computing fitness facility now it's called
10:52:33 the STC scientific data center. Computing Center and then you know like it's stored in the tape robot system here. I have a couple of pictures. So, these are 60 totals or 20 of those machines are already installed.
10:52:48 So this is what it looks like it looks like any other rec. Okay.
10:52:52 I find this you're very impressive. This is a buffer boxes has hundred and 214 terabyte discs right a lot of this space is is one of six that you show you protect really really heavy right all these disks and we need special rates for these guys but that's
10:53:13 what what I have. So one of six and give 20 of 60 so 1121 tix basically installed.
10:53:14 So, the data structures that we are producing in the end, look like this and you have seen this slide, probably a number of times, so each front and in general, couple of exceptions but generally contributes what we call a packet to the overall data event
10:53:34 event data structures here and there is a packet ID that uniquely identifies and never changes detected component and the front end card or whatever it is, where it really comes from.
10:53:45 And then there is another field that says rk and in this package, the data are encoded in this particular format. And then ultimately that's selected decoding algorithm.
10:53:56 And that's really important because we can change and improve this binary format the encoding all the time we just signed a new hit format for packet at any time and in the past in Phoenix, we have done this many times.
10:54:10 And so this is really the mechanism to insulate the offline software from changes in the online system so we have changes, none for nothing changed for the reconstruction for the user of the same spirit the same answer.
10:54:23 even though this is just encapsulated completely different. So, we have about hundred 80 246 packets in the typical less Phoenix event. And in the case of triggered electronics than a packet usually comes when as a response to a trigger.
10:54:41 And I mean it's still the same data structure but has a slightly different meaning for the streaming readout. That's what I'm showing here and the next slide.
10:54:48 So this is for streaming data this packet paradigm, even though it's still the same data structure and all the changes, meaning just a little bit and becomes more like a packet in the voice over IP sense right you hear my voice, clearly, my audio stream
10:55:02 here is chopped up in conveniently sized network chunks and being transferred to the network and coming out of your speakers again so this is just a little waveform here.
10:55:14 Actually this is from the song collider you know like from the CERN band, actually. And so this is just being chopped up in packets and that's really what we are storing.
10:55:23 So what I showed here is just a little list of an event that we took at the Fermilab testing here.
10:55:31 That may have been in 2019 or so you know I can see, this is streaming data so we're reading out the TPC you see this is a format IPC front and version two this is a particular format, but the point here is we're reading out streaming data here.
10:55:47 And so, so now I said, we have still legacy triggered electronics and we want to combine this with streaming and the, what I'm showing here are what I call the must have data.
10:56:02 And the idea is of course you want to have for a particular collision you want to have the thorough detector. So what what's shown up here is the stream that is coming off the TPC Is this a civilization not the real thing but I mean like this is the data
10:56:15 rate coming collectively of all the readout channels from the TPC here. And these red lines are triggers in the triggered system, okay what you need to do is, then you need to okay when he was trigger, you need to preserve the waveform and the pits and
10:56:33 what forms a TPC as long as the drift Time goes and correlates this was the other thing so but if you wouldn't do anything else, then they are here's a trigger.
10:56:43 Now I have to take data from the TPC as long as the drift time is about 13 microseconds. So, and then you know like so you have to actually preserve this data here's another one, another search in microseconds and so on.
10:56:56 So now here you have a bunch of triggers which are close together. So it's sort of like a reset, of the time so you just extend the time right accordingly and your cover them with this particular chunkier multiple collisions here's another one of those
10:57:11 and here are a couple of them. And this is really what you get them is what I call the must have data, where the triggered and streaming radar detectors are correlated, you have data from everything so this is really your baselines This is what you do
10:57:26 get on the physics.
10:57:28 So then we came actually is it was General came up with this idea that we are, we have seen opportunity to well put opportunistic tracking only data on this here and this out again.
10:57:43 What I showed before these are the must have data. And then what you can do if you have room in your storage system I told you I just have gone down the numbers a little bit, then you are just extending the stream time and add tracking only events, without
10:58:01 the color meters right so you know just don't stop streaming and just do a little bit more why it's important that you just don't stop and don't start a new stream here.
10:58:11 a color meters, right. So, just don't stop streaming and just do a little bit more why it's important that you just don't stop and don't start a new stream here. So, you know, because when, when you do this extension. This adds more than proportional crossing counts for
10:58:20 counts for additional data so think of it this way, right, we have certain microseconds of drift times that we have to cop to cover this a 13,000 nanoseconds.
10:58:36 And so you have to record the waveform length for the TPC to cover, cover one crossing right so this is you know one crossing one collision happened, you have to stream for searching Microsoft ok so now you add hundred and 10 nanoseconds to it, one crossing
10:58:45 9.9% length. But now we cover two crossings right so because, you know, like the next crossing is now also covered. and you keep on doing this, and each hundred nanoseconds here and add some miniscule amount of the data, but an additional cross, cross
10:59:02 coverage and then we can get 50% more data and perhaps even more by just adding this very small amount of additional brief time on this year and then we can choose is how much we do, to the extent that we do this to backfill our storage limit.
10:59:19 Okay, so we do not have an event builder, that has been a long discussion, and so many many many arguments that this is actually a good decision here so each of the CBS and ABC sexual before they write a file to an individual data as an individual file
10:59:39 file storage, and then they contain some event from the detector in question so that you know like for some of them the name separate Seb sub event buffer.
10:59:49 At the end we are writing 60 about 60 or so files concurrently and then the six buffer boxes, you know that each Buffer Box has about 10 files or something like this, that's shown here so these are individual of contents, or systems here, and then you
11:00:03 like these are individual collisions. And then, you know, and then you know like they are going into individual files. Now when you do your reconstruction and analysis, we have to combine them.
11:00:15 And this is a little bit beyond the scope of this presentation here but we have really demonstrated this in like 500 million events, so that we can either read the 60 streams on the fly, and they work really nicely, or we can combine them into a combined
11:00:41 and work on that we have both options we are exploring what's best at the moment but bottom line is this all works. And this is really the best way of going about. Okay, so coming about to the gains that we're getting was a stream streaming readout.
11:00:49 And this is a little bit what I have been well obviously thinking about the streaming readout paradigm in general a lot and you know every month or so I find yet another cool application why this is better.
11:01:04 And when you have a classic trigger event. So you accept the trigger we got to detectors, you're done next event right so one event is really well contained thing.
11:01:14 So there's a feature of many readouts systems that I found that I have come ask considering the defining property. even if you ultimately trigger your front end.
11:01:27 And this is really into readouts there is really no synchronized end to give an event that means, while event, n is still streaming and other places event and minus one minus two isn't really finished yet.
11:01:41 And this is very clearly the case in our TPC so the beta are coming like and so the two big chunks, and then, you know like, even the 32 bit chunk has data from different crossings in it so this is ingrained in the data format but there are other spirit,
11:01:55 and the show an example in the moment.
11:01:58 the show an example in the moment. But this is really where the speed increase can be significant, even for classic systems and let me show one of. Let me just show you what I mean with this.
11:02:06 So if you have these, not, not really well defined and so this what I show here on the second line here in this picture here. This is ok I reached out to us then I'll be done since then every data center we don't say so but what normally happens, is what
11:02:20 happens is that you have some interleaved so right i mean like so this in this system readout channel or without system you will already go on and read out so next event was this one isn't finished yet and then this is pretty much what happens in many
11:02:34 places. So how you could actually apply a busy to get yourself out of this and convert this into that but you can already clearly see the near slowing down.
11:02:44 And if you insist on these event boundaries and you could buffer those events and fragments and back memory assemble them, and then write out.
11:02:53 Now, you have actually bought event builder back and you have to do it online and you have to do it right the first time. And keep in mind that you cannot often ask the device to give you a state of and it's convenient for you but such as network packets
11:03:06 that have come in for example or something, you have ready to catch them. So this sorting is not very easy so you really are on the path of getting the real eventful a back you don't want to do this.
11:03:17 Okay. So, what I have been doing lately. A lot is, I'm just doing the offline sorting of streaming data into events and crossing and this little data file that I showed like a little dumb from in slide before from the TPC is exactly what happened here.
11:03:34 So I'm maintaining the streaming data offline pool that holds a number of crossing like thousand draw whatever. And then it's sort them by crossing number which is for us to improve value here so this is what you'll find looks like but what your, your
11:04:01 gets after this thing has been sort of like taken in the pool and the standard stuff has been pulled together. This is again what you get, but it's all done offline so the data acquisition can be faster. And so, you know, and you release a deck from all
11:04:04 just block, a lot of bookkeeping tasks and you gain speed. So here is a very good example that is at its core it's triggered system, but it still has this feature that there is no synchronized and to an event so this is a dream electronic I'm not quite
11:04:21 sure if anybody from a secular isn't this meeting today.
11:04:24 But, so this is from our lab so we have this a dream, basic that we are gas detectors. And so what you see here are two of the front end unit unit fe us here.
11:04:38 Normally there are a lot of more and this is just some data packets on the network.
11:04:44 And you need to catch them as they come so at again in this system here we have only two sources but you can have 10 or whatever. And then you know I can be many, many more.
11:04:53 And the reference without implementation, makes it so that installs of packets from different avenues in different files and to kind of implement simples quote unquote streaming.
11:05:06 These are the network packets, come on here and you see they are not synchronized right so this blue event and so on. This is just not at the same does not actually overlap 100%.
11:05:15 This is what belongs together what you want to end up with in the end right. So what I'm doing here, I'm just store them as they come and this is even faster than this reference implementation here, because I save all this decision making this package
11:05:29 is coming from here it goes to this fight oh this is coming from there it goes to another file. I don't do this year, and all of a sudden this done offline and this is really the way to go.
11:05:37 And this is really one of the hallmarks of streaming reader, why it's so cool. Nevermind what you can do with triggering and filling out your data stream.
11:05:46 So, you know, your front end can actually work faster with us. Yeah.
11:05:54 So obviously when you do this, the correct clock value is really of utmost importance right and then this trigger number that you probably had at some point event number what have you, by and large users, meaning was streaming readout.
11:06:07 And for us, it see Rick been crossing counter, counter we call it here.
11:06:14 And it's really what matters. And we had to demonstrate that. Well, we can do this right so that we can combine the 60 file so that I was going on about.
11:06:26 And, you know, put put this crossings together here. There was one little problem, our legacy kilometer digitizes that don't receive and they don't know, an absolute beam clock value, and they count independent.
11:06:42 Okay, it's not a problem. You can see that they are consistent amongst themselves right you can actually figure out if one of them bludgeon clock there is a different, but you cannot really tell out of the box.
11:06:55 If they are really synchronized with the rest of the system so that's what we had to demonstrate and you know like I was just going to show you how this, are we, how we did says here.
11:07:05 So, we this is a picture here of our GL one global level one, bought an FPGA boards that does all this as a debug header. And what we did is we brought the current beam clock counter value out as a bit, I mean the thing has 64 bits look at least significant
11:07:24 At least significant bits here. We brought it out on this feedback header, and then we just enlisted eight individual digitize the channels to just digitize it's completely stupid to do it this way but I mean like you know the digitize a one off those
11:07:38 inputs and then each digitize the channel with just digitize one bit here.
11:07:43 and then each digitized the channel would just digitize one bit here. And so in this way you could just put the honor roll over was 200 255 you could actually just figure out what's been clocked value really was the absolute one was for this particular event
11:07:56 event here. And this is, you know like what you see, so the bits are in some strange order here just believe me that I put together right through you know like the picture here records bonds to the restaurant the number, 192 and 94 and since 82 you see
11:08:11 but what can be done on business off here, and so on, just keeps this number in mind hundred 9294 82 sequence here this is what it is. And what we have done here is then, you know, these are the beam clock, that the GL one, the values that the beer.
11:08:32 The digital one sent out. This is what we got.
11:08:31 So this is for 250,000 events, and you see this perfect correlation. I said at the end, you'll see there's a number one there is one hiding in here that doesn't actually have a deal because one gets through but number one I mean so the the all of them
11:08:48 actually came came out right so this was really what we needed to demonstrate here, that everything is really okay and, and, and Brooks great.
11:08:58 Okay, so that's where we are, again we have a couple of, You know more talks about individual detectors tomorrow here. So let me just summarize for now that as Phoenix is on a good track to taking data in 23.
11:09:19 So we got a jumpstart taking streaming data actually this was in the year before the last year, okay it was supposed to PPC so last year we didn't actually go anywhere to protect data.
