| 00:03 | the notes to see. I don't a bit about that. Not only |
|
| 00:10 | do the programming of it, but are the different architectures it's or it's |
|
| 00:17 | a gene. You somewhere there. . And sometimes it works. |
|
| 00:22 | then sometimes it doesn't work so So the view's not the universal |
|
| 00:31 | Carol. And this is kind of computing device. You don't have things |
|
| 00:38 | memory accesses and streaming choice. right. So And I saw talk |
|
| 00:53 | and a bit of a police So one thing, the first |
|
| 00:57 | Why should you be interested in the before your committee? Nevertheless, |
|
| 01:03 | I'll give you some scope here. give application accepted Post talk a little |
|
| 01:08 | about Thanks. Oh, Rene Army is that science is this chemistry was |
|
| 01:15 | drove the development of computers for That case on engineering started to |
|
| 01:25 | uh, specific example slides then too ago, very much. In the |
|
| 01:34 | decade or so years, business intelligence starting to become money talks update on |
|
| 01:43 | , right? Yes, but it some time ago. A telecom companies |
|
| 01:52 | oh, and figure out how to pricing schemes. Do so no. |
|
| 02:01 | . Now there's a morning No. . So much data is digital service |
|
| 02:05 | or things are accessible. And I say the most recent one is not |
|
| 02:12 | real time applications. So I will to give examples of all of |
|
| 02:16 | Get something fun, right? So sudden is where things depends on |
|
| 02:29 | And here's my God, one of examples specific ones. And why also |
|
| 02:39 | actually trying to save businesses usual doesn't any more. Nuts follow. One |
|
| 02:46 | , goes up. I wouldn't say couple, but Intel struggles to figure |
|
| 02:53 | what is coming next. So, , things are changing quite a bit |
|
| 03:00 | terms that used to be about 30 or so. That was kind of |
|
| 03:06 | on the microprocessor, everything news. processors very good at it. Processing |
|
| 03:15 | . And for what? It's come stuff. Now things are not department |
|
| 03:21 | , so now there's kind of more in terms of future architecture. In |
|
| 03:26 | ways, the success rate use is element of it. But even on |
|
| 03:31 | sink there's more diversity. It's happening for that fun. The options yes |
|
| 03:40 | is trying to figure out what to . Yes, you may have seen |
|
| 03:44 | buying companies of all different kinds of where so I think it's also to |
|
| 03:52 | it's important moving forward things. You . So I know where there's |
|
| 04:02 | Chapel. So that's nothing said on like you keep on something characterized self |
|
| 04:10 | Car suspects the supercomputers on Well, order to do the processing, |
|
| 04:17 | processing, navigation, role of the of itself on its A very much |
|
| 04:23 | that needs to be processed. So on top? There is a super |
|
| 04:28 | or yours a little bit what it , Um, next. So the |
|
| 04:32 | on this pump your is about to courts. So one other things. |
|
| 04:37 | do you put two tennis courts in the car? Doesn't kind of |
|
| 04:41 | So you need to shrink a So that business as usual does not |
|
| 04:47 | . I think this computer driving these No, no, it's not |
|
| 04:53 | This is not just say what Okay, This kind of cape |
|
| 04:59 | this is what you have. The driving car it would just package in |
|
| 05:03 | little box in the car. How you do The other part is, |
|
| 05:10 | I said to, I guess you the victory about Yeah, The volume |
|
| 05:15 | about 700 square meters, including all infrastructure for running this computer. |
|
| 05:21 | that's one part, and then it to get under the hood in a |
|
| 05:25 | box. So then cost is another , this type of super computer. |
|
| 05:35 | so it's on. If you're going have it in the car, maybe |
|
| 05:39 | can cost 1000 bucks, but not . So there is a significant reduction |
|
| 05:45 | price necessary in order to get We really need that much. That's |
|
