WEBVTT - TechStuff Classic: TechStuff Looks at Supercomputers 0:00:04.240 --> 0:00:07.240 Welcome to tex Stuff, a production of I Heart Radios, 0:00:07.320 --> 0:00:13.720 How Stuff Works. Hey there, and welcome to tech Stuff. 0:00:13.720 --> 0:00:17.000 I'm your host, Jonathan Strickland. I'm an executive producer with 0:00:17.079 --> 0:00:19.360 How Stuff Works and I Heart Radio and I love 0:00:19.480 --> 0:00:24.000 all things tech, and today we're going to look at 0:00:24.040 --> 0:00:28.800 a classic episode that aired back in July of two 0:00:28.800 --> 0:00:33.000 thousand twelve, July six to be precise, and it is 0:00:33.000 --> 0:00:37.839 titled tech Stuff looks at super Computers. This was a 0:00:37.920 --> 0:00:41.200 fun discussion to have with my former co host Chris 0:00:41.240 --> 0:00:45.360 Pallette because I, you know, grew up in an era 0:00:45.440 --> 0:00:47.920 of super computers and only had sort of a vague 0:00:47.960 --> 0:00:51.880 idea of what that term meant for many years. I'm 0:00:51.880 --> 0:00:53.320 sure there was a time when I was a kid 0:00:53.400 --> 0:00:55.120 where I thought of it as a computer that wore 0:00:55.120 --> 0:00:57.240 a cape. But as it turns out, it gets a 0:00:57.280 --> 0:01:00.960 little more complicated than that. Let's listen in So, Chris, 0:01:01.080 --> 0:01:02.920 if I were to ask you, just off the top 0:01:02.960 --> 0:01:06.080 of your head, how would you define a supercomputer? What 0:01:06.120 --> 0:01:09.240 would you say? Well, if I hadn't already made the joke, 0:01:09.280 --> 0:01:10.959 I would have said it was a computer in the 0:01:11.000 --> 0:01:15.360 cape and tights. But no, honestly, I would say supercomputer. 0:01:15.560 --> 0:01:20.120 Is a computer that can do a lot more calculations 0:01:20.160 --> 0:01:23.000 in a shorter period of time than the machines sitting 0:01:23.040 --> 0:01:24.959 on our desktop. Yeah. I think of it as sort 0:01:24.959 --> 0:01:28.320 of the bleeding edge of what a computer is capable 0:01:28.360 --> 0:01:32.040 of doing. Something that that still feel fills a room, 0:01:32.080 --> 0:01:35.280 even though typical computers these days don't need to fill 0:01:35.280 --> 0:01:37.640 a room because it's that big. It still has that 0:01:37.760 --> 0:01:42.280 much computing power, right, right, And the term comes from 0:01:42.560 --> 0:01:46.240 the nineteen sixties and uh. In order to really kind 0:01:46.240 --> 0:01:50.680 of understand the the the span of this, I think 0:01:51.000 --> 0:01:52.960 I was going to talk a little bit about the 0:01:53.520 --> 0:01:56.600 last computer I could find that was a powerful computer 0:01:57.080 --> 0:02:02.120 that existed before people started talking about super computers, which 0:02:02.200 --> 0:02:07.360 was the IBM seventy thirty Stretch. Yes, that was the 0:02:07.360 --> 0:02:13.280 one that was made with elastic. Yes, ye gain a 0:02:13.280 --> 0:02:15.640 couple of pounds. Your computer can still you know fit. 0:02:16.040 --> 0:02:19.320 It was Mr Fantastic of the computer world. No, because 0:02:19.360 --> 0:02:22.000 it was not a super computer. It took up two 0:02:22.080 --> 0:02:25.480 thousand square feet back in the day, this being the 0:02:26.120 --> 0:02:32.200 early sixties. Two thousand square feet. It costs thirteen million dollars, which, 0:02:32.240 --> 0:02:36.680 if you were to translate to today's cash, would be 0:02:36.800 --> 0:02:40.120 ninety one million dollars. It's a lot of money. It's 0:02:40.120 --> 0:02:43.280 a lot of cash. So that was the fastest computer 0:02:43.360 --> 0:02:46.320 at the time until a fellow named Seymour Roger Craig 0:02:46.440 --> 0:02:50.360 showed up. I s Mr. Craig. Yeah, Craig ends up 0:02:50.400 --> 0:02:54.160 being a big name in supercomputers, particularly in the sixties, 0:02:54.200 --> 0:02:58.000 seventies and up to the mid eighties. That was the 0:02:58.200 --> 