11:09:26 We have demonstrated that we can properly. Align and combine our individual data or the streaming ones. And we have a really good concept, combining triggered and streaming readout in the experiment, and also let me say that you know like what I mean.
11:09:43 Several of us are deeply involved in the ESC community and I'm coming clearly coming in from the data acquisition angle here and, you know, again we are continuing to support the various data taking efforts in the IC arbitrary.
11:10:03 So thank you very much.
11:10:06 Okay, thank you, Martin.
11:10:09 Questions for Martin.
11:10:17 Okay I'm on slide nine.
11:10:21 If you want, I mean, it's, it's where you talk about the that the crab crossing actually helps your, your rate into the TPC but I expect real collisions aren't what's driving the, the rate in the TPC that it's probably been gas interactions.
11:10:49 Actually I mean to me.
11:10:52 Well it does, it does help right because the upper picture here.
11:10:57 There are a lot of collisions and collisions are always making particles and charging the TPC. That's basically just useless for us right but it's happening here outside of the acceptance.
11:11:12 And that goes pretty much away.
11:11:15 Alright so this is really the the thing yes has also been gas and all the stuff, but actually probably Jen has a better number for for this, but this is actually taking friends at TPC about, give or take half data rate out basically because if you're
11:11:34 loading your TPC. I mean you cannot know when you're reading it out whether it's called a hit you just have to record the charge as it isn't if you're putting less charge into the GPC really immediately translates into less data volume, coming, coming
11:11:50 out, and I'm not really don't have a number for the mixture of, you know, like other background beam gases or maybe Jen, can chime in. But this is significance.
11:12:01 So this is not a small con it's one of the major contribution to back our background in, and unwanted charge in the TPC.
11:12:12 Okay, so if I, if I can. So, in closing reduce the article acceptance, cross accretions by about a factor of 10, without affecting the D. Being word hex collision much reductions about the 20% and so that were previously we almost every his way to King,
11:12:32 Is it justified one to find power creation. Now most of our heads, taking GTC is good condition that we can use with the silicon tracker that's why it's so dramatic as a marketing illustrated here on the good news for Rick though is so compelling to us
11:12:50 The good news for Rick though is so compelling to us is we news for Rick is that the key cross section so high. I saw some time higher than EIC in a way that being that interaction is much smaller fraction comparing to use it.
11:13:02 So that was our previous operation from Phoenix is that the luminosity is really collision driven it's not a theme that's driven. So that's what we expected it to DC most data will be fun collision.
11:13:18 There will be neutron big ones or we just slow neutron flying around. And does that to the contributor to half of CDC data and disaster irreducible from the detailed view, but we can do more about that in computing so I think something we are going to
11:13:31 touch on the computer in session to yeah that's okay thanks Jen, Chris. Go ahead.
11:13:38 Hi. I was wondering how your choice in streaming Rita affects the trigger logic in in the detectors.
11:13:47 Um, so you just show.
11:13:55 Just see here. Um, I think, This here is solved by illustrating it.
11:14:01 So we have a trigger logic board right it's a big FPGA that has gets all the, the input. At the end of the day though it's not actually that much, that much input that, that you might think.
11:14:18 So for gold gold right we decided to run and minimum bias trigger.
11:14:24 And that's basically made by the minimum bias detector by and by and large, right, and okay and so these. What I show here in the slide these red lines.
11:14:39 We have to obviously communicate the fact that the rest of the system has been triggered, to the streaming readout front and, which is the Phoenix cards right so I need to be aware that oh yeah we have to peel out these particular chunks right so we have
11:14:55 to distribute this right so this is really what what what's happening. But other than that, it's pretty much still a classic thing so a big fat FPGA sitting there, making the logic decision was various triggers that coming out you have scaled down so
11:15:13 you have a busy logic, you know, all that stuff. So,
11:15:20 of course it's it's a lot of effort to do this but I don't think the fact that we are streaming here is really that complicated. Okay, so, I think you might come from a different angle.
11:15:31 When you say okay I stream, like in the upper picture here, the entire thing.
11:15:37 And now I have to actually look at the data to sift out individual hits and do sort of like a soft software trigger or something like that, that is by and large not what's happening here so by the time we have these triggers, so everything is said and
11:15:54 done so we are taking everything that resolves into this red line so there is no sort of like a higher level trigger behind this is that what you were kind of getting at here.
11:16:05 So, so I must have your detectors really get an external trigger and then you just read out for a certain amount of time. And those go straight to desk almost.
11:16:14 I don't watch. I mean, I think that would say just in a way I'm tempted to say yes but it wouldn't just do justice to like what's happening in the screen read out the text, because they already I mean like this is here, a data rate that is coming off
11:16:29 the detective after they have been going through all that clustering and get recognition and all that stuff so that is already happening in the streaming part of the streaming read Alex by then.
11:16:41 Right, it's not like oh yeah take this entire waveform right like you know you would record a CD with audio or something that's not what's happening like this, here are really representing already hits and so on.
11:16:49 and this the TPC happens to be quiet. Then there is nothing coming but this is really the the aggregated data stream after all this data reduction step already.
11:17:03 And what I mean this is do not take this here so as a naked role waveform or something that's coming up. These are already recognized hits and clusters and what have you.
11:17:12 Right. Okay, so if you want to, I mean that those could be available for triggering logic in a process for him. Yes, it could be. On the other hand, with the rates that we have.
11:17:27 And there's a reason in particular for the heavy I'm wanting that we are doing minimum bias, right, because it is fantastically hard to trigger on anything useful and if you have the ability to take every single minimum bias, without a significant dead
11:17:43 time you better do it, because then you also get processes that you can not trigger on, Mike.
11:17:53 And, you know, we have the luxury of being able to do this, whatever results into this red lines here, we can actually store so just run with it right and go like, like, like, dude, that's a little bit different in, when we are running PPC rate is just
11:18:08 too high for that. Then we will have, you know, like, a level to trigger that I didn't talk about it all here. And, but but it's also then just results in sending out these red lines here it's not really what what you would consider a high level triggering.
11:18:24 Right.
11:18:26 Okay. Thanks, Dave.
11:18:33 Yeah, hi. Hey Martin. So I'm curious, you guys.
11:18:41 How do you manage your, your dead time and what do you do with the streams, when, for instance the triggers are inhibited.
11:18:51 Or like, how do you how do you stop them or to use throw the data that way what's the what's the back pressure management there in this hybrid session.
11:19:01 So when a trigger is issued. Right, so it goes on to the front end.
11:19:07 It is guaranteed that the data that has been triggered, as the landing place. Right.
11:19:15 For the trigger data, you will but yet we'll wait a second right okay so for the trigger data. The data once triggered it has a landing place so we have this multi event buffering and all this stuff so but at the end of the day, when we, when the system
11:19:32 recognizes that oh if I will trigger one more. We might actually clog up the front end.
11:19:35 This is basically when we are not issuing a new trigger so the GL one the global level one trigger is either issuing a trigger or not. Right. And that then also translates in one of these red lines not happening.
11:19:51 Right. So right i mean like if the system is going busy let's say you have some cluster of things here. And then the the system is going busy these guys are guaranteed to have a landing place.
11:20:03 Fine.
11:20:04 But then you know like we wait until the pipes have sort of like been drained and then we open the next trigger again, this is basically the management here for the, for the trigger and then translates into the streaming readout because they are obeying
11:20:18 the trigger, even if they are doing their own thing was recognizing hits, and so on, so on they are sending the data which are correlated to a trigger, even though they are not triggered themselves.
11:20:29 And then we have, as I explained a little.
11:20:33 Another, you know, tacked on opportunistic streaming component here.
11:20:39 So that's basically how this works and also keeping Miami like you know yeah I mean in the extreme case, when we really load the TPC all the other detectors might be an issue the Felix cards but job behind the TPC and streaming things.
11:20:56 It has a little bit more capacity than our legacy electronics or the legacy Electronics is really what drives the that that time and too busy going up.
11:21:06 But still, we are shooting for a lifetime that is above 90% in minimum buyers wanting.
11:21:14 And so, you know, like we probably, I'm not quite sure yet if we haven't led that chose busy I mean that we had in the old system. And so, we don't want to see this go up to my.
11:21:28 Does it explain it or, yeah.
11:21:30 Okay, thanks everyone I think we need to move on. So, Martin if you could give up to the screen. And our next talk is about DAQ at the spin physics detector Bye.
11:21:47 Nana, often as.
11:21:51 So, you can share your screen.
11:21:54 Nice thought, Let's try
11:22:04 Yeah, looks good. Okay.
11:22:14 So.
11:22:09 Oh, go from the beginning.
11:22:14 So I'm going to present the data acquisition which is planned for you experiment which is called in physics detector, which is placed on your collider which is built, which is under construction of the Institute of Higher Institute of nuclear physics.
11:22:34 Nuclear nuclear research in dogma.
11:22:43 Here's some
11:22:42 information about Collider.
11:22:47 It will be it will provide beams for studying from proton, up to gold, and having to across the globe crossing point one, one is for multi purpose detector, mainly for iron physics, and another point is for spin physics detector, which is will be useful
11:23:12 polarized beam.
11:23:23 So, now we can accelerate and the collider under construction, and it looks as it was Pete you were close to the project project and current status of construction.
11:23:38 So, spin physic detector is going to use polarized beam protons and neutrons, with a first luminosity up to 10 to set it to whisper polarize it wasn't 70% all possible
11:24:02 orientation. When you do it alters first of all on polarized.
11:24:08 So ideals experiment is to study
11:24:13 with polarized and polarized beam glue on components, inclusive reaction mainly Chairman you open chair and blunt photon productions production.
11:24:29 So, this is a main idea.
11:24:32 So we are going to observe Harmonia open shop and.
11:24:41 And from photon mainly at the range of energy up to 27 gb
11:24:48 souls the start of collaboration is a form conceptual design report was designed and mainly accepted by program advisor committee in Gina, in general, the soul, the next step for us is a development of the technical design report.
11:25:13 And he has a few slides with more articles about physics, which is, we may be interested in for for physics. So, it's about main concept of physics, and this will be a paper about physics, for sure stage one accelerator was not at the highest level luminosity.
11:25:38 So there's references for all be publishing already published
11:25:45 status of collaborations. We are not on the stage of protocol operation. Let's see.
11:25:52 There was one meeting girl brought up proof protocol operation. Officer offline in 2019. And now all remote, there was to remote protocol operation board meeting.
11:26:07 So for the moment.
11:26:11 Already a city about city Institute from 20 states signed is going are going to participate.
11:26:27 Scientists conceptual design report and are going to participate in our experiment. So what is the main idea of the experiment. So it's a four p collider experiment, which is rubbish use the magnetic field as a separate coil six separate Co.
11:26:44 Super conductive magnet with, with about one Tesla
11:26:51 magnetic fields.
11:27:09 silicone voted for drinking for voting center.
11:27:13 There is, we are considering two options, five layers, cheap or free real three layers of maps and two ways of still have strips.
11:27:27 Next, detector is a tracker, which is going to be is going to be to be able to on stroke cubes, with about certain double lives of stroke,
11:27:44 which will provide the accuracy about 150 microns.
11:27:53 Next, we're going to put a political identification, mainly for not time off like property.
11:28:04 So there is two options.
11:28:08 guest judge, Detective PC or silicone based.
11:28:15 It's not decided which options will be selected.
11:28:21 So here's the property.
11:28:24 Okay required property for template system.
11:28:29 Buddhism, 16, because seconds.
11:28:33 But it's not it's not enough so much better as small as possible as usually.
11:28:41 So next detector is a medical area.
11:28:46 We are going to use type of query letter with CPM energy resolution required energy resolution, not better than 5% energy.
11:29:07 Energy resolution, not better than 5% energy. So what.
11:29:14 Here's the size. So the size, about five meter.
11:29:14 I want you to know, and transfer says more than three meters.
11:29:25 And most heavy and most okay not not most important but quite important part of our detective is a so called drainage system immune system. It's be done.
11:29:38 main idea is to use it yields will be absorbed as a straight line almost straight line, but hydrants will produce the shower. And also, neutrons can be absorbed, with this was this system.
11:30:08 This system is almost the same as we're already built a building for panda experiment, and we're close to the system which is already used on in the compass, or amber experiment at CERN.
11:30:28 So we are we are not invited something very new for the system, and almost all private system are already based on system which already developed, or for some experiments which already on stage of construction or already running.
11:30:50 So the four most important for the positions that
11:31:02 works with front end electronics which is required
11:31:05 to be satisfied for experiment.
11:31:10 For the moment moralists
11:31:14 found solution for electronics is forage system, and for strong, the store that data.