| 05:50 | people claim right now, whether it's enough. But that's in the notion |
|
| 05:54 | supercomputer. So this is Saskatchewan, example. Okay, get the scale |
|
| 05:59 | things that needs to happen. Something that. Yes, the energy consumption |
|
| 06:03 | this computer is about 10,000. The slides out wrecked, so So |
|
| 06:09 | you need to days saif the need reduce and is your consumption. You |
|
| 06:17 | to reduce costs by orders of not just, and it's supposed to |
|
| 06:22 | in the next few years and increased . If you can reduce the power |
|
| 06:29 | to work impossible to put it in car now cool. It wouldn't be |
|
| 06:33 | much so on in order to power kind of thing. At best, |
|
| 06:40 | want to agree. It takes no amount of things yester power one of |
|
| 06:45 | guys. So megawatts is kind of these today and that's more or less |
|
| 06:55 | hopes. It's a small town to us apart. It's, I |
|
| 07:01 | that all the competition technology rebounded. our consume. Oh yes, |
|
| 07:11 | no. It is for our for computing on communication is today about |
|
| 07:24 | of all consumption, and it grows than most. Well, number |
|
| 07:33 | Yes, it is. It's already design, and that's part of the |
|
| 07:36 | . I was right, you So this is just saying what it |
|
| 07:42 | take. And that's why you should to see our exchanges in, among |
|
| 07:48 | things, and computer architecture will come . I'll show you what that |
|
| 07:53 | You're smart, quantum, only about one that will happen. But you |
|
| 08:00 | different things about so anyone so big , and this is a little bit |
|
| 08:06 | get the calibration. What big day means industry talked to many academics. |
|
| 08:16 | talk to him, they find a big day advice or something, and |
|
| 08:22 | that is not much so. So I'll give you very quickly. And |
|
| 08:29 | not, yes, to give you . Little bit scope. What industry |
|
| 08:34 | . Yeah, this is just a of data in terms of what's you |
|
| 08:41 | information. There's a little bit, , some of the commercial of Wal |
|
| 08:48 | . That's a few years back, many are off. The companies that |
|
| 08:53 | petabytes or contents are hundreds of not talk about the unit. He's 1 |
|
| 09:02 | is a retainer. As you that's to keep track of customer |
|
| 09:07 | This I love this backboard. Do like more? But this is what |
|
| 09:14 | have. Extend the action of squeezing out of it. Um, so |
|
| 09:23 | here's a few other things. Someone heard about, Uh, what's |
|
| 09:31 | uh, is trying to best known issues and health care this use it |
|
| 09:40 | to mine in some ways, or or no cut jobs, medical |
|
| 09:47 | in order that no sis actions. so the Knicks successes. But sometimes |
|
| 09:57 | has been quite successful. 20 years . So in terms of them, |
|
| 10:03 | records compared to human beings trying Things are start trying to extract information |
|
| 10:13 | in terms of what it's worth. of what it would work. |
|
| 10:20 | Yes, improve the efficiency in terms health care. It's serious money if |
|
| 10:25 | can make it that in these things , what's the little money like from |
|
| 10:33 | ? Yeah, it's travel. Here are. Yeah, well, it's |
|
| 10:36 | such a question off, you making better use of data that |
|
| 10:40 | um And it's not necessary. Here's one. So mammography generate sources, |
|
| 10:50 | of data. And here. So is, you know, kind of |
|
| 10:57 | city energy industry it Mrs Exploration, , someone you may end up there |
|
| 11:09 | to with them. So it's again of data, Davey or war. |
|
| 11:16 | , And they also have you no tens or even understood petabytes of |
|
| 11:23 | I know that again. Try to to make find out. Party should |
|
| 11:28 | drilling in. No, that's because depending on where you're trying to |
|
| 11:35 | , if you miss so Andy Ocean $3,000,000,000. So if you miss it |
|
| 11:47 | , fine. Right. So the will get this information and then so |
|
| 11:54 | the lies example of how you reduce . Yes. Yes. So |