0:03:03.160 name in supercomputers. And he was working for a company 0:03:03.200 --> 0:03:07.360 called Engineering Research Associates or e r A, which actually 0:03:07.400 --> 0:03:12.080 grew out of a naval operation, um being the U. S. Navy, 0:03:12.120 --> 0:03:15.000 not belly Button. I was gonna you were looking at 0:03:15.040 --> 0:03:18.600 me like joke. No, No, not that it was a 0:03:18.720 --> 0:03:21.840 navy project. How about that as in as in the 0:03:21.880 --> 0:03:25.360 military force, not the color It was a Navy project 0:03:25.600 --> 0:03:28.480 that was all about code breaking, all right. So there 0:03:28.520 --> 0:03:31.280 was this project about code breaking that eventually kind of 0:03:31.320 --> 0:03:34.359 spun off and became an actual company all on its 0:03:34.360 --> 0:03:37.920 own called Engineering Research Associates, and it branched out beyond 0:03:38.000 --> 0:03:40.520 code breaking, although it took all the code breaking work 0:03:40.560 --> 0:03:44.760 it could get. Yeah, we talked about the Enigma UM 0:03:44.880 --> 0:03:48.760 some episodes back UM and we were talking about the 0:03:48.880 --> 0:03:53.320 bomb UM and yeah, there was early uh that was 0:03:53.360 --> 0:03:58.080 really the early application for supercomputers was you know, needing 0:03:58.120 --> 0:04:01.720 to crunch a lot of data very quickly, and there weren't. 0:04:02.680 --> 0:04:04.640 There weren't the kind of applications that we have now. 0:04:04.680 --> 0:04:06.560 We'll get into that, I'm sure in a few minutes. 0:04:06.600 --> 0:04:09.880 But but yeah, I mean that was you know, why 0:04:09.920 --> 0:04:13.120 would you need a supercomputer? You know, that was That's 0:04:13.120 --> 0:04:15.920 probably about the only thing I could think of where 0:04:15.920 --> 0:04:18.840 people were needing to crunch that kind of information as 0:04:18.920 --> 0:04:23.719 quickly as possible. And defense. Yeah, typically, especially with the 0:04:23.760 --> 0:04:28.560 early supercomputers, they were really designed for very specialized computing, 0:04:29.360 --> 0:04:32.960 so not necessarily specialized from the ground up for a 0:04:33.120 --> 0:04:35.840 one particular type of computing, but they were They were 0:04:35.880 --> 0:04:38.400 not meant to be general computers. They were meant to 0:04:38.440 --> 0:04:43.039 do tip no admiral computers because they were the Navy, 0:04:43.120 --> 0:04:45.640 that's true. Uh No, they were. They were meant to 0:04:45.680 --> 0:04:48.560 do a specific task very very well, and that's that's 0:04:48.600 --> 0:04:53.080 all they were meant to do. Now, Craig had an 0:04:53.120 --> 0:04:55.800 interesting philosophy he said, and this is this is a 0:04:55.880 --> 0:04:59.480 quote from him. He said, anyone can build a fast CPU. 0:04:59.760 --> 0:05:02.840 That trick is to build a fast system and that 0:05:02.920 --> 0:05:07.040 was the secret to create creating the first supercomputer. He 0:05:07.120 --> 0:05:10.640 realized that if you created a processor that was really, 0:05:10.640 --> 0:05:14.120 really fast, that did not matter if it couldn't get 0:05:14.160 --> 0:05:18.560 the data it needed to execute operations upon fast enough. 0:05:19.279 --> 0:05:22.680 So he saw the need to create a system that 0:05:22.720 --> 0:05:26.960 would move data through very very quickly, not just processed data, 0:05:27.040 --> 0:05:28.720 but move it so that means it needs a lot 0:05:28.720 --> 0:05:31.880 of memory, it needs a very fast pathway from memory 0:05:31.880 --> 0:05:34.479 to processor. There are a lot of pieces that have 0:05:34.520 --> 0:05:36.760 to be put in place, and he saw this very 0:05:36.800 --> 0:05:40.800 early on, and so using that philosophy, he designed a 0:05:40.800 --> 0:05:44.440 computer back in nineteen two that was called the c 0:05:44.760 --> 0:05:49.520 d C six six hundred. Now CDC stands for Controlled 0:05:49.600 --> 0:05:54.120 Data Corporation. Yeah, um h E R A was taken 0:05:54.120 --> 0:05:58.400 over by Remington Randy UM. And that's, uh, that's the 0:05:58.480 --> 0:06:01.640 name I remember because you know, UH still remember a 0:06:01.640 --> 0:06:05.680 lot of those old machine names UM from stuff that 0:06:05.800 --> 0:06:09.000 I found in my uh dad's collection. Of course, he was, 0:06:09.040 --> 0:06:12.520 you know, a mechanical engineer UM before he retired, and 0:06:13.160 --> 0:06:15.400 you know, so he was interested in all kinds of machines. 