11:31:23 So we are going to use up all creatures use coupon the experimental for Compass experiment
11:31:32 electromagnetically MIT, and silicon detector is not still not there is no still final decision for, which electronics, we will be used for this and the directors.
11:31:45 So, concerning data flux and the water, word for data acquisition.
11:31:52 But expected bunch cross and will be issue, nine, each at 18 seconds with closing rate from 12.5, my fingers expected the collision rate at the highest luminosity will be three or four mega, mega tears.
11:32:12 So this means that there's no idea how to put a hole in what trigger for Sasha fluxes. So we, when we are going to use trigger lists or freerunning dq to avoid for hardware bias estimation for data fluxes know about 20 gigabytes and gigabytes per second.
11:32:37 So we detect musicians should rock with should accept such such flux of data.
11:32:48 Here, requirement for the front end electronics which is
11:32:57 full from the ideas of the free running dq
11:33:04 to exclude at least some 00 which usually exists.
11:33:12 Trigger electronic books, electronic so front end for the book sale for in Franklin from the free self trigger digitizing on board, and so on.
11:33:27 Mainly main ideas is electronics front end, which will provide timestamp include to output format to, to build events
11:33:42 or insincere free running in Iran and what
11:33:48 with requirement forms dq additional requirement for frontier to front end electronics, it's coming from idea to use dq based on ideal for amber or compass experiment.
11:34:13 And also one additional wonderful idea to use white rabbit system for that time synchronization.
11:34:22 So idea. He has his.
11:34:26 So idea here says they do okay concept conceptual idea of the data acquisition which is based on as a told on idea of compass experimental invited way either corner group from Technical University of Munich.
11:34:44 So for those of your stage.
11:34:56 We use the protocol for transfer all sitting in from Google Satan, and synchronization information to the front of God. Are we using the using time controls the controls temperature control system.
11:35:12 And when all data are collected by set up for multiplayer wishes on each level, collect data
11:35:31 and format it to, to the, to the continuous stream.
11:35:32 So front end concentrated already existing multiplex or which is used for inner compass experiment. Next to level of multiplex or under, under design, which will have a much more capacity and much, much higher but it's
11:35:58 also the last level. This multiple multiple Xf collect data from all detectors.
11:36:04 We are going to, we are expected to us to such multiplexes, which collect data from all detectors, and when on the output we have stream stream stream of
11:36:21 events, continue as a continuous time. And when and the final level.
11:36:27 Using the network link.
11:36:34 We distribute these events.
11:36:37 Up to online feed through online filter, which will be required to suppress data with factor.
11:36:52 From 50 to 100, to provide it required flats, which is acceptable from point of view of saving data.
11:37:06 Whoa, it's looks
11:37:20 as cool as it looks.
11:37:13 It's okay It's already used in the campus experiment multiple, multiple access with, which is inserted on as a mezzanine to the vv framework.
11:37:37 If we will use social multiplex are. We need to the front and walk around 400, Sasha, Sasha unit.
11:37:41 And will you will put okay more wisdom, 50 VME creates.
11:37:50 Now, this.
11:37:59 be me be me what is really, it's only customer construction for carrier for this multiplex are now.
11:38:14 Okay, they're going to put free for such multiplex on inside one units in format of Africa.
11:38:27 Period. It will will to certain significantly
11:38:33 more
11:38:36 compact more compact placement of the electronics,
11:38:43 as an input for computer we're going to use commercial commercially available.
11:38:51 The same a buffer.
11:38:55 Based on.
11:38:57 If PG.
11:39:00 And people from Munich already test such a, such a device, and if it was all the prophets that one.
11:39:13 One giga bytes per second is acceptable for such a device.
11:39:23 So, he has some idea of how synchronous synchronization. We will, we are going to distribute to 125 mega gifts from white rabbit for all detectors and scale, and the scalar inside to all from one thing to Eunice will be will be used to synchronize events
11:39:48 inside each detectors, and when we will run a real call using this scalar will collect data and build the continuous set of events. It's all shown in.
11:40:05 In previous talks, some idea how to do it so we not developed very deeply, all this idea how to separate the wins.
11:40:20 How to build the wins. In our case, it was will be overkill overkill.
11:40:38 So summary.
11:40:39 It's been using your approval experimental project, which is will be placed on the Nika collider agenda for study of polarized proton and neutron beam with Harriman no city, with all possible combination of polarization
11:41:01 is thinking detector is a universal multi purpose for p detector, which includes all your standard detector for particle indemnification and
11:41:16 measurement of the tricks and so on.
11:41:22 And we are going to use the free running dq as only one possible solution for the moment.
11:41:32 It's been open for new ideas and new collaborations the main the main message of my talk is that you were invited new participant was interesting to this physics session, electronics for such, they were to contribute to session data acquisition system.
11:41:54 Okay, thank you.
11:41:57 Okay. Thanks very much.
11:42:01 Any questions.
11:42:04 Martin.
11:42:06 Yeah I'm, I'm thinking.
11:42:09 The thing with the different I mean like I forgot the number 25 or 40 VME crates and you have she's concentrators and so on and so on. I, I believe that in last couple of years right and it has been recognized that PCI Express Bus.
11:42:28 the very best of the old days right instead of just doing this dedicated electronics, couldn't you just take something, just like for example the Phoenix card or, or, or, you know, like any of the FPGA cards, which are already sitting on a PC and have
11:42:47 them takes place I mean I can sit of 12 inputs that you showed there was what looked like individual LC connectors right, you would have this multiple fiber input, you know like imports or whatever have you and get a lot of more information into a board
11:43:06 which is already sitting in the PC and can be easily used to just configure the entire thing.
11:43:13 Okay. As I know the feeling says all Linux is going to use this very simple. Not, not PCI Express but something, I don't know, I don't remember exactly which interface but very similar card which is going to be installed inside to come inside.
11:43:34 In computer, let's say, so it's very cool solution.
11:43:40 Felix, or social board, which is available.
11:43:47 It says there is no big difference. Only the advantage of fail exceeded. That is good. This system is used for such a big experiment as obvious as I remember.
11:44:02 And it's a big advantages, but on performance, it's almost caused cause solutions.
11:44:09 Yeah.
11:44:11 Excuse me, I mean this is what I'm getting it right i mean you are incurring all this. Well if it's ATCA have to deal with that. You may get another bus system you may get another shelf system, instead of just putting it straight into the PC that was
11:44:26 basically my better off going to something that is sitting in a PC to begin with.
11:44:32 Okay, our idea for the moment for the moment I know, I know, I don't know if it sees that where it says the final solution.
11:44:40 Where is the very beginning, We are trying to follow.
11:44:44 For the people who is building this similar system for amber experiment at CERN at SPS.
11:44:52 So we're going to do. Okay, but part of our people who is participating in SPV is also participating.
11:45:03 So,
11:45:03 okay, we are going to see all the problems which they will meet and run will at least don't make the same for the same problem. So for the moment they're born to this direction with specialized FPGA based multiplex so well.
11:45:25 For the moment we're trying to to follow them.
11:45:29 Okay, we're ready to consider another approaches.
11:45:35 But to put them off the way I don't see a big advantages of
11:45:42 Phoenix.
11:45:44 Felix, Felix the idea.
11:45:47 Okay, okay, let's see, let's see.
11:45:53 Okay. Other questions.
11:46:01 When do you anticipate
11:46:05 the accelerator being completed and last Sunday I did it just pulls it is no such a such a transparency so sorry there's going to be at the end of 2000.
11:46:19 Okay, 20 2020 2022.
11:46:25 But it's only four islands and.
11:46:31 So, okay, another experiment to MTV which called mpg in Nico, oh experiment is going to start much earlier than 2025.
11:46:42 Okay.
11:46:44 Okay.
11:46:45 Any other questions.
11:46:51 I don't see any hands or anyone speaking out.
11:46:55 So thanks again for her a nice session in the morning.
11:47:01 We'll meet again at two o'clock this afternoon Eastern Time.
14:00:07 Okay, welcome back.
14:00:10 hope everyone had a nice siesta.
14:00:15 So we'll get started with the afternoon section session.
14:00:23 Dr Han Ji knew I hope I'm not mispronouncing your name too badly.
14:00:29 We'll talk about the solid experiment, and in particular the electromagnetic calor emitter flash a DC and the forbidden words trigger.
14:00:44 So, Angie Go ahead, share your screen and you can start whenever you like.
14:00:58 Okay, thank you, um, that's pretty good pronunciation. My name. So I'm going to talk about a solid Jq for Chris on the color may turn reality and it's triggered by design.
14:01:07 So this is on behalf of the solid dq group, and the here are the people who really deep the hard work, and I'm just the person who test their work.
14:01:18 So, my, my slides probably going to be less technique and quite basic here, this is an outline of my talk. So I will first give a brief introduction of solid and solid dq requirements.
14:01:36 And then I will show the decay modules that are going to be used and they are features and 30 I'll talk about the color mentor and the trigger design.
14:01:50 So solid, it's a high luminosity device with the large scatter and go acceptance and the floors in this as a meal so, Angela acceptance.
14:02:02 A supports multiple wrong group experiments, which can be divided into two configurations. Why is the semi inclusive deep in the last in the last six gathering, and the other it's the paradigm overall rating Diaz.
14:02:20 And here on shows the configuration of the two of the two set up, and you can see they share the detectors but the detectors up place in different locations.
14:02:36 So for the PVDSDQ.
14:02:41 Because of the Ph. Ds is an inclusive experiments. So, it only detects those scattered electron.
14:02:48 So the main trigger Columbia electron trigger which is a coincidence between the trunk of into the color amateur.
14:03:08 kilo hertz, which is on the screen possible for the dq to handle.
14:03:15 But the good thing is, we can divided in the system. We can divided into 30 independent ecosystems corresponding to the 30 sectors of the detector.
14:03:29 And as a result, the, the, the cheaper rate for individual sector is less than 20 kilohertz.
14:03:44 The rate is still high so it going to help tie ups.
14:03:41 So we're going to use the app EDC Ross and promote and I will ensure, I will explain later.
14:03:50 And I will ensure, I will explain later. And the correction to the playoffs will all be done in offline analysis.
14:03:57 Because, because the epi the ADC record records the full.
14:04:03 The, the shape of the, the shape of the signal. So the data rate is quite high, it's about 2.9 gigabit gigabytes per second, and for each sector is about 94 megabits per second.
14:04:12 So, which is much smaller than the credit limit
14:04:24 for the semi inclusive configuration.
14:04:29 Because there's some inclusive experiments, a detector detects both the scattering electron and the pions. So the trigger going to be two types winds the electron trigger, which is a coincidence between center later and the color meter and the trunk of
14:04:48 for the one won't win for the full word and go forward and go detector.
14:04:56 The electron transfer rate is about 148 kilo hertz.
14:05:03 palantir Columbia queens dense between the center later and the color meter, with a lower threshold, then the main trigger, it's the Queen's dense between the electrons and pile.
14:05:17 And this surprise the trigger rates are to be less than 100 kilohertz.
14:05:22 Another very important experiment to using the CD as configurations literary site experiment.
14:05:29 So in the GC experiment, we want to detect. Both.
14:05:34 Both scattered electron, and also the GC decay the pair.
14:05:50 Peer production events. So, so the trigger gonna be a three queens dance trigger and the trigger rate is about 30 kilo hertz.
14:05:51 So as a result, the requirements for the dq going to be, it should be handle 100 kilo hertz trigger rate with small time, and also issues support reliable clean, clean incidents triggers.
14:06:10 I'm so sorry there's a very high luminosity device and he has lots of channels.
14:06:18 For example, He has about 4000 trigger channels. So this makes the dq kind of a challenge, but with the top JUVDQ development that has already been used in the class 12 and blacks, so it's not, it's, it's possible to, it's so those requirements are achievable
14:06:42 with the new developed modules and and technologies.
14:06:48 Are you know 12 JVJ lab dq development to the electronics group to is to use the way access creates.
14:06:58 It has all the features the VM equal retest, but we sent additional feature, so it can synchronize and post signals between the modules in the play law, payloads laws and switch slots through the, the back plane.
14:07:18 In the pillows laws, we can have every DC.
14:07:20 So FEC is a 215 megabytes per pipeline flash etc. So simple. The simple of the signals error for nanosecond.
14:07:31 It can run into mode.
14:07:33 Winds the raw sample mode and the other is an integrated mode. In the raw simple mode at recorded the data, the records the shape of the pulse. So here is a cloud showing an example of the raw mode, and well in the, in the integrated mode at will only
14:07:53 recording the total charge of the, the signal and also the time when the signal cross a threshold.