|
| 12:07 | The right finger? Oh, I this especially Earth on this. I |
|
| 12:17 | so years. And then what? , or markets Or let you need |
|
| 12:35 | find old. It's best No, does. Where you Texas work Porous |
|
| 12:53 | that contains awarded Get the where it's giving In summer intern years we're gonna |
|
| 13:15 | destruction This is just so the way works Acoustic waves I closed down |
|
| 13:33 | You in their knowledge. So on dope has almost been There's barns full |
|
| 13:53 | attempts I'm all this. I speak Um, basically Ivar the bullet tips |
|
| 14:01 | not a real joke, I think said all the data around. |
|
| 14:05 | you It has been. So here's one. I think it's also but |
|
| 14:12 | power I want to win our companies the manufacturer is one that reduces power |
|
| 14:23 | sentence. So what to do you service from astrologist? What? All |
|
| 14:32 | of sensors out there that Mrs Waiting locations, they get historical records on |
|
| 14:40 | patterns and then they used to Martha of in Vince. That's what I |
|
| 14:49 | . And that's one aspect of the 2/4 in from of the competitions. |
|
| 14:56 | , of course, he also designed wind turbines patient. And then I |
|
| 15:02 | to do that in the this Then the also monitoring, controlled, |
|
| 15:09 | use. Leave the blades in such you control the generation or when |
|
| 15:15 | Really? Um, here's another You the monitor. So one thing |
|
| 15:27 | about your faction that, uh, of the they produce 6000 fan blades |
|
| 15:36 | your enough okay, properties of that family. That's a fact. Then |
|
| 15:48 | do they all say? Instrument on monitor. We have to what they're |
|
| 15:58 | at someone today. So giving the or what they get then they |
|
| 16:04 | What if it's time for us straight little. That's one thing. But |
|
| 16:09 | something is about to go bad, hope for the catch it before it |
|
| 16:12 | goes back. And not only then what to do Their best sent |
|
| 16:17 | tomorrow with the plane. His d'oh It's finer. Things that can repair |
|
| 16:24 | the plane is at some airport was to having to take it out of |
|
| 16:30 | , so it's pretty impressive. When you He was just another example. |
|
| 16:37 | Electric is another one that built turbines trains lactating and other things. So |
|
| 16:45 | do these things. Airbus's is one the company's in. That's it. |
|
| 16:52 | , in a way. But it's motor Dane. I think the next |
|
| 16:55 | shows a little bit. So they been collected, Um, off things |
|
| 17:02 | operational. Down rises, tear These are hold the line here, |
|
| 17:20 | version of the temple. So again instrument More, more court. Just |
|
| 17:27 | degree of time. Trust me to . Make it as an operation |
|
| 17:33 | Check the state and some of you minimize the potential out of service |
|
| 17:42 | Yes, yes, I guess that's nine years. So generous. |
|
| 17:54 | that's 15. Is that a safety just to give you a point old |
|
| 18:01 | , things that nobody thinks that's any rail stuff, but even they try |
|
| 18:06 | monitor what happens. Trying something, , trains and don't move fast. |
|
| 18:14 | they're trying to make sure that there's breakdown on the train. So one |
|
| 18:18 | the things is wear and tear and they try to monitor times. |
|
| 18:23 | train doesn't break down on the track clogs. So winners on this is |
|
| 18:30 | about when you see a lot about of the wheels somewhere. Yes, |
|
| 18:39 | is no developed because of him. something frequency, things that others |
|
| 18:47 | This should be consistent. Yeah, here's some more. Crew scheduling is |
|
| 18:54 | one thing that is airlines do and it also his. One thing |
|
| 19:00 | to do it for the week or . Storm seats whatever and planes but |
|
| 19:06 | and then have to reschedule things on notice both ways. Cruise planes and |
|
| 19:14 | . Other things is an engineering example . Degraded simulations design both the woods |
|