0:06:15.480 --> 0:06:17.400 And I didn't know what I was looking at at 0:06:17.400 --> 0:06:19.400 the time, of course, you know, but they were all 0:06:19.440 --> 0:06:22.960 these UM science and computing magazines laying around and that 0:06:23.040 --> 0:06:27.120 name I recognized also UNSIS because Remington Rand became Unities 0:06:27.680 --> 0:06:30.040 UM and probably a lot more of our listeners are 0:06:30.040 --> 0:06:34.120 familiar with that name. But he partnered with William Norris 0:06:34.320 --> 0:06:38.360 to start Controlled Data Corporation UM back in nine seven 0:06:39.120 --> 0:06:42.120 UM and really at that point, the UNIVAC from Remington 0:06:42.240 --> 0:06:46.760 Rand and IBM were the computing companies. And you know, 0:06:46.800 --> 0:06:50.239 IBM has been the heavyweight for so long, but CDC 0:06:50.560 --> 0:06:54.600 was the first, uh you know, upstart to really make 0:06:54.640 --> 0:06:58.720 a dent in there, uh stranglehold on the industry. And 0:06:59.000 --> 0:07:02.880 Craig wanted to join CDC fairly early on, but apparently 0:07:03.279 --> 0:07:07.919 he was needed for a project UM that would not 0:07:08.120 --> 0:07:10.560 let him leave exactly what he wanted to. So once 0:07:10.560 --> 0:07:13.840 he did leave, that's when he designed the CDC sixty hundred, 0:07:14.920 --> 0:07:19.280 which was officially announced in nineteen sixty four, so designed 0:07:19.280 --> 0:07:21.840 in sixty two, announced two years later, and it was 0:07:21.880 --> 0:07:25.720 the first commercially successful supercomputer, with a price tag of 0:07:25.800 --> 0:07:28.520 between seven and eight million dollars, sometimes going up as 0:07:28.560 --> 0:07:31.240 high as ten million, depending upon the configuration that the 0:07:31.280 --> 0:07:34.840 customer wanted. UM. Now, in today's cash, that would equal 0:07:34.880 --> 0:07:39.239 about sixty million dollars, so thirty one million dollars cheaper 0:07:39.240 --> 0:07:44.960 in today's money than the Stretch computer, and it was 0:07:45.000 --> 0:07:50.680 actually much more powerful. It had four hundred thousand transistors 0:07:50.720 --> 0:07:54.280 and one hundred miles of wiring, and it was the 0:07:54.360 --> 0:07:58.240 size of about four filing cabinets, so it's also significantly 0:07:58.320 --> 0:08:01.920 smaller than the Stretch. Didn't take up two thousand square feet. 0:08:02.800 --> 0:08:06.600 The clock speed was around a hundred nanoseconds, and it 0:08:06.680 --> 0:08:10.920 had sixty five thousand sixty bit words of memory. So 0:08:11.400 --> 0:08:13.680 this is kind of an odd time in computing. We 0:08:13.720 --> 0:08:18.800 hadn't really settled on the thirty two sixty four bit 0:08:18.960 --> 0:08:23.440 kind of model. This was before that. Um. It also 0:08:23.520 --> 0:08:26.080 used six high speed drums as sort of a temporary 0:08:26.080 --> 0:08:29.440 storage area. It had a central storage that used magnetic tape, 0:08:29.840 --> 0:08:33.760 and it used the four trans sixty six compiler. UM 0:08:33.800 --> 0:08:36.440 the equivalent to today's machines means that it would have 0:08:36.520 --> 0:08:42.280 about a ten mega hurts processor. Yeah, well that could 0:08:42.320 --> 0:08:44.520 work up to forty mega hurts and speed. Well, it 0:08:44.559 --> 0:08:47.880 could do a three million floating point operations per second. 