14:08:05 The FCC also has some other bonus. Other features a can, such as a can provide the, the scalar cons for each channel.
14:08:16 So the CDC has two distinct data paths. Winds the trigger data path.
14:08:30 So, it continuously streams trigger data to wake up through the backlink, and the other is the readout data path. So that VDC on continuously stores the data in a circle buffer and a won't read out intuitive receives a trigger.
14:08:43 They, the other way important module is the VTP, the way access trigger processor. So the ATP is placed in those which slot.
14:08:55 It's, it's connected to the modules in the payload slots with four for the black serial links that can operate up to 6.25 gigabits per second.
14:09:09 Through those links the wiki be can receive data from from modules in the payloads laws. It can also receive data from the other way TPS look at the other creates using the optical links.
14:09:25 When the ATP receive all this data, it can it can generate a triggers based on the data.
14:09:35 Another important develop the event, blocking result.
14:09:42 So instead of reading. I read out the data invented by invent it packs a number of events into one blog to read out. And typically, we have 20 to 40 minutes new one block.
14:09:58 So this makes it possible to run the dq was triggered bigger than 100 kilo hertz. Otherwise, it will be limited to 10 kilohertz.
14:10:12 So, um, So with all those develops.
14:10:17 We think it, so it's possible to run the to meet to the requirements of solid dq, and we have the, the pre r&d for the solid uq to understand the capability and limitation of the, of the design.
14:10:36 Apply off the prayer and D can be divided into three parts.
14:10:40 The first part, it's about the IPCC result.
14:10:45 So, except to jam basically all they all their detectors will be riddled by Fe DC.
14:10:54 So, in the following I will focus on those two topics. Once the development. The development of the FEC password out through the BGP, and the other is the prototype color mentor trigger algorithm.
14:11:13 So the reason I I didn't mention the other Queensland trigger designs because the color and winter trigger gonna be included in all the, all the electron triggers so this is like the first step of we have to, to do.
14:11:30 And also it's quite it's relatively the complicated one. So, if we prove that we can do the color mental trigger algorithm. All the other eight coins then should be on straightforward.
14:11:43 The second part of the apparently it's the trunk afraid out.
14:11:49 It wouldn't be talked here in this talk.
14:12:01 The third party the jam Renault using other the VM m three chip or the APV 25. The Alex will cover this part in his talk
14:12:07 on social so for the first readout.
14:12:26 So previously the FEC data it's riddled through the VM a result controller, which is limited to the data transfer transfer rate is limited to 200 megabytes per, per second for the interest rate.
14:12:27 The, our testing results showing that this will give about one person died time at the immense rate of 20 kilohertz. For one, epi DC on to improve that.
14:12:41 The engineer upgrade to the FEC radar through the ATP.
14:12:48 And here's a nice slides from Ben rudo explaining how what he did for the, for the upgrading.
14:12:57 So, basically, before we run the four lanes that we run that the four lanes that two links at EDC and wake up at 2.5 gigabits per second for the trigger, they tap half.
14:13:20 And, Hey improve that to 3.125 gigabits per second to be enough for both trigger and also the inventory.
14:13:27 Also for the inventory it out. So this will gives our 200 megabytes per second, data transfer rate for each ADC.
14:13:56 The system has been updated to handle combine the trigger and the event data streams.
14:13:56 The ATP has for free. 10 gigabits per second.
14:13:53 Ethernet optical links, and currently we only use one so.
14:13:58 So we. So there are lots of space for the future improvement.
14:14:06 And also, by you, then this false readout.
14:14:13 The in the humans blocking is probably not needed.
14:14:14 So currently on the phone we're for the fast radar has been complete completed and the coda expert it's on so called expert it's trying to incorporate into the coda system.
14:14:29 and once, once they are works of down they going to pass it to me and for the, for the rates compatibility testing.
14:14:42 Okay then. Let's talk about the solid color meter.
14:14:50 so solid color. Color military use the shush like sampling color meter. So each blog, it's what it's, it's riddled by one PMT, and the here's a layout of the blocks in one sector.
14:15:03 There they are 147 blocks in one sector.
14:15:09 So, traditionally, the way we do regenerate color in winter trigger is by, I take a sum of all those signals of all the blocks signals and the policy through a threshold to say whether it's a electro not.
14:15:29 This is not very efficient, a more efficient way should be.
14:15:36 If we can, we can identify the fire to plasters, and only and use that as a trigger and only riddled the fire. The fire the blocks data.
14:15:47 And this can be dying BGP in and and is already implied. It's already using the cocktail
14:16:00 idea of the trigger. So, the idea of using ATP to generate triggers. So one sector. So, so all the blogs in in one sector, going to be read all by the FE DC located in one create and the way TP in this crate can gather the data from those Ephesus.
14:16:24 Also, it can get the data from the nearby to nearby Chris VTP to get the, the edge blocks of the nearby sectors. So once this all the TP has all those data in hand, we can develop autism, to identify the fire the clusters.
14:16:45 The trigger algorithm is designed using the model he is.
14:16:52 So he may rather he is, we write the class to funding algorithm in c++, and the software will on convert the c++ function to the HDL language, and in the other two we will go tues will map, then this HDL function into a to the FPGA chips.
14:17:18 So, so in this way scientists who don't have the electronics engineering background can can get involved in the trigger algorithm design. And also, we can fade the, we can fade simulate simulated data to test the trigger all resume before a quite the
14:17:39 chips.
14:17:41 This actually save lots of time.
14:17:45 So the trigger all resume. So the trigger for the kilometer algorithm, it's, it's designed as following so will we loop all the loop all the blocks in one sector, to see if we can find some sedate.
14:18:04 So said hey this hates that passage, the city threshold. And once we find the city, it will, it will be taken as the center block.
14:18:19 And we will check the nearby six blocks safe they they they have and the hits, and the in order to be to form a cluster the nearby blocks should a hatch.
14:18:32 The and the energy deposited in the nearby blocks true to be less than a cinder block, and also the time difference between the between the center block and a nearby blog for less than a nanosecond, and a clause terms of how more than two blogs fired.
14:18:51 And once we have a cluster. The, the total manager of a cluster and the XY position of the center block or the time of the cinder block and the number of hits in this cluster will be reported.
14:19:10 So once we have all those development.
14:19:15 All those familiar developed.
14:19:18 We all tested in our test and beauty, beauty to UMass.
14:19:23 We will mirror the app etc dead time with the password vault applied, and we will test the ability of app EDC to mirror.
14:19:33 Tomorrow small symmetry at the order of 600 particle bill. parts per kneeling at a trigger rate of 20 kilohertz. So this is what's PVDS condition going to be, and also will test the color winter triggers and the data sharing between two to create.
14:20:03 I'm so, so the previous layer is all about. So what we are doing.
14:20:08 We are doing and applying the in the prayer nd, but they are also potentials. We can potentially we can also using streaming readout in TV DS. So DVD as it's, you know, as I mentioned before, it's like 30 independent, the ecosystem, each, each, each system
14:20:33 is quite simple oil. It has a kilometer with 147 channels and no church or uncle with night channels. And here's the data rate. So it's, it's, it's a less intense systems for the streaming rate out.
14:20:50 But the issue is the streaming radio to resources is not included in the budget.
14:20:56 So, we don't know if we're trying to do those.
14:21:02 And here's a summary so the electronics development made for the laptop JUAR will make it possible for the solid right high triggered with small dq daytime.
14:21:26 And currently, their, their activities going on for the solid prayer and D.
14:21:25 That helps develop and under understand the capability of the dq system. And if it will want to use screaming readout for PDS, we probably need helps from some extra resources.
14:21:42 That's all I have. Thank you.
14:21:48 Okay, thank you. Julia questions.
14:21:56 Martin.
14:22:03 Sorry, I had to find the unmute button, um, let me just understand the flesh at CES right they existed, I mean like, I'm, I'm always the year 2021 right I'm always asking why are we still going with bus systems.
14:22:24 However, I mean I ATCA VXS, whatever have you, if at the end of the day, it is still a bus system right and you fundamentally time multiplex while you can have gazillions of fibers going, whatever which way.
14:22:40 And, and have it much easier so I'm, you know, what's this in like an existing thing that made you choose this or what was the reason behind this.
14:22:54 Can we tend to reuse the bigger inventor, better your flesh, etc. So yeah.
14:23:09 So, yeah, it's just to be compatible with the inventory your flesh etc We already have. If we had money to develop some new electronics.
14:23:32 And just the way that we thought so, yeah, that's the reason I might, this is David, unlike chime in here.
14:23:27 For Martin's.
14:23:29 So, so Martin, I guess the first thing is is that, well, so so yeah we have, we have a ton of these VXS crates.
14:23:38 But these we access crates, the VXS really shouldn't be considered a bus system. I mean it's a it's a dual star high speed serial, you know, interconnect.
14:23:50 And so this is basically all you know all bunch of high speed serial lanes all, you know, free for the taking to do whatever we want with them, just built into the crate the existing crates that we have.
14:24:03 And so this is really fundamentally allowing us, you know, on the cheap so to speak to basically implement know both have a faster. This is.
14:24:23 Sir, sort of getting into my talk on Friday but that this basically allows us to create a faster system that we can support both streaming and and triggered readout on much of the existing infrastructure at j lab.
14:24:33 But, but fundamentally down at the at the core that the way to think about it is that's exactly a Felix card.
14:24:42 Only it's it's it's the Felix card is RVTP, and it sits in a crate instead of a CPU and and so that's really, you know, what it amounts to.
14:24:55 So the i mean i you know i mean i'm looking at the world. Well I have, at some point looked at this, different flavors there.
14:25:03 I mean this is all this is a little bit like a PCI express it you have point to point.
14:25:10 It's all point to point the links that support up to like five gigabits per second. Yes. Okay. All right. Yeah, that's it's it's a little bit of history to this is Chris Martin i mean you know what, 12 Jeff came around, 2009, you know, kind of similar,
14:25:28 maybe not the scale of course but you know we were thinking and looking for technology than then knowing we couldn't use via me in a 20 year lifespan but commercial off the shelf would have the access that was, You know, we knew the parts we're going
14:25:54 We went from vertex fours all the way they would, you know, ultra scales now so that back plane, you know has really kept up with the technology quite well. We can scale it, but yeah no sense will, you know, will be me.
14:26:01 Limitations be used moving forward.
14:26:04 Okay, thanks. Very quick follow up question I mean like I saw the APB 25 chip mentioned there in the same sentence as high speed.
14:26:16 Well, it's not really I mean like you know i mean this was about the demon three was mentioned that I by the APV 25, I don't think you get to raid steel.
14:26:37 that.
14:26:39 But yeah, we plan to focus on JVMM three. Anyway, if we need to save money we could we use some existing electronics which has a pre 25. Yeah, I am. it's not that we don't last year between 25.
14:26:53 Okay.
14:26:56 Okay, thanks.
14:26:59 Okay. Other questions.
14:27:07 Okay, then. If not, Let's move on to the next talk. Thank you.
14:27:17 Alexandra. If you want to share your screen.
14:27:20 Yeah.
14:27:32 breaks creating.
14:27:50 Okay, my screen.
14:27:53 Yes.
14:27:55 You might want to go to full screen.
14:28:00 Right.
14:28:00 So, I'm going to put you on the street, that you can read out.
14:28:08 So, we'll do your
14:28:21 track heroes and requirements.
14:28:18 Right away Don't cheat.
14:28:24 Welcome to Universal ledger of the data rates to download and I discussed about streaming option.
14:28:32 So certainly the to experiment. So, yes, you can so electron and independent sector.
14:28:43 It takes a turn. So you need to temperatures to reconstruct the momentum of a trend.
14:28:55 And we have the CD set up setter, well you get to see when the exclusive right trend, so you need to detect both the ektron and you got to understand all this madness.
14:29:08 So, You don't know where the party. Turns out, I'll go with you to
14:29:20 figure.
14:29:22 So, the trigger right here else so that we can take precedence better if we could do more than the undercard so just to put a stake here or just to get a transpose that's what we're trying to do.
14:29:47 So yeah, so it's 52 nights so when we divide by 36, which pictures and teachers. Guest Actor and less than 200 years falls experiment to detect pathogens.
14:30:03 So as the chemtrails
14:30:08 made of.
14:30:10 So we have five kinds of gym fractures, so you can just inefficiencies different plane. So that's about hundred 64,000 calories of Kim's.
14:30:26 So, in someplace in the game expert on this but yeah you're right.