| 19:27 | in this case feet monk dying. this is from a cup you work |
|
| 19:36 | here, and this is a an SUV next to it, and |
|
| 19:44 | other things also so And you don't these, you know, every week |
|
| 19:50 | every month. Even so, it's of most of them are one of |
|
| 19:53 | kind, so it has to be once. Let's go back in, |
|
| 19:58 | okay, we'll fix it. So sort of a computing that needs to |
|
| 20:04 | done including thinking hot to transport these because it doesn't justice. Yes, |
|
| 20:13 | good words. And the mechanics and well attic apartments you don't want are |
|
| 20:21 | . The most contaminated particles come um, reactor designers, something you |
|
| 20:33 | as a big discussion eunuch for houses . On the other hand, it's |
|
| 20:38 | for him, waste product. It's , I think, in order to |
|
| 20:44 | things clean, there's it's protection that would need to rescind component energy. |
|
| 20:54 | to do good simulations of that, also highly commanding airplanes. We talked |
|
| 21:01 | coming back to the drinks here in of the engineering saw I speak |
|
| 21:07 | Yeah, shot very intensive minutes Samples are you think? Well on |
|
| 21:13 | train. That's real. Is it boring thing, it turns out not |
|
| 21:18 | be so. The curvature, it's . They're flats with kind of thing |
|
| 21:24 | protect this the off ramp. But curvature of the surface on right, |
|
| 21:31 | important other things is that happens because you joins us if you've been on |
|
| 21:40 | , you know, you know when current so you know otherwise, things |
|
| 21:47 | , seats, curves. So that's of the things to do the dynamics |
|
| 21:52 | control system of high speed and other is it pushes separate the good air |
|
| 21:59 | around it. So when two of French meat that you know before under |
|
| 22:05 | an hour, you don't want so a whole lot of issues, |
|
| 22:11 | do serious simulations. The face that's that man may not think |
|
| 22:20 | And here's one Procter and Gamble's. may. No anyone. Proctor. |
|
| 22:29 | do. All kinds of consumer goes court, so they use supercomputers and |
|
| 22:38 | up. But I do it for . We're going to shape to pack |
|
| 22:50 | , at very high speed. So takes a very particular shape. I'm |
|
| 22:54 | pregnant. It's not because it's supposed look good or anything that it is |
|
| 22:57 | shape of this. It's for them me. No, I mean, |
|
| 23:02 | started with shape there with a Yeah, I don't know if it's |
|
| 23:10 | from first. Principles are trying to read it other things that you |
|
| 23:15 | they do. Diaper design and other . Abscess surprising that eggs. |
|
| 23:25 | So that's but it's again so and things that you know a few little |
|
| 23:32 | . All kinds of things to give examples here. So So I |
|
| 23:39 | So there's something that is in the to see space people tend to used |
|
| 23:46 | bragging rights. You know what notes some this Oh, most powerful computers |
|
| 24:00 | kind of are more or less Not necessarily because parts of the government |
|
| 24:06 | doesn't want to tell you so is known. And this I didn't do |
|
| 24:13 | just because most of you probably end in the industry research up from this |
|
| 24:21 | . Computer savvy used. Yeah, settings go through this list there will |
|
| 24:31 | out that predominant it today it is weather forecasting. So you will see |
|
| 24:38 | bunch of netapp that offices are on go has these computers. Then there |
|
| 24:47 | petroleum or oil companies that tends to also some of the biggest systems. |
|
| 24:51 | then in a minute, a couple on this thing other parks take |
|
| 24:59 | So that's in terms of who gets types of computers. The other part |
|
| 25:04 | is important in terms of all this say terms of probably so offer manage |
|
| 25:12 | and this how to manage things at , both so here you see the |
|
| 25:20 | number? Of course, honey. of 1,000,000. Course. Want an |
|
| 25:32 | ? Do this off course. Then may, uh, execution St. |