0:08:47.960 --> 0:08:52.520 Yeah those areas, Yeah, so three million, that would be 0:08:52.520 --> 0:08:55.720 a mega flop, three mega flops, right, so we're gonna 0:08:55.720 --> 0:08:58.760 get into lots of different flop terms later as well, 0:08:58.840 --> 0:09:03.520 and they get incredibly huge. Of course, you have to 0:09:03.559 --> 0:09:05.439 keep it cool because otherwise it breaks out into a 0:09:05.480 --> 0:09:08.400 flop sweat. And that's true. Uh, well, not the flop 0:09:08.440 --> 0:09:10.000 sweat part, but you do have to keep it cool. 0:09:10.360 --> 0:09:13.280 As we know electronics, when you're running electricity through them, 0:09:13.280 --> 0:09:15.960 one of the byproducts is heat, and heat, as it 0:09:16.000 --> 0:09:19.959 turns out, is not a great thing for electronic components. 0:09:20.360 --> 0:09:25.199 It can make stuff expand contacts can lose connections, so 0:09:25.240 --> 0:09:29.360 that stuff starts to malfunction. An entire system could shut down. 0:09:29.480 --> 0:09:34.920 So the CDC had a cooling system that was provided 0:09:35.000 --> 0:09:39.880 by a special chemical free only. Yeah, they used free 0:09:39.920 --> 0:09:42.760 on to cool the system. In fact, it was they 0:09:42.760 --> 0:09:45.920 would use free On for a while before finally having 0:09:45.960 --> 0:09:49.400 to switch to a different coolant because free on just 0:09:49.559 --> 0:09:54.400 wasn't efficient enough. Eventually, now at it was still doing 0:09:54.440 --> 0:09:59.720 the job. So Cray was also an innovator in another way. 0:10:00.320 --> 0:10:05.880 The stretch IBM stretched um was sort of a hybrid machine. 0:10:06.400 --> 0:10:10.520 They had transistors and vacuum tubes in it um and 0:10:10.559 --> 0:10:13.480 that's I think why one of the reasons why craze 0:10:13.520 --> 0:10:17.640 machines were smaller. The six four, which proceeded the sixty 0:10:17.679 --> 0:10:21.800 six hundred UM, was the one of the very first 0:10:21.840 --> 0:10:29.080 to use transistors only, so there was also a transistor machine, 0:10:29.280 --> 0:10:30.960 and so it would take up a lot less space 0:10:30.960 --> 0:10:33.480 than the vacuum tubes. And I would imagine that based 0:10:33.480 --> 0:10:36.000 on my knowledge, my personal knowledge of vacuum tubes, might 0:10:36.000 --> 0:10:38.439 have been a little cooler simply because of that. Yeah, 0:10:38.440 --> 0:10:40.200 I would imagine that they wouldn't have had to have 0:10:40.520 --> 0:10:44.160 as dramatic an a C system to keep the the 0:10:44.240 --> 0:10:47.760 room bearable, because vacuum tubes do put off a lot 0:10:47.800 --> 0:10:52.720 of heat UM. Another interesting IBM C d C connection 0:10:52.800 --> 0:10:56.760 here is that Thomas Watson Jr. Which was IBM s C. 0:10:57.000 --> 0:11:00.600 He was IBM CEO at the time, wrote a famous 0:11:00.640 --> 0:11:06.160 memo that time too IBM employees, and he said, last 0:11:06.200 --> 0:11:10.840 week Controlled Data announced the six system. I understand that 0:11:10.880 --> 0:11:13.600 in the laboratory developing the system, there are only thirty 0:11:13.640 --> 0:11:17.760 four people, including the janitor. Of these fourteen our engineers 0:11:17.800 --> 0:11:21.199 and four our programmers. Contrasting this modest effort with our 0:11:21.320 --> 0:11:25.040 vast developmental activities, I failed to understand why we have 0:11:25.200 --> 0:11:28.040 lost our industry leadership position by letting someone else offer 0:11:28.080 --> 0:11:32.560 the world's most powerful computer. Craig's response was a reportedly, well, 0:11:32.600 --> 0:11:36.400 there's your problem. Essentially, Craig was saying that, you know, 0:11:36.640 --> 0:11:44.080 perhaps IBMS approach it was a little burdened by size. 0:11:44.320 --> 0:11:47.840 That IBM had grown so large that to manage a 0:11:47.880 --> 0:11:51.640 project like this was very difficult to do because it 0:11:51.679 --> 0:11:55.360 was just too big. So that's an interesting idea that 0:11:55.840 --> 0:11:59.920 an organization needed to be kind of small and nim 0:12:00.040 --> 0:12:02.199 bowl in order to pull something off like creating the 0:12:02.240 --> 0:12:05.839 world's fastest computer. He followed up the c d C 0:12:06.160 --> 0:12:10.760 six hundred with the seventy, which had a sixty five thousand, 0:12:10.840 --> 0:12:13.559 five hundred thirty six sixty bit word memory and a 0:12:13.600 --> 0:12:16.559 clock speed of twenty seven nano seconds uh and actually 0:12:16.800 --> 0:12:20.680 in practice ran about five times faster than the sixty. 0:12:22.000 --> 0:12:25.160 But then Cray left c d C and he formed 0:12:25.200 --> 0:12:30.480 his own company, Kray Research, and in nineteen seventy six 0:12:30.480 --> 0:12:33.560 he introduced the Kray one, which if you've ever heard 0:12:33.600 --> 0:12:37.000 the Kray supercomputer, that's what this is. It's the crazy 0:12:37.080 --> 0:12:39.840 One was the first of those. It had a clock 0:12:39.920 --> 0:12:43.320 speed of a well, its processor ran at eighty mega 0:12:43.360 --> 0:12:48.240 hurts and back. At this time these supercomputers were still 0:12:48.360 --> 0:12:52.400 using a single CPU, so that was kind of interesting 0:12:52.400 --> 0:12:55.240 to these were single CPU systems. So it had eighty 0:12:55.280 --> 0:12:59.120 mega Hurts processor sixty four bit system. It ran at 0:12:59.120 --> 0:13:01.559 a hundred thirty six mega flops, so one or three 0:13:01.640 --> 0:13:04.960 six million floating operations per second, and it had one 0:13:05.000 --> 0:13:09.080 thousand six d sixty two printed circuit boards that made 0:13:09.160 --> 0:13:13.040 up the components of this computer. It costs between five 0:13:13.040 --> 0:13:15.120 and eight million dollars, depending on how you wanted it 0:13:15.120 --> 0:13:17.880 set up, and in today's cash that's about twenty five 0:13:17.920 --> 0:13:21.760 million dollars. So we see that the processor speed is 0:13:21.800 --> 0:13:26.319 increasing and the price is coming down. Often the size 0:13:26.440 --> 0:13:29.520 of the computer is decreasing as well, although that that 0:13:29.679 --> 0:13:34.480 also flip flops over the years because while the solid 0:13:34.559 --> 0:13:40.319 state electronics definitely brought the size down, eventually the way 0:13:40.480 --> 0:13:45.400 we pack in more speed requires more space. But we'll 0:13:45.400 --> 0:13:50.280 get into that. Okay. So after the Cray one came 0:13:50.320 --> 0:13:55.520 the Cray x MP. Yeah. This is uh, this is 0:13:55.600 --> 0:14:00.240 interesting because realized also in addition to the fact that 0:14:00.360 --> 0:14:03.439 he knew that the components, the all of the components 0:14:03.440 --> 0:14:07.800 the entire machine was important and not just a processor, 0:14:07.840 --> 0:14:12.240 he also realized that uh, early on that parallel processing 0:14:12.480 --> 0:14:16.840 could also speed things up. UM. Now it's common for 0:14:16.920 --> 0:14:20.080 us to have multi core processes in our desktop machines 0:14:20.240 --> 0:14:23.440 or laptops, or in fact, now we're starting to see 0:14:23.440 --> 0:14:27.880 them in our mobile devices. UM. But you know, at 0:14:27.920 --> 0:14:30.080 the at the time in the seventies and eighties, this 0:14:30.240 --> 0:14:34.240 was still something sort of newish, UM, and it's not 0:14:34.320 --> 0:14:38.360 something that everybody realized. Uh. So the x MP actually 0:14:38.720 --> 0:14:44.760 was to Cray one computers linked together, UM, and using 0:14:44.760 --> 0:14:49.320 those two machines together in a multiprocessing effort, UM, they 0:14:49.360 --> 0:14:54.200 could triple the performance of just one Cray one UM, 0:14:54.240 --> 0:14:59.600 which is something interesting to note. And uh Cray two 0:14:59.720 --> 0:15:02.800 had four processors in the same machine