14:30:42 We have this gentleman out from the center. And we can guarantee so James.
14:30:52 So because all of the acceptance for the CDs experience or right now to finish.
14:30:58 Kim's experiments.
14:31:02 So, why don't you talk about flushing disease, or is it detected on, so can be done chunk of the beer and done by the CDC.
14:31:13 We add to that, gym right out right now.
14:31:20 human speech you will cease to function at
14:31:25 each speech, which can give you a sick triggering what needs to be triggering so looks for when you purchase which is a beneficiary the between the processes.
14:31:45 So, as a speaker. It is up to them, of this signal into your server slow Well, it can be to be decent.
14:32:00 Does it have a pretty faithful, that it can be read out to
14:32:15 the
14:32:12 power of God's Word so direct path. Well you have 10 days.
14:32:24 Available store accounts used for three doctors
14:32:28 who funded 30 million by African to catch it so that's separate
14:32:51 So you can't have fun with
14:32:42 this.
14:32:51 Some, we do so with the option is what to read you free.
14:32:51 Huge for the gym, but it's MapQuest key and we are not showing up with tensor projection.
14:33:06 So, that would need to be so focused in terms of you
14:33:17 most of the wisdom. During the fall so.
14:33:26 Be creative carrier, you'd be better path is a, It's a timeless to follow Yes, actually. So, you have a faithful.
14:33:45 She can take it, events, and you have the ADC, which can process. Do that with the action stuff don't want to
14:33:47 read outside for window you ready to Tibetan.
14:33:55 So, we have some challenge, which would then it rates, we've been getting you're using your direct output from the event so zoom when we try use the three years.
14:34:17 So in order for seniors are available your key
14:34:24 in a second.
14:34:24 And we would send that to you bone, and instead of using it to us for a class. Would you do that without to read your read section under
14:34:45 10 to 15 minutes to action.
14:34:54 So I know you've been finding
14:34:56 your chip.
14:35:14 That would be sent to
14:35:11 me.
14:35:10 So, to Roger Bannister direct mode so
14:35:17 every time valence the teacher which is detected.
14:35:22 It's going to be the 35 now. And then the second and center to the
14:35:33 add value Media Center.
14:35:42 Content director. And can you read out by your home. So right now, you're just triggered.
14:35:50 You assume a phone number six gender timing window with HB to always feel it just triggered get to be here.
14:36:04 But, yeah.
14:36:08 So what would you put the table.
14:36:12 So right now, but this thing we're missing here is the time bond, so it's going to remain free, because it were a PGA, and
14:36:34 nobody done it.
14:36:45 Oh, you're
14:36:32 so focused Yes, since it's the deep end let's start trying to get the idea, get done.
14:36:44 During a tail and make a coincidence, is the trunk of the so effective Johnny a reasonable figure 800 us with a coincidence to be trying to Jeremy john chambers.
14:37:01 For this, just three or four years, a country you better with damage so whenever we have the capital.
14:37:12 And we have paid in Tweeddale.
14:37:21 So when we use a coincidence. We were 30 minutes.
14:37:21 Rich about us.
14:37:26 So, if you like this video right, We've
14:37:35 got the rights to the gym, you can evaluate
14:37:39 that I expected from the background to tell. So,
14:37:51 don't tell them you're right to get second for GPS, to talk kind of sectors
14:37:59 and facilities.
14:38:03 Yeah, face to me now. Yeah, I bet so tonight of turn the chaos. So that's not frequency to get by guessing, so.
14:38:21 you're never too diverse hundred qq.
14:38:33 We're going to try.
14:38:33 Hello.
14:38:31 Can you know,
14:38:34 hundred Qxf trigger right so Jackie what's your cell phone, just go to Target, So, yeah.
14:38:50 But, But
14:38:49 our
14:38:55 be recorded.
14:39:00 So.
14:39:04 So, those are those rights issue we bet that we send that are you can get your thing.
14:39:11 But if we have resources to process the data, we could apply the tracking our glory, which right now is that we do find that online.
14:39:24 So, for this limited to be your point for music and for us and for artistic director it's about
14:39:38 cause and foster young girl.
14:39:44 That's tricked
14:40:03 into right to use a U turn.
14:40:02 Cheese model so in the end it's trying to put Bye bye.
14:40:03 Supposed to be prevalent as fast.
14:40:06 So fraudsters trimming right now. So, like, no, no to reduce request from the budget we had fun doing, we just assume that they're behind trigger mode, and we are recording on tape.
14:40:24 But if we can have additional resources to do some processing. If you can do the tracking things that would reduce that are significant to the effect of your full year
14:40:41 history of switching is a demon free system paper that you need to pay attention to the conduit for feelings. So,
14:40:55 I guess.
14:40:57 Okay, but you can raise the canvas, keep the doors you that are receptive.
14:41:07 So, contingency strikes them for those of you.
14:41:14 If you fall under hundreds of channels.
14:41:17 You see the hundreds of thousands of channel that propensities so much. You're welcome.
14:41:25 So, if we assume we have a cheaper like some path around your do we put up the agenda and if you end the bridge
14:41:38 that Jackie to push the thing node.
14:41:43 So that's pretty picture of a lot of data.
14:41:50 When tiptoe.
14:41:52 So, right now, you can but by the time the three dragons are, this we could consider, again, contrary to the links. So it's not too crazy but the addition that goes that.
14:42:12 So, as soon as you guys are just some parents are, you can focus your question, and Teacher, what you feel is Junior, when we get to the next number.
14:42:26 So these are great sets you need to construct your budget tracks.
14:42:34 So I put in the number or the penalty for reconstruction.
14:42:39 So that's about 2500 cross sector is the tracking. So, that's approximate because, because you're using your data from security tell us the order of magnitude.
14:43:00 I think is the most basic way more efficient to activities your professional shoots or you don't deserve to be the center of all human fi, which
14:43:14 only triggers on the person.
14:43:17 So in this case
14:43:38 So, the
14:43:49 people that are tracking your inventory for this.
14:43:56 So, you're like, now you have to stream the trips
14:43:59 to be some time so that's a very large have lots of data.
14:44:02 So that's a very large a lot of data.
14:44:14 But
14:44:07 this from the computer tender process.
14:44:14 The counter. It's the two crazy number of nodes
14:44:22 that are tracking and
14:44:29 looking at smart.
14:44:33 Notice, find a year or so.
14:44:41 So, if you do the job to push under, under to
14:44:46 stand up, people just have to unfriend to us notes. So I said to be.
14:44:54 so focused. If you go for premium support him is your community demand base is cook but he strips. The. You have to process, which attracts foreigners when she sees you can find the tracks for responding to.
14:45:17 So like rescuing Yes, for bit information back channel
14:45:24 with four channels down.
14:45:30 When we get, you get to but for.
14:45:45 So it's an inch.
14:45:45 10 me so you can just window abroad.
14:45:43 And we need to increase the number of things.
14:45:48 So, this is my boy vector addition to your
14:46:03 computer joins.
14:46:03 So we have certainly the experiment
14:46:10 to defect known as the best liner is today order right now, under.
14:46:24 It
14:46:25 Under the cross to pin.
14:46:39 He was just doing sir tracking you and reduce some follow.
14:46:40 And if you go for this premium.
14:46:43 You need to catch pause, just a conduit for the Pepsi challenge and increase the PC requirements.
14:46:57 So, that's the crazy numbers the
14:47:09 details. But, yeah,
14:47:09 Right.
14:47:10 So, Like in touch by pointing your forms of them don't. So, so we'd be well for 15 years.
14:47:33 You can get this mission to figure out
14:47:27 if you can find.
14:47:31 each ticket.
14:47:34 Okay, so yeah, that's me.
14:47:39 Okay. Thank you. Questions or comments.
14:47:44 Yeah, Jen.
14:47:47 Okay, Thanks Thanks Alex, so it's a great to hear updated from solid.
14:47:53 So I have a question or personal one on electronics the one on computing, with most of it is on slides 11.
14:48:04 Yes.
14:48:05 Yeah, so, access to execute using Wi Fi is, it's a very large refugee rights auto scale complex RPG. And it's also rather expensive by PJ and whatnot, to serve that it has to use the ultra scale complex rather than artists to do the to do the same job
14:48:27 I have to check with Ed but I think as far as the human free is sending some signal in the GPS. And so I think this should take those notes directly.
14:48:43 Yeah, This is it, is that actually here.
14:48:45 Fighting comment.
14:49:02 Alex is right. Basically, the skill series is the first cheapest one that has native seating of of the a signal which company.
14:49:01 I've seen some
14:49:04 work around for the artists inquire external termination part of the data sheets I'm hearing about that. Some people have done it with chips.
14:49:18 chips.
14:49:39 But normally company's internal termination is 128 single family. So, with the auto scale right it's, it's more expensive ship.
14:49:33 But it also has better radiation times
14:49:38 more inside that typically we have data actually for the for the diamond keep data for the single very upset, four sections and you can see definitely trends.
14:49:55 As you go from artist seven radio chips to the author scale and beyond for section for panel and decide the next thing is quite frankly provided $400.
14:50:08 So, a final
14:50:13 price discount want to watch number will be expecting.
14:50:18 If can win awards number will be expected. You have to prove also that it's the radiation mitigation and using triple or the logic in voting logic that I can correct your own office scrubbing the creation of the API that can work, and we create the conditions.
14:50:39 Your PDA that can work and installation conditions. But also, I mean, our final scheme might involve a sick, dedicated AC to do this. And we have to prototype that anyway in the next year.
14:50:51 So I think the right path is to do this year now and decide how we proceed that layer to the theatrical direction.
14:51:01 I think so, it's probably also so my next question that's exactly you also have 16 GTH that you can directly go to the 10 gb transceiver si p pass. Right, so you don't have to go through GTX.
14:51:09 And maybe I said too much to manage your question.
14:51:27 So,
14:51:23 Three times faster link speed to dance. Right, what we have, what he's showing us on file so we have more meat out there. One is directly from the FPGA can be fever, and that will be usable.
14:51:39 No radiation environment.
14:51:41 But really in a hurry environment, fail
14:51:49 to design the FPGA make the gigabit links and the FPGA will not work at the high radiation so what we've done is provided second path by lowering the sexy points
14:52:04 that community director.
14:52:13 really high radiation and
14:52:16 this makes it easier to plug in, you don't have to have a beautiful card, you have one Felix card.
14:52:22 Kayla, but if people want to try this card, you can just plug in
14:52:29 PC back with a network that we have to read our channel so it's the testing of low res unary you the
14:52:46 screenshot makes sense um so it's an interesting bring up it's a difference in religious intolerance I think in Costco tomorrow he's also going to talk about the religion test tests on our feet, and hope that so be useful to you too.
14:53:00 Okay, and cope and. So thanks again. And my next question is on the tracking speed on this on
14:53:12 this about two minutes second per second, the tracking.
14:53:17 What kind of bag on the input, what's what kind of background was it tasted ways. I remember we have created a larger background as a random hits on the gym chamber.
14:53:28 I have to check we receive it, I think it was done with put back on.
14:53:33 Okay, so that's great to know. And also, also some, some events with Hamas peace TV me drag is attacking.
14:53:43 Just feel what this event and they just drag them down the tracking speed. So the, the buffering and pipeline you to be.
14:53:51 The battering and pipeline need to be. Please be careful so he's not handling the average rate and Jessica can be very curious about this estimation, but it's good to know he's a fullback on the weekend.
14:54:05 Okay, that's not Thank you.
14:54:07 Okay, thank you.
14:54:09 Okay. Other questions.
14:54:16 No, then we're ahead of schedule, I like this.
15:20:45 Oh, Kenny, everyone back.
15:20:48 I never left.
15:20:52 Good for you. Hope you were doing something productive.
15:20:57 Okay, so
15:21:01 we got next. So we've got streaming readout for the class 12 detector and Maria and gel, Angela Bondi will, we'll talk about that. So,
15:21:16 Maria and gel, Angela.
15:21:19 Yes So Mia, good evening at all. Yeah. Can you hear me. Yes. Can you share your screen.
15:21:26 Sure.
15:21:37 Yes, sir. Can you see my screen. Yep. Looks good.
15:21:43 Okay.
15:21:46 Okay, go ahead.
15:21:50 So, in my presentation. The likely to, to give an overview of the underway have to do for lab to develop a slimming down to a desk to make it down to the acute for across to the doctor.
15:22:13 In particular, I'm a focusing on what we had done. Up to now, so let me.
15:22:21 Sorry problem with my present dish.