|
| 25:48 | . Getting things to work at scale an important aspect. Oh, thank |
|
| 25:53 | , Esposito. Just use if you so oh. Oh, and |
|
| 26:20 | So, uh, these two No. Oh, these numbers are |
|
| 26:28 | by running your particular cold. I call it an application, but it's |
|
| 26:35 | I'm the ghost in the nation. , that used to be the only |
|
| 26:41 | that was used to right? So . And it's That's you. What |
|
| 26:52 | ? Okay, performance. Uh, what woman Over one. Doing it |
|
| 27:09 | Monday for direction. Execution. That bunch One instance for something versus |
|
| 27:40 | No, bring it. Course. many of these. I think our |
|
| 28:03 | what they're being used for. I'm usually more power. I know |
|
| 28:09 | most of the time it is. huh. It's all our efficiency |
|
| 28:23 | Where are you? Irritation. Far courses, forms. You are |
|
| 28:39 | Very important force don't work. Is of this measurement here in terms those |
|
| 28:53 | , rations? Operations? What? . Oh, you. But it's |
|
| 29:15 | big difference if you do it. the whole system on this Come over |
|
| 29:28 | heart, Get the names, Before just so this ends. Nation |
|
| 30:07 | , radiant methods. CS students, students. Yes, it's very simple |
|
| 30:20 | a little bit trickier than matrix pontification questions solvers appears required. Jackson Impression |
|
| 30:30 | are in science, and interviewing are court but Erica or dense major |
|
| 30:39 | So that's my scotch and definition of for a long time. You did |
|
| 30:58 | . It's over. Well, do try Thio, get some conclusions. |
|
| 31:04 | must have something You saw most of case that it happens, you |
|
| 31:09 | in Iranian darning on neural networks. somewhere in there there is a simple |
|
| 31:13 | operation. So So that's no No, sir, it is, |
|
| 31:26 | wrote, Get more insights. That's goes on in. This is a |
|
| 31:31 | bit of sense for the scale of that even industry does. Of |
|
| 31:36 | there's just not necessarily small companies. here. It's just more of the |
|
| 31:41 | and I can see drop a little down to this. All right, |
|
| 31:48 | soul young the first time it's just it up off the Internet companies showing |
|
| 31:59 | . They're not all this report to consumption of their systems. And then |
|
| 32:05 | can also see him on your Work through put systems on the hours |
|
| 32:13 | if your problems. This is from supposed to put systems. That's why |
|
| 32:21 | The same caliber off interconnect between the predicted a 15 or 20 years our |
|
| 32:28 | will be really important. It is asked me. Yeah, it's none |
|
| 32:34 | the overall conserving power. Well, started out with you at the ship's |
|
| 32:45 | , right? Yes. What s that's what to start waas So in |
|
| 32:52 | . About 15 orders from 15 years . They had to, uh What's |
|
| 33:00 | ? Yes. So are you doing ? Nothing was One force turned |
|
| 33:06 | couldn't cool, so it was about be released, and they have to |
|
| 33:10 | back to They release product. They're it burns too much. So |
|
| 33:20 | So yeah, they're over. Show slides later on. The best gives |
|
| 33:27 | is more of this. Our vision do what? Slower, Increasing. |
|
| 33:39 | , so that that is you cannot enough. That's it. So intensity |
|
| 33:50 | So heat density chips about 20 years . Well, it's kind of |
|
| 33:55 | Heat density in the nuclear reactor, ? No. So here's other |
|
| 34:04 | You do a medical emergency in terms computing. They want to organ |
|
| 34:10 | Nothing's requires. A lot of computing one of my favorites. Example |
|
| 34:13 | in terms of you need to do terms of and this candle and I |
|
| 34:20 | my cross put made. So here's you do Get my crystal mike or |
|
| 34:30 | out of this great nation function. the car's a free muffin computing. |
|
| 34:39 | you know, if you are electrical , this thing terrible. C'mon, |
|
| 34:48 | . And that's that's the first That's a good idea about imposing all |