and that was 0:15:02.840 --> 0:15:06.320 the first to exceed one billion flops as Britainic it 0:15:06.440 --> 0:15:09.280 tells me. Yeah, uh, it actually could have up to 0:15:09.680 --> 0:15:14.880 eight CPUs the create too. UM. The these machines often 0:15:14.960 --> 0:15:19.080 over time were upgraded, so the initial step specs you 0:15:19.080 --> 0:15:21.440 would get when they were first released were one thing, 0:15:21.520 --> 0:15:24.320 and then by the end of the run of production 0:15:24.520 --> 0:15:26.720 they would be better. I mean, which makes sense. I 0:15:26.720 --> 0:15:28.440 mean we see that in computers all the time. We 0:15:28.600 --> 0:15:31.560 definitely we tend to call them different model numbers now, 0:15:31.600 --> 0:15:35.200 but the same sort of thing happens. So back in two, 0:15:35.200 --> 0:15:38.760 you had this the Create XMP with a hundred five 0:15:38.840 --> 0:15:44.080 mega hurts CPUs running around two hundred megaflops each. Uh, 0:15:44.200 --> 0:15:46.680 then if they had up to four CPUs you could 0:15:46.680 --> 0:15:50.800 get eight hundred megaflops going, and that was pretty impressive. 0:15:50.840 --> 0:15:53.400 It had the equivalent, by the way, of a hundred 0:15:53.440 --> 0:15:57.040 and twenty eight megabytes of RAM, So yeah, you think 0:15:57.040 --> 0:15:59.440 about that one or twenty eight megabytes of RAM in 0:15:59.560 --> 0:16:03.040 nineteen a D two that was considered bleeding edge for 0:16:03.080 --> 0:16:07.920 a supercomputer. Um and the storage units for the Cray 0:16:08.160 --> 0:16:10.320 XMP were the size of a file cabinet and they 0:16:10.360 --> 0:16:13.360 could hold up to twelve gigs of storage. Because they 0:16:13.360 --> 0:16:16.200 have a flash drive in my bag with me that 0:16:16.240 --> 0:16:18.720 has eight gigs. Yeah, and you can find you can 0:16:18.760 --> 0:16:22.440 find you just find twenty gig or more flash drives, 0:16:22.480 --> 0:16:24.160 which you know, you think about that, that's something that 0:16:24.360 --> 0:16:27.440 is small enough for you to carry on a key chain. Well, 0:16:27.480 --> 0:16:30.880 back in two you had a file cabinet sized device 0:16:30.960 --> 0:16:35.040 that could hold twelve gigs and that was considered massive, 0:16:35.160 --> 0:16:39.400 like a massive amount of information. So yeah, time really 0:16:39.440 --> 0:16:42.040 does change things, doesn't it. So, yeah, the Cray two. 0:16:42.600 --> 0:16:46.480 That's when they switched from free on to Flora Nert 0:16:47.400 --> 0:16:50.680 as their coolant. I'm sorry, but that sounds like a 0:16:50.960 --> 0:16:54.760 made up alien name from a from a an animated movie. 0:16:54.800 --> 0:17:03.200 Technically all names are made up. I know. That's I 0:17:03.280 --> 0:17:06.800 just blew your mind. What if there were no hypothetical questions? 0:17:08.440 --> 0:17:13.359 Turn So the Floria Nert. The reason why they switched 0:17:13.480 --> 0:17:16.200 was because they had at that point packed the components 0:17:16.240 --> 0:17:19.040 so tightly together that free on was not efficient enough 0:17:19.040 --> 0:17:21.240 to cool them. So they switched from free on to 0:17:21.280 --> 0:17:25.480 Flora Nert. And it's a little Floria Nert I've had 0:17:25.520 --> 0:17:28.320 around somewhere. And then they also had to figure out 0:17:28.320 --> 0:17:31.520 a new way to access the memory on the create too, 0:17:31.600 --> 0:17:34.280 because at this point they had reached that that point 0:17:34.280 --> 0:17:37.920 that Kray had mentioned earlier about creating a CPU I 0:17:37.960 --> 0:17:41.040 can process information faster than it can pull information in. 0:17:41.800 --> 0:17:46.800 So they found they would actually dedicate processors to getting 0:17:46.880 --> 0:17:52.840 data from memory and funneling it into the central processing units. 