15:22:29 Sorry,
15:22:32 I didn't know I can go.
15:22:40 Okay, so let's meet a little bit to describe introduce the deck last well room cluster with facility, a solid the experimental will be at the difference on lab, but it is a septum spectrometer for depression after 12 gb electron beam, and it's huge.
15:23:15 Scientific program is basically to provide the substantial progress in understanding the quantum of priority Not me, so.
15:23:21 So, in the next year, the collaboration plants around attire luminosity a particular was identified the stage of the project, increasingly the luminosity of a factor to in the next two three years, and increasingly gloominess Ito factor one under the
15:23:45 in the next seven years. And do you see you can see that the son of the luminosity.
15:24:09 good a good idea to system.
15:24:11 Due to the high.
15:24:13 Right.
15:24:14 involved with them. and in this sensor.
15:24:19 The collaboration is taking into account the possibility to use of the streaming to be dealt solution in particular in this industry last here, we are the setup prototype the trigger trigger less said uq using the existing components they're starting
15:24:43 from the, from Dan good and arriving to the became the software. And this is a dq was the use it in.
15:24:54 In real condition. In fact, was performative to test on Beeman and wherever were there on electron beam or been both keys and electron beam or pretend gv impinging on our target.
15:25:11 And in this to test.
15:25:21 And then dq was that triggers dqz us at the two.
15:25:30 The goal of this is to test it was the first of all, of course, to start the first four months of this dq in Israel condition so arena test on beam. And also we're all very, we decided to use our physical channel, and basically their per day by zero production
15:25:51 from their interaction interaction of the letter memo we did a target.
15:25:58 And, to, to, to assess the performance of these of these.
15:26:10 This system. Now to let me a little bit to describe their experience experiment they'll sit up. Good. And then the dq particular here you see the part of the cluster world where the tax code that we read this team in the mode.
15:26:31 And here you see some picture of the for what tag era Yeah, for example, for work together Gary stolen. And you see here on our schema, you would have the Fort Worth that girl.
15:26:40 This is a detective the sub part of the cluster detector is a component of by trip after Kennedy method, made by lead the Thomas and crystal, or discover my device last Sunday later in the several layers of omega omega.
15:26:57 My Divine Love, Cindy later in the several layers of omega omega six. In the case of with the streaming without tested we read, just the decoding emitter and the scope of course this detector is boys, this time and talk to the tech the electron and the
15:27:26 fourth one, in, in a poorer angle between the two degree, and five degree.
15:27:21 These detector was read that with this trigger said that chain. And, as I said before, we use existing components from the starting from different then the a and arriving to the defense softer dry doesn't the face of the with the agenda to data analysis
15:27:51 instruction as now I need to be described with a different part of this train.
15:27:58 Starting from a different down the electronic different than the software as a board we use, they have a bola electronics so dismissively see the delegates listen to see and Debbie to be more usage of the color in this configuration divvied up, read their
15:28:13 flesh, the flesh, the flesh, the flesh of the sea, and streams stream the data to a new component.
15:28:22 This new component.
15:28:25 The other is our intermediate translator between the different than the, and the, the backend software. And, of course, in order to to use the different than that they're available at front end electronics in streaming mode, the, the, the, the regional
15:28:47 software so coda was modified that was modified them about that because software tried us.
15:29:14 These software was a developer originally for military master particle experiment is made by different several, several components that each of them with a specific task, and the Here you can see a diagram of the DS.
15:29:27 This is softer particular the data collected collected by the coda translator are sent to the manager.
15:29:34 The manager subdivide the collected data in times lies. Each longer 15 minutes.
15:29:42 And this time slices
15:30:05 The CPU. The trigger CPU, apply the even building, and the algorithm selection, in particular for the, for what we use.
15:30:08 First of all, l one swallow with level three girls and Edwin Vance's case consisted of a data within a time window of the 200 nanoseconds around on it with an energy, elbow, Stephen.
15:30:42 In this case, in this example is two GB and down today, and one selected events are fed to that to algorithm selection. And after that, events that they selected events.
15:31:03 They got selected invited to.
15:31:03 Some
15:31:01 are sent to the event manager and the Event Manager brought the bride to these, these, these events in the process trigger file.
15:31:06 These events in the process trigger file. About to della to trigger in our, in our setup but the cell to trigger algorithms that are Jana to plugins and what is Jana, Jana is c++, and move to thread the framework.
15:31:26 Move to thread the framework represents critical sections where we are, we have developed with the power of seduction.
15:31:34 In, in this framework. And, and the sort of the benefit of these. Is it the sweet algorithm a developed the in the in the offline analysis are available also in the online, construction, and in this way you can.
15:31:55 You can write the euro euro all good it's more accurate at a selection regalia based in Beijing on on on the on the on the record section. And moreover, also their translation table that the calibration courses that are available in this away.
15:32:18 Our agenda plugins. And in the you can have a different server condition server alert Jana plugins so 750 area, and you can run in there, in the same time in parallel, it is different their selection criteria.
15:32:50 And this is what we've done. Now, in order to do in the face of this, to this word.
15:32:59 Try this plugin, was developed but do not have to treat the face of course this is Jana, and that right yes.
15:33:09 Let's meet a little bit to describe what we have done in the first in the, in the two in the two in the to run into two tests, and a little bit I won't lie, I would like to show our results, particularly Are you see some results for the wrong one.
15:33:35 So the first round performant the in.
15:33:40 In the winter winter of 2020, and this case that we use it an electron beam impinging on our team lead to target them.
15:34:01 In the first round and we read, just the the calorie meter so just three under 32 channels. And for this, for further, further on one, the offline analysis was profited don't indemnification of a defined zero events.
15:34:10 Now, these are these are, these reactions so this physics channel is isn't a simple because the, the, the gamma gamma decay particle, and you can detect the with the same.
15:34:27 We do the same detected basically they're called the different words they get calorie meter. And in this way the experiment that sells out.
15:34:39 It is is more simple because it just is just that the factor.
15:34:47 And for the offline analysis that we use, we apply, of course today. The same medical staff actually use it to India or line that a construction so in their, in their library construction of basically reuse that offline or dare dare dare dare to construction
15:35:04 to framework. And the only see some results that show the quality of the data, in particular use here you see the number of the, of the ethical stuck the class, there are pretty bad sir.
15:35:19 And you see that today for the most of the events.
15:35:28 We identify just one class. And just for four models less 10% of events, the DOD two or more clusters.
15:35:41 But events.
15:35:43 Here for example you can see that distribution the position of the cluster. And you can see the footprint of the calorie meter, and this way you. Okay, is the data, big data, I suspected basically data we thought was on the dare to cluster events, and
15:36:06 we calculated that environment.
15:36:30 we see too big, but it is not suspect is is not strategies if you to the experiment that's about for basically.
15:36:42 Yeah, for example, you see the schematic view of the target to use it, they use it into test.
15:36:51 As I say the before in Now what does that we use it with the electron beam impinge on the on the on the on the lead to target, but the downstream that made you sad but there is a
15:37:10 window, made by aluminum and disease, another target for us, basically. And in this way the first speaker is the other.
15:37:24 Master is due to the production of the idea production interaction of the electron beam agree that the principal with the domain Lee.
15:37:34 Let's start with it. and the second speaker is muted interaction, the production of the Divine zero, from these the second the aluminum thought, of course, that position of the law where everybody in mass is due to the wrong assumption is that the best
15:37:52 exit is located that after the lead with God with the position when we calculate them body mass.
15:38:00 But basically, we, we, was able to, to, to see a clean physical senior from the, from the data for the first for the restaurant.
15:38:14 Now, let's meet.
15:38:16 Move on.
15:38:18 Sorry. Okay. Yes, another think about that on one for the brown one data analysis offline that analysis, we was able to to developer
15:38:36 casting the sweeter beta down here.
15:38:40 gusting the sweet cerveza down here. And here you see Kobe's on between what we have painted that line hit their AI class setting the angel read the in red.
15:38:52 And what we are painted with the, with the sound of the glass seeing that in the motor less today than there is a season in agreement.
15:39:05 This.
15:39:06 This.
15:39:09 This is sweet, sweet developed in the, in the offline analysis. Dan was using this in the tech on the ROM in the second test performance during the summer of, 2020 factor we run real time idi, and we use, we use AI class setting to to develop, to, to,
15:39:36 to, to build our selection criteria to sex and pretty.
15:39:45 So, about the sacred ground. As I said before it was before men in the summer of, 2020. In this case, the experiment of setup was a little bit differently because we use, again, an electron member.
15:40:02 After 10 gb, but in this case, the target was differently because, because we use it the Arctic.
15:40:11 Gosh. To start, and then FEMA aluminum target and.
15:40:20 In this case, we increase the number of channels, because we don't have God.
15:40:38 God's glory meter, but also the for what are the scope of this drama was that the goal of this was that not to the first of all to study.
15:40:40 Detroit so profiling and study the platform of the system, changing a little bit of the condition so basically she changing the day from Dan the threshold that, that one threshold and started the best for a month to changing the day to selection condition,
15:41:04 and about the experiment of sitar. I said before is it was a little bit different. And for the physical channel is a little bit, be the disadvantages.
15:41:16 Why because, because in this experiment the setup, but there was a third version a divine dead and funk Center, located there in front of the death for what do you see a schematic view of the, of the customer detector.
15:41:39 So let's see here the city Carla, the target and the you see here there was a tick absorber and the tickets are were shared the most of their for what they got a sentence as well.
15:41:59 Was it in this way, is that is that these are the two to see the physical sound and so they're quite zero production. And what these these disappear we are so the the shift is that for a second.
15:42:15 And is it is it is, is visible. Already in their whole life scale. So, for example, Here you see the flexible see scalar or for the, for the calorie matter, and you see the events right there most of it even, even three comes from the outermost crease
15:42:37 outside of like a really major, and the offline analysis for rare.
15:42:45 In this case the
15:42:47 performer that using the record the channel. Jana to record instruction show this, this, this behavior. So in fact the year, you see that clustering the position.
15:43:04 And you can see that the most of events that are associated to class there in the outermost decreases, that are machine that the by data caps orbit.
15:43:21 And despite the, The disadvantages experimental condition.
15:43:25 We try to calculated to be with him but infamous for the two gamma gamma, and the are you see the results, and those who is engaged. Is it possible to see a peek around and then must invite myself debate zero above, of course are very huge background,
15:43:47 but also in this case.
15:43:51 Despite the disadvantages the condition is it is possible to select a clean the physical channel.
15:44:04 So, in conclusion, we, we have.
15:44:10 We had set up but our
15:44:18 seemingly doubt, seemingly without dq system, a prototype disintermediated our system.
15:44:28 We are using the heat into successfully on being tested. And in this test we are the readers swimming, the more that are back to the class 12 with the texture in particular the tiger calorie meter and the footwear and for what that scope.
15:44:48 And we are, we are, the phone today. The analysis in particular for the second around, but from the results that we already have already obtained it was possible to extract clean physic senior in form of five zero embodiment must speak from this data.
15:45:13 And let's see, thanks a lot for your attention.
15:45:18 Let's see. Thanks a lot for your attention. Thank you.
15:45:20 Questions.
15:45:29 No questions.
15:45:44 Okay.
15:45:35 Were you able to do a comparison, sort of before and after you switch to streaming reader.
15:45:45 And yes, we have.
15:45:47 Yes, we have. We have take data, we did the sound of the two system or so forth. In the second day in the second round, and the analysis that is undergoing is ongoing.
15:46:02 Okay, so you from the preliminary analysis is enough in in agreement. Oh, very good. Very good.
15:46:09 Okay. Other questions.
15:46:20 Okay. If not, then we're a little ahead of schedule, but let's go on to
15:46:33 Alexander.
15:46:35 Come soon and go sorry I'm reading the wrong.
15:46:41 Timing entry, Sergei boy not boiling.
15:46:46 Okay. Yeah, sorry.
15:47:03 Okay, good.
15:47:07 All right, good. So yeah, I'm gonna give you a nice presentation on our first experience was a three month.
15:47:14 And, you know, just on the small portion of clubs detector was used, it was a for tiger and forward.
15:47:22 And the last detectors surrounded by cxS crates.
15:47:32 But the entire class, much bigger, and I will tell you what we're doing to get credit for the entire class.
15:47:38 get from our very sophisticated, level one. It's a trigger to some some quiet for statement, and probably a few remarks on the stream and result in general in Costco.
15:48:04 And one general remark so it was relatively new experiment we started three years ago.
15:48:13 So all our electronics and the trigger system will just built on the sensors, not all to we already have some nice features already implemented, and that lets us.