|
| 34:52 | noise. You need to check for more responsible, what they're looking |
|
| 34:57 | Unfortunately, this is molecules. If do your burner, man, you |
|
| 35:02 | destroy this specimen. So the thing really to try to extract information? |
|
| 35:08 | . See pictures or images. More , clever, outward, obsessive. |
|
| 35:18 | more energy stuff. Couple. access of these today they do about |
|
| 35:34 | of about for Armstrong two, and also must measure yeah, 10 to |
|
| 35:42 | minus 10. 10 to the minus I'm still winning. Yes. |
|
| 35:48 | it's Mommy. Four 12. Yes, I know what I'm |
|
| 35:56 | Get this. The macro molecules are a bunch of times. I'm trying |
|
| 36:02 | do my word for it, It's always someone I used to see |
|
| 36:11 | , and it's basically and what last . And so there's probably this image |
|
| 36:27 | of the image it on you and ? Oh, and those are the |
|
| 36:34 | you used to process. Yeah, oh puts the same mama que but |
|
| 36:45 | government angles. You know, they align themselves nicely for you, so |
|
| 36:50 | don't know which angle their acts you about such island errors. Tonto |
|
| 37:01 | So and all right. And then other things People do simulations breathing. |
|
| 37:06 | here's a little bit of scale coming . You mentioned time scales, scales |
|
| 37:14 | range in order to do space. memo, weather prediction. The doctor |
|
| 37:22 | hurricanes we know about have a little . So tornado predictions is one of |
|
| 37:29 | things also that needs to be done accurately. Unlike hurricanes, you shall |
|
| 37:37 | a fair amount of time to So what? She worried not |
|
| 37:44 | Harvey, These hurricanes, you you may not decide to evacuate the |
|
| 37:50 | about something. A few 1,000,000 people the road from a small area. |
|
| 37:55 | is not true of exercise. And can't do that for tornado that is |
|
| 38:00 | a very small foot hurricane as big . Do it. But this was |
|
| 38:09 | nature out that has you read any hungry solution, addiction. These |
|
| 38:15 | they get pretty good figuring that tomato climate smuggling. You know, environmental |
|
| 38:23 | is picture Houston, that's not so . Yes, yes, but what |
|
| 38:34 | if you're lucky, like, is the same soul Part of the things |
|
| 38:41 | I work with a few years I work with some people in the |
|
| 38:45 | . You're excited department. So they up another sensors here in the Houston |
|
| 38:49 | , and you try to make Oh, seriously, open weather. |
|
| 38:54 | you can, you know, tell district exercise and breathe too much trouble |
|
| 39:01 | . So it's just a way of taking more of your time management |
|
| 39:06 | Given what it is warm, people couple exercise and jogging on. It's |
|
| 39:12 | a because you don't see in a are gases. I need to know |
|
| 39:21 | another thing Is that records lots of with fire simulations. What fire? |
|
| 39:30 | Woz, huh? Or a few ago. And one of the college |
|
| 39:45 | . You try to come in some that you do simulation fires in the |
|
| 39:51 | . It was no traction. Nobody . So what happened? And may |
|
| 40:08 | place fly at 10,000 meters out to out this morning. 10,000 meters. |
|
| 40:18 | , things in here. So astronomer do degenerate tons of data. The |
|
| 40:29 | and there are many things. So why don't you Maybe from |
|
| 40:36 | I know what's students You. so particle physics is one. I |
|
| 40:43 | know what visits you do, did ? You doing the class project? |
|
| 40:52 | know what? I was just your wrenches always right. What |
|
| 41:13 | Big, Except our intern that, ? Um, no, it's partially |
|
| 41:20 | . I'm partially friend. So I know if you're going to start. |
|
| 41:28 | . Oh, and I said that petabytes of data each year. So |
|
| 41:34 | is big again. And then to all the data that comes, Do |
|
| 41:40 | ? Which physics department is most The part science sent us. Are |
|
| 41:51 | to order groups just working on Uh, so, yes, I |
|
| 42:03 | I put miss in terms of thanks do things in real time. |
|
| 42:11 | it's actually beautiful. Well, I been attacking. I'm so reliable power |
|
| 42:24 | it turned out you did this or . Very congressman, uh, |
|
| 42:32 | Their next work's work. So back 1965 big blackout with, you |
|
| 42:43 | city and 100 Lost Power CD. wasn't very popular. So what I |
|
| 42:53 | to start to figure out, get much better assessment how vulnerable their systems |
|
| 43:03 | . Far for that money to do lot of competition. So in that |
|
| 43:07 | , parts matrix equations. One of things I did most not to do |
|
| 43:14 | efficiently. Duckman I work for managed love less money than the competition. |
|
| 43:23 | one contract based on then it became the product of the company two |
|
| 43:40 | Another thing worked out for them on , um, thinking machines, |
|
| 43:47 | the next machine. What? They first system the first list first, |
|
| 44:04 | powerful system on that this year. system, this company, the benchmark |
|
| 44:14 | group was responsible for us in terms SBC two examples where you see in |
|
| 44:22 | case that one what? Your performance got a good business on created a |
|
| 44:39 | And this is the limit. This Oh, mark zone. So it |
|
| 44:45 | you some size. So this a of more words before the what is |
|
| 44:54 | Much examples Science formance high performance were it's, uh not, you |
|
| 45:04 | scientific and engineering competition where many people this, why you of it? |
|
| 45:11 | prowess. But it is a question . Essentially using profession resource, you |
|
| 45:20 | . You need not case now a bit about algorithms architecture catching off |
|
| 45:29 | Then you need to have programming tools matching you want to achieve. So |
|
| 45:39 | and the other part is, um , so this is a high |
|
| 45:43 | It can't get the efficiency on this . No, was a lot of |
|
| 45:52 | . So I was speaking about that has a clue. Typical efficiency that |
|
| 45:58 | as a fraction off p forwards Peter performance and she on typical cold that |
|
| 46:07 | on these things. Fraction of this for so have fun. Yes. |
|
| 46:22 | , you know Whoa, wait. are way too high. Okay, |
|
| 46:27 | was hedging. Wait. Yeah, know. Typical cold that is still |
|
| 46:39 | of perceived to be reasonable. Written . 10 about no one of science |
|
| 46:56 | that works on these big machines. got in the water for getting 30% |
|
| 47:05 | . So no. So, you , I don't care too much. |
|
| 47:09 | laptop efficient. It is by having go back to my earliest signed for |
|
| 47:14 | 102 $100,000,000 computer. If you just one minute's worth of it. Your |
|
| 47:22 | , Where's the but conditions in terms power, you know? So here's |
|
| 47:34 | part. How much of it? I will be using resources. Part |
|
| 47:42 | the things. Course the tools to some insights in tow where bottlenecks use |
|
| 47:59 | in principle. Yes, in The best thing that always like |
|
| 48:04 | Percent on that design. A specific . No, Jake. The standard |
|
| 48:14 | . See you on. Did she matrix? Multiplication? Honey, I |
|
| 48:24 | , but you can get it on . You can cast things in things |
|
| 48:28 | are similar than you. Yet you get it. First part's matrix |
|
| 48:36 | There's no talk. More wine, . So this is another bet in |
|
| 48:44 | . Off are the particle axis on , Aurens. It's Gideon. Some |
|
| 48:58 | , Instructions per second. So name . Yeah. So no and |
|
| 49:39 | You want power and you want high . No, I don't want to |
|
| 49:45 | in this corner. So I was of a joke. If you want |
|
| 49:49 | have something where? Up into the , That's a good thing to be |
|
| 49:52 | up. Means are kind of, , here on its butt here. |
|
| 50:12 | in terms. All the parts Assumption you. Well, fire capability tend |
|
| 50:24 | the fire in terms of if it's efficiency. So what's the difference |
|
| 50:31 | Is the difference Hers, don't Are you decide? What is |
|