15:48:37 Switch to more modern litigation Islam for the Quran already for a longer on the time. And the vicar of Mr. We're already running several years and we will around, probably two three more years with the same system so if you're not super us to do changes,
15:48:47 and the same time we have a boil. And we know what's going on.
15:48:52 And I don't talk about this until we get on credit for that, and the statement is not our next step. It's our second step, But we don't get the.
15:49:06 Look Mr.
15:49:07 So first for content electronics and slideshows general diagram for our litigation system. So currently going through, there's about 30, years.
15:49:22 And the.
15:49:39 Our next great have to bring costs 200 kilograms, that what we have to do in about two years from now.
15:49:51 VTP both installed. Those crates, who participate in the degradation. So we already partially walk a path to can kind of freed from Vimeo, Vimeo Buster though, to, you know, more the ones that are those using federal lands.
15:50:05 So together we have two great elements element and components which are still being me and have a low performance, lower than hundred delegates.
15:50:18 And we have three, three places we have to upgrade first.
15:50:32 We use MKMT dishes.
15:50:29 And we have 25 immigrates America installed but most of electronics and those credits discriminated.
15:50:35 So, actually to install judicious we need.
15:50:39 If we combine them in one crypto delicious on one trade we need on the six, six grades six, maybe eight dependent on cabling so that our goal what we're doing now we are very drinkable we integrate to be accessed right, you're replacing Bitcoins.
15:50:56 And we're replacing current issues with Joe have made the dishes.
15:51:03 I will show you a few slides after the details, but right now I just want to give you a summary. So second place where we underperform is micro mega current micro macro though to based on the dream chip and the week in Syria, we can get to 50 to something
15:51:26 like 54 girls.
15:51:27 But the different level know but in Costco contract and beyond.
15:51:30 So we're comfortable with micro mega team and the effective for exchange of information, and currently we consider in modern Isaacs.
15:51:41 And we want to build new window for micro mega, which will allow us to go to get us on later to go to streaming. And of course we consider NKVMM three, which was kind of several groups already considering, and recently we started to also consider Samba.
15:51:58 It's all driven by availability. Last two chips, being considered a tunnel up by other groups, and for example we already invested in some AC production, some money like committed committed money.
15:52:14 And so we're trying to stick together with other groups to minimize you know your hit and so results will be benefited for several groups, and the BMI of 30 on some points to ice AX, which actually consider in a lot and just trying to stay with other
15:52:30 people and considering them.
15:52:33 And I think it makes sense. Those two Isaac Simpson most of the ones.
15:52:39 For the moment, and it will do what what we need, what we want to do. And the third component as it was discussed on previous
15:52:47 about solid was a part about sort of the director, Vp board, which was originally designed on color photographs for trigger purposes.
15:53:03 It just happens it can be also used for.
15:53:06 And that's what a nice side effect of, take your system which we built and the.
15:53:14 After we install VTP credits at the bottom every of the access great. We can complete London readout from Jeanie buss and switch to remote from Cheryl and using Sarah lines from every news station.
15:53:32 and the Spirit can be up to think five gigabit per second from our station, which is much higher than.org me right, just like 200 megabit per second.
15:53:43 All right, so let me show you a few slides is details. So, the slide shows this veteran leadership board, which was designed in a lot.
15:53:59 We need the board for journalism, and we already received it. And we already upgraded biplane and six grades to access. So we are now able to say one sixth of our class called detector one sector.
15:54:21 measure. The most important There's our time flight resolution. And, unfortunately, we cannot measure it was a cosmic very linear the beam. Was it.
15:54:42 So, our plan is to install, electronics, and finally we'll get him. Next time, maybe in the fall, they will take data and probably to be safe to switch back to km, and then after data analyzed and we will make sure that everything for him then we'll go
15:54:44 get them to play some third system.
15:54:45 So we always trying to play so in that sense. So total price for place and tissues, founded at six o'clock Okay, it's estimate.
15:54:55 And just remember good don't give us money chance we're not in a great mood, we don't have a great moneymaker from the traditional money so partially using capital for money to do changes and the problems and when we have to switch to federal Federal
15:55:10 Minister and we will get some money from from difficult so.
15:55:15 So next slide from Michael McDonald.
15:55:19 Again, it's summarize some parameters, mostly came from a local development also from a micro mega people, and the price estimate here about 250 K.
15:55:33 So I don't I'm not sure if it's maybe it's a little bit more but at least what people estimated.
15:55:40 So again it's not not not a huge money and they feel decided to go with a sample, instead of VM I'm sorry.
15:55:48 Well probably till the price will be comparable.
15:55:55 So that project not started yet still expecting some published books available in all up sometime soon.
15:56:03 And I think it will be report on this workshop about it. And the, the me based boards may come in the end of summer.
15:56:13 And we all will be offering them on the test different detector so hopefully we can get at least one work of which came in close to wealth to get real data from a micro maybe and see how it looks like.
15:56:23 Because there is a lot of issues, and the micro Monica is not simple detector if you have to be careful in building for those sort of electronics, but since our micro mega team fairly involved, as in every single mobile phone.
15:56:43 And my last slide about VTP board.
15:56:50 Again, it was designed for trigger but nicely can be used. Also for adult and that can go up to four links 10 gigabit pitch.
15:56:57 That kind of verify right I think we have about 60. Great. Remember if you multiply one to another theory future right and will not expecting that much right even instrument mode.
15:57:13 Right. And we're not expecting that much right even instrument mode. Because I was done with about 50 gigabyte per second maximum.
15:57:27 Alright on for 32 by remaining, which appears many of the world from the key.
15:57:24 So, we're all, we're talking about, let's say, 400 k, maybe, and during several years.
15:57:32 And let's kind of reasonable money and hope to if don't buy everything will progress a lot on that direction.
15:57:44 So that's kind of my summary both front end.
15:57:47 So our first step will be upgrade 278
15:57:53 accelerate, people calculate that on simulators, he forgets its water word but to be safe, set the goal 400 plugins.
15:58:19 So I will still stay in the database mode, because we. If you have to switch to Canada to get it from two years we will not definitely have any some credit for streaming in a global in both background and our professional calgarians, and the Sophie and
15:58:21 the 400 target author and totally unnecessary to target them What can I can do for him on that level if you go beyond counterterrorism yes you probably better to switch to streaming.
15:58:34 So most of our board skills DKMTDC celebrity replaced my grandmother will get new though.
15:58:43 We have to check also Silicon Graphics performance for color minister on me because it's another task, we will do.
15:58:52 Hopefully we can use existing electronics but just to be safe to do some special test.
15:59:00 Then some room I started to use some VIP symbols trigger and without not even 400 gets triggered system. There is one is that we already filled couple grades, with only one book installed for in third grade, because single board over the story with me
15:59:19 both boundaries. So it feels face to Wikipedia though we can minimize the number of credits, that's already nice.
15:59:26 We might need some upgrades and dedication called the software.
15:59:33 And the will work in steps, as I said, this fall, will know how when you get issues before.
15:59:40 And then, competent next day we'll get some prototypes for micro mega, so we will implement those changes one by one. taste and every cent very carefully.
15:59:49 And the lessons.
15:59:51 I think a good strategy so total time.
15:59:54 And I would stay for two years of course it might change.
15:59:58 I don't think it will be shorter but it may be longer actually depends on many factors.
16:00:05 And then after it's done and we're ready for kind of upload your trunk, and we're running could convert orders we can start to pull them for streaming.
16:00:20 Of course we might build some new, new Vimeo electronics in the next few years, but it's also the second compatible differently.
16:00:29 All our new electronics development, go the experiment comfortable from now on we will not implementing electronic seven and triggered mode, which cannot be used in the future in a statement.
16:00:41 So exactly that configuration for cloth will flow statement will be decided.
16:01:02 Later, because right now since we don't have background.
16:01:00 Already, and several group development several frameworks and plus network equipment development very fast and the view for today, different groups using different techniques, I was a multi stream multiple files and whatever.
16:01:06 So, although some businesses not settled yet so we will just throw the wait and see what other groups will do.
16:01:12 And then few years we'll just pick some solution convenient for us.
16:01:17 So in your stream and Pearson of coders needed being developed. Hello, and thanks Gail, I would say, it's sort of the five years on three.
16:01:37 Most likely, but who knows.
16:01:38 Again, as soon as Canada gets greatest finished, we will be ready to start seriously think about streaming.
16:01:40 So my conclusion about Phantom Phantom electronics have a great instrument to Steven tomatoes on the way, and they will not anticipated Can you see those problems.
16:01:57 My best optimist.
16:01:57 At least, right now I don't see any problems way to be stuck somewhere.
16:02:00 Alright so now my next section about trigger and how to migrate from our trigger system to streaming system.
16:02:11 I can't say that without very nice and very powerful trigger system based on wikis, which again are specifically designed to be used in classical in the Gospel trigger the system that slides for example, so electron trigger works for several components,
16:02:32 participate and configure first first initial chunk of is differently differently electron and also the chances for the tetra gives you time for your trigger.
16:02:43 Then we have a brief chamber which consists of six, super layers.
16:02:47 And notice what better way to do.
16:02:51 We call it a segment find the seven funding process. The next stage of triggered the link between those six layers and it's done by road dictionary.
16:03:03 And then another part of trigger.
16:03:08 Search for clusters in the electromagnetic perimeter of the photos.
16:03:13 And on the next stage of triggered system with the geometry match between the chamber and the collimator, and also in the process of course we're doing time in case it cancels between different verticals that system because a lot of different parameters
16:03:29 you can do.
16:03:31 Now of course it's time uncorrelated parameters when you do your delay curve to set correctly time delay for every component has a energy related parameter if you can set a threshold for peak, and then threshold for integrative of your big.
16:03:51 And so anyway, it's a very long list so system quite sophisticated.
16:03:57 And it works very well. So the lyrics as it was designed.
16:04:03 This is a the diagram for trigger system.
16:04:05 And as you can see everywhere it's that VP VP VP so everything was done on the VP, except stage three were a previous version of trigger words so speed is used.
16:04:17 That little bit less sophisticated but on stage we're not doing clinical author and Co. Director He can create just doing mostly 10 cases, so we don't need any sophisticated refrigeration.
16:04:30 So that system works for him, and
16:04:34 go to next slide, system full operational efficiency close to 100% with have quite developed system and prices were to estimate trigger efficiency, we're repeating this procedure every few months.
16:04:52 Every time when we changing something configure configuration between run periods, and the efficiency stays close to 100% that can be 96 97% 98, but we're close to the purity of those three gear depends on, you know, around, around period, but the number
16:05:14 good around so stays on average about 50%.
16:05:15 Sometimes some can get to 70% sometimes go down to 40% but the average I would say 50%. So that system was very much.
16:05:23 Good enough, I would say, and it really it system allows us to run without significant changes for empirical physics problem.
16:05:34 But, of course, as soon as we will start to use it.
16:05:37 We started to create a list of Samsung which will be good to improve.
16:05:42 And here I put some, some of those improvements which are waiting to be implemented. I don't want to go into details of that just to show you that it's fairly typical a typical stereotype related to some German translation related to some multi particle
16:06:00 clusters.
16:06:02 Time consistency improvements, so very much. Every part of the system can be improved, in some ways, and the, of course, the question is, what to do, to go back to our FPGA based system, and with GPS and start to modify firmware, or just to implement
16:06:25 software, which is by the way me something like collectible because of number of reasons.
16:06:30 So, current particular system solely based on a PJ, there is no loyalty component.
16:06:45 And they discuss that and we decided that, don't touch exists on a level one trigger for now at least one of the reasons, several reasons. One of the reasons that entire level one thing you have to be completely abundant when we go to statement.
16:06:51 So it's unclear if you will, you want to use some solutions in streaming data processing or giving a completely new solutions.
16:07:03 And another reason of course we want to try new technology, first of all, machine learning and artificial intelligence and the level three is a very good place for that.
16:07:17 And in another reason. One reason of course as much easily easy to develop something called the software to modify firmware it just takes a long time.
16:07:28 the beam do with additional procedures again.
16:07:40 And that kind of it there is key.
16:07:46 It's a risky. And it probably it's unnecessary we really want to go, go forward from the implement Samsung software based.
16:07:52 And another nice thing about software a trigger you can run it and thinking mode so if you implement some algorithm, you can put results into data stream and I'm always I'm offline.
16:08:02 On slide and then see if your engine. If your searches for it you can start to use it as a trigger. So, anyway, done a level three component was added.
16:08:13 And the older girl improvements will develop their.
16:08:16 So,
16:08:21 and as I said, it will not only allow us to implement our improvements which we want to implement but also we can develop solution for future tremendous physically with solutions still can be configured bundles it idea but since we know a lot about the
16:08:39 a lot about the PJMO very few about, you know, software based triggers, so it's probably it's time to learn more about it, and then we can, when we know both sides we can decide what more precisely what we want to do.
16:08:52 And the initial test for level three of you
16:08:57 that have chamber segment finder and machine learning curve with tech tech fund.
16:09:03 Why is it there chamber is traditionally the most troubled component and closed world, if you go to Carla minister and simply because of acupuncture acupressure becomes so clear that even a flying dragon becomes less and less efficient so they completely
16:09:22 rewritten over multiple heats and segments from different services. And if we can implement some kind of better selection for example or we can give a better prediction for a second position of Director position it.
16:09:41 First of all we can decrease the rate data right and the ones right and also we can hope for improvisation.
16:09:45 To speed up offline processing, and that will be even more important than the streaming that when we have to do, going
16:09:57 live going to spend a little time on that slide and explain you would do for fun our direction.
16:10:03 So, I will Def Jam or concerns so on.
16:10:07 So parallel equidistant parallel of six layers and hundred 12 words and every layer. So it's a matrix so one foot by six, and the current trigger does not see those circles, which corresponds to the roof time current trigger Europe so based on the sheets,
16:10:24 so it takes center of every circle center of basically our position, and the found found the segments, and those segments are linked linked into track on the folding stage that very, very nice for this example which I should we, but in reality, we have
16:10:42 to prominence first.
16:10:44 You might have a bunch of segments and the single hits around this area, which will be older it.
16:10:57 Secondly, we're more complicated problem inefficiency sometimes instead of six.
16:10:58 You may see five for for kids in three shoot sometimes.
16:11:02 And that's become certainly.
16:11:04 So your selectivity of your older it drops significantly was that inefficiency.
16:11:10 So that's why we decided to try to use the time and brief time on very early stage, and hope that they will give you a more precise second position a more precise angle.
16:11:25 And, in theory, so they'll let you
16:11:32 be more selective, then just hit post and the
16:11:38 to implement that the with of three kind of operations, which have to be performed on whether we use GPU CPU FPGA first iteration.
16:11:50 You put your date in some kind of great and you do shift and some operations that's very easy for PJ, and probably it's also not a problem and the regular CPU.
16:12:01 But then the second stage you have to do pictures in two dimensional array.
16:12:06 And that operation. Well, for example, seven groups as a function to search for two dimensional to pick some to the mystery but of course you want to have something very, very fast.
16:12:18 And the final stage is a linear feet.
16:12:23 And the, what we will do in a folding mom sort of months, we will explore how those three operations can be performed on FPGA on GPU and CPU and the in process we hope to learn a lot about those devices and in the fact that we can pull in our future strategy
16:12:41 more precisely because right now I personally don't have much experience.
16:13:02 you need some were sophisticated algorithms, unless you want to build villains mode forum is probably not a very good idea.
16:13:12 All right. Hello threat trigger, as I said that's implemented.
16:13:16 Currently it's implemented in our version of using.
16:13:20 So it's based on our one transfer system on the old we can build we can build the chain of Protection Units, the chain can be as long as we want. Well, it's a reasonable price of course, and in every not substitute to 48 threads.
16:13:41 And the, you can imagine how you can organize your protection for example and just not you can do segment find them on second point you can do dragon concert mode you can do some cut and the final target issues and so on.
16:13:56 I think it will do some talks about artificial intelligence, I will not talk about it but to test our machine learning algorithms we are going to set up your visit to GPU cards.
16:14:10 So that server will be delivered, probably in few days, and jalak computer center point exactly the same service, those servers can hold up to 16 GPU curves.
16:14:30 And it will be really powerful Protection Unit for us right now to save money we decided to just two GPUs, because if you know how fast it runs on turns up if they can extrapolate for 16, and phone number of nodes in it.
16:14:38 So that's what we'll do in
16:14:46 full length Time.
16:14:47 Okay. Our last set of Section worship, what they want to say well it's $300. So in few years.
16:14:55 Hopefully, colossal sometimes electronics should be ready for streaming coverage.
16:15:01 As I said, upgrading to country together and we keep in mind stream uncut version.
16:15:06 Welcome supposed to be developer as a group so we're not dealing with background, we just upgraded front and Kobes that job group will deliver something for us.
16:15:15 We are not pointing to use a favorite system. So, one time will come we will speak switch completely from trigger to streaming doc.
16:15:24 And I think it's the right decision otherwise it becomes too complicated and their illness.
16:15:30 And the last thing was a cloth cloth Well, we're not developing.
16:15:36 Also, but we provide them test bits for streaming development from relatively small to setups, with view access crates but Miranda was thinking about, and they can provide up to like 40 credits cable system.
16:16:01 We're full scale streaming, that can be just because as I said the risk of right now 42 we typically a BGP both installed.
16:16:06 And the way I thought them to kind of systems because the flesh agencies and with our words. If you drop, for example, threshold and goes to device the flesh edition The, the chamber doubt, you will generate huge amount of right and the little bit difficult
16:16:23 the test to become solution to say how much you can process.
16:16:28 solution to say how much you can process. And so, we're ready for that.
16:16:31 We cannot deliver tomorrow but I think if we have to do it and we will request we can do it in is not very big money and scale a few months, and then your for your current system will be ready for testing.
16:16:46 Alright, so my conclusion Costco from time upgrade.
16:16:57 Blend was in progress. It was a purpose to encourage them and try to convert your guess is exist and figure out mode, and to make them comfortable with this drink.
16:17:09 preparation will be developed and tested and frame of new level three trigger product. And it will build them in for a more current target population.
16:17:15 Think go, a level three is, if it doesn't fit for any software solutions you want to test.
16:17:20 And we will be working closely with Zilla, and also groups systems and business streaming that become development and providing testing facility.
16:17:29 Alright, so let's go.
16:17:46 Okay, Thank you. Okay.
16:17:36 Questions to Saturday
16:17:46 Jaan, or sorry, Jen.
16:17:50 Yeah, it takes a very, very close. Yeah, anyway.
16:17:56 And so my question is, it's a four bits of open ended a discussion. That's a photo community of this workshop. And I think it's very useful to bring up surfing and photo full screen reader for how it's kind of like for Isagenix where you find out by increment
16:18:21 And therefore, my question is for cast Well, I'm sure there will be also physics opportunity that can be accessed through streaming deck. And I'm going to put your coming down dogs, and whether that, whether by emphasizing this community that would help
16:18:42 you to get to the founding needed which was a new hardware that's up with the things that they're good at system because the limitations and not every physics can be
16:18:50 executed right.
16:18:53 Yeah, so I'm just wondering, where's the data or example from Costco. That could be highlighted.
16:19:01 When we discussed the streaming readout upgrade.
16:19:13 That's, I think that's a very good technical arguments. So I'm wondering whether that's also a parallel physics argument that can be used when the community it's moving closer $3 million.
16:19:21 Well if you're talking about like technical specs on both for kind of a Philadelphia, that much right the main reason to switch to streaming. If you're talking about physics.
16:19:35 I don't know.
16:19:39 One example when we were limited trigger system, but it was not related to the fact that we're an integrated system, it was related to the limited FPGA size for progress not very good example so far.
16:19:56 I think the key is a.
16:19:58 But again,
16:20:01 you can imagine the trigger system is a level three only but the trigger the system goes level three only the basically streaming system.
16:20:09 So, I can imagine my data management strategy,
16:20:17 completely abandoned, our target one level one trigger and the fish to level three.
16:20:23 What we will gain.
16:20:29 Yeah, I think, whole data processing can be organized, but, for example.
16:20:37 The example I showed you is the number of construction.
16:20:50 The reasonable time and collaboration, basically zero to get your drift time. The time to distance. And in the current system because way it's the code signal that when you're doing this through the roof from the trigger, you don't have to zero, because
16:21:06 the zero exists and different block of DJ. When you're doing everything together in level three or in like streaming data processing. You can organize everything such a way so all needed information will be in one place and it's much easier to organize
16:21:20 it this way rather than to do it on a level one because, as I said, level one took quite a lot of efforts to to develop and the one is developed when it's implemented.
16:21:34 It takes a lot of time to change something, for example, one good example, our cluster and clustering figure in the Canadian pedophile electromagnetic color interface design to fish for electrons, and the because of the limitation on FPGA, we have, we
16:21:51 If we can keep one of the four clusters.
16:21:56 And we decided to keep philosophy as a maximum manager, because for electrons that's what we need.
16:22:01 But late in the process of for some other physics. We need to also get closer for a minimum ionizing particle, which obviously have a much smaller manager.
16:22:15 And she also has a number of clusters solar the limited by four, you have a nu, if you already feel for beginners or you feel good chance to lose a mini womanizing cluster.
16:22:25 And there is it was a one of the requests to change. Unfortunately we cannot change it because we learned this FPGA vertex seven we use them already on some 90% something busy which is crazy.
16:22:39 And the, we will implement that in level three.
16:22:42 So that's one solution which will be easy to implement a level three.
16:22:48 But again, I'm not expert on the wife I was back on the organized on streaming, but I think it's important to be able to process to keep all knew that information in one Protection Unit.
16:23:04 And right now, for example about developing a weapon the people showing him some others words, people doing partial aggregation so which means, because of the limit, limited will do is you can
16:23:17 read from some subsection of a detector.
16:23:21 And, again, the same problem different processing units don't don't have complete information about your event, about from all components of your detector.
16:23:31 So I don't know how to organize and streaming.
16:23:35 But it will be great if we can give to every Protection Unit complete templates from enter detector is that to build the ideal case.
16:23:47 I don't know if it's possible to do it with current development of technology. So like, you have, let's say, 64, Microsoft a millisecond from entire class detector and it goes to one computer to process and the next 64 millisecond goes to another computer
16:24:03 on the next computer and so every computer will extract certain number for every time flies, and also once we have the complete as they don't have to talk to each other.
16:24:14 So, in that mode will the ideal I think I'm not sure I hope we can get to it in like few years.
16:24:23 Well that's what what I can say, Well, I agree with that technique argument. Yes, end up competing with a time block is look like.
16:24:34 Even the building by the window streaming scale is an interesting problem here.
16:24:38 Thanks for the comments.
16:24:41 Okay.
16:24:44 Alexandra.
16:24:47 Alex now.
16:24:50 Oh yes, you have a rough estimate of how much it would cost to live it free.
16:24:57 Not right now, because we have to mother Oh, already on this machine with GPUs.
16:25:06 Then we can estimate how many of those we need.
16:25:09 So I think we all know this number in few months, but not now I don't want to speculate.
16:25:14 Okay.
16:25:21 Okay, any other questions to Sergei.
16:25:30 Yes, I, I have a question. Let's assume you remove a level one. There you go to a thought, and though I rely only on level feet, what your beta rate will be extrapolated what system was a famous model, what will you expect the data right, so we're level
16:25:46 three forum forward to, well estimated data rate will bear on 50 gigabyte per second.
16:26:03 Okay, since we are not going to remove for level one, working with level three level four level one only when we switched completely to streaming, and instead of level three will have something, but I assume.
16:26:11 Yeah, Mike, which is a war consisting of motto be 50 gigabyte Yeah. Yes, that's a yes estimate for her luminosity for intercultural.
16:26:31 Okay, any other questions.
16:26:39 Okay, I'm just one question from personal.
16:26:46 The.
16:26:50 Will the you this upgrade to the class 12 readouts is going to go on over several years from what you say.
16:27:08 Will the physics that will be studied with class 12 remain the same.
16:27:11 Well, that's the question whether the physical groups, as I know we have a button for next two to three years, even before we switching to her luminosity and after we will switch to Carla minister a few material winners a few years.
16:27:26 Otherwise why would switch.
16:27:29 So, I think my favorite put over what I know is I don't know much about it since I'm mostly involved in, you know, data protection can litigation but as I know we have a good program for some number of years.
16:27:42 Okay. No, I was just one to check that there was a way to compare physics performance under, under your old data acquisition with your new data acquisition system.
16:28:00 So we're here, I expect you're doing the same physics but a tire luminosity so now.
16:28:10 Okay.
16:28:15 So that's it for today's session, I just flash up the agenda for tomorrow morning, where it's
16:28:38 hardware vendors and chip manufacturers, talking about the the latest products.
16:28:38 So, that starts tomorrow at nine in the morning.