| 50:40 | Your Yeah, Lord, life in . Here. You know, funding |
|
| 50:55 | or Yeah, it's just you can't apart this and on guard a book |
|
| 51:14 | armies arm here, process. Do right. So there's more than our |
|
| 51:45 | produce insignificant. Think us in. being the dominating I'm so course being |
|
| 51:58 | . Wishes are These are for this . So And here is another sign |
|
| 52:13 | no coming back for this notion of and how much you get four and |
|
| 52:28 | in their sample is using patients, ? Right, Right. No. |
|
| 52:44 | , I know that. Right. you can see here Execution. 9,000,000 |
|
| 52:55 | a second farces. This is one . So software is important. |
|
| 53:17 | the difference to be in these two . 5000 pounds. So again. |
|
| 53:27 | that's partially thing, you know, to emphasize in divorce. What? |
|
| 53:38 | , for months, Languages. People sight. Yeah, this side is |
|
| 53:51 | from courses. So there's something more less that So that's why I started |
|
| 54:02 | say missions that required efficient state for , what you need or what's beginning |
|
| 54:12 | happen. There's a proliferation in so and you get enough on a |
|
| 54:17 | of silicon to execute bunch of the , it's again something. It's important |
|
| 54:29 | space to be aware that that happens though architecture in this course. But |
|
| 54:37 | understand that things are happening. Be . You okay? So this process |
|
| 54:46 | have my execution course. No, don't even use all its course. |
|
| 54:55 | . Well, it depends again the and you're used to No. |
|
| 55:01 | No. When it does. Isis, that is made you smoked |
|
| 55:05 | example in 98. 99% of vicious got even on. But of |
|
| 55:17 | you know, if the application don't the parent was used his horses. |
|
| 55:27 | here's a guess. A little bit this. So that's again Dolly. |
|
| 55:38 | didn't know you have video, but been very active in this space for |
|
| 55:44 | long time student that when I was And by the way, there he |
|
| 55:52 | , giving a talk tomorrow. Rights a cop? Yes. So in |
|
| 56:00 | ability, I encourage you to go . Listen to me. I don't |
|
| 56:04 | Exacto title of his talk. It's different guy and get very Justin Fox |
|
| 56:15 | to rise. It's in Duncan you know, c'est apart. |
|
| 56:21 | you know the winterized campus. No. So you can find it |
|
| 56:30 | on the world. But if you Rice campers, you know the fancy |
|
| 56:36 | off main interest with them with our , then the courtyard behind it. |
|
| 56:42 | if you look at that bending meters the right of that is done, |
|
| 56:52 | see what else hears things again. , go. The Internet companies there |
|
| 57:02 | that's again by cool and I don't if I have it here. But |
|
| 57:09 | designed started to design their own And so that's Facebook. Standard |
|
| 57:18 | terms of our consumption and then, the other hand, large enough. |
|
| 57:23 | they decided. Here's but the service be. And then they tell his |
|
| 57:26 | have been going with this one up That's what's happening, and I'll talk |
|
| 57:33 | about that. Distribute example. So a sense, it's another infrastructure and |
|
| 57:39 | . It's not bunch of racks on , other things that people are also |
|
| 57:44 | about. This again, part of . Staying green about one and 1/2 |
|
| 57:52 | the costs. Lifetime cost. The is probably and a few insults. |
|
| 57:59 | I would talk about that. My is up. There's a big slice |
|
| 58:03 | monster Shane showbiz this. It's where with a bar to chance therefore also |
|
| 58:11 | heart of this gypsy. So because got to up, so here's more |
|
| 58:19 | less similar to the other. With , you asked about different level, |
|
| 58:25 | deal efficiency off. I just talked Okay, just roll this nonsense |
|
| 58:40 | So I hope you stay in the . Two, Of course, |
|
| 58:49 | it's important you know is Of if you don't think it's for |
|
| 58:52 | it's It is okay. Thank Do they You work for that? |
|
| 59:07 | |
|