0:17:53.200 --> 0:17:56.600 And uh, this was this was really important. It was 0:17:56.640 --> 0:18:01.479 what kind of led the way into threading and loading memory. UH, 0:18:01.720 --> 0:18:05.399 CPUs that have that capability to load information from memory, 0:18:05.520 --> 0:18:08.840 preloading things that kind of came out of this work. 0:18:09.040 --> 0:18:11.679 In fact, a lot of the uh, the advances that 0:18:11.720 --> 0:18:15.439 we see in personal computers um are really possible because 0:18:15.440 --> 0:18:17.840 of the pioneering work that was done in supercomputers. It 0:18:17.920 --> 0:18:21.159 was stuff that that found its way from the engineering 0:18:21.240 --> 0:18:27.360 of supercomputers into personal computers, often a completely different sense 0:18:27.400 --> 0:18:31.560 of scale, but a similar approach. Now, after the Cray Too, 0:18:32.320 --> 0:18:38.080 that's when Japan started to produce some supercomputers that were 0:18:38.200 --> 0:18:41.200 uh that were actually faster than anything that was being 0:18:41.200 --> 0:18:44.480 produced in the United States. So up until this point 0:18:44.480 --> 0:18:48.119 it was all US that was they dominated that that 0:18:48.200 --> 0:18:54.040 country dominated the supercomputer industry. But in so this is 0:18:54.119 --> 0:18:58.679 you know, again, the Cray craze, if you will, lasted 0:18:58.680 --> 0:19:01.160 front the sixties all the way into the eighties. Well 0:19:01.240 --> 0:19:05.200 ninety six, Japan introduced the s R twenty two oh one, 0:19:05.800 --> 0:19:09.119 which had two thousand, forty eight processors. So remember a 0:19:09.200 --> 0:19:13.399 Cray too that was up to eight processors. The s 0:19:13.520 --> 0:19:17.520 R two two oh one two thousand forty eight processors. Yep, 0:19:17.359 --> 0:19:20.640 ye I count two thousand forty more processors with that computer. 0:19:20.720 --> 0:19:23.919 Then with the Cray, do my math could be off 0:19:23.960 --> 0:19:27.200 of an English major and it could it could have 0:19:27.280 --> 0:19:32.360 up to six hundred gigaflops of processing. That's kind of crazy. Um. Yeah. 0:19:32.400 --> 0:19:34.640 I also I also feel like we would be remissed 0:19:34.640 --> 0:19:39.320 to mention the efforts of Danny Hillis um W. Daniel 0:19:39.400 --> 0:19:43.040 Hillis was a grad student at m i T. Massachusetts 0:19:43.040 --> 0:19:47.320 Institute of Technology when he realized that distributing computing was 0:19:47.800 --> 0:19:50.520 the way of the future, if you will. Um. He 0:19:50.640 --> 0:19:55.840 started thinking Machines Corporation ine UM and this CM one, 0:19:55.920 --> 0:19:58.119 which was the first of his machines to come out 0:19:58.160 --> 0:20:05.560 in eight five m It had six one bit processors 0:20:05.680 --> 0:20:10.120 grouped sixteen to a chip. Interesting. Um, that's a that's 0:20:10.160 --> 0:20:17.879 a really interesting approach tiny tiny processors. Yeah, so you 0:20:17.920 --> 0:20:22.359 know wow. But yeah, I didn't come across my um my, 0:20:22.359 --> 0:20:24.800 my research, which is why this is actually really like 0:20:24.960 --> 0:20:27.320 I'm my mind is really as I'm thinking about that 0:20:27.480 --> 0:20:30.280 sort of architecture. That's really an interesting approach. Well, it's 0:20:30.280 --> 0:20:33.600 interesting too to see how different Uh see, Jonathan and 0:20:33.640 --> 0:20:36.439 I do our research separately on purpose so that we 0:20:36.600 --> 0:20:40.399 uh come up with different things on the cases and um, 0:20:40.640 --> 0:20:43.640 so it's funny that that I would have come across that. Also, well, 0:20:43.640 --> 0:20:46.320 I think of Danny hillis because I've seen his name 0:20:46.359 --> 0:20:49.439 a lot in things like a long Now Foundation and 0:20:49.520 --> 0:20:52.399 people with he hangs out with people like Stewart Brandon, 0:20:52.480 --> 0:20:56.760 Kevin Kelly, UM, fascinating people. But UM anyway, Yeah, that 0:20:56.880 --> 0:20:59.760 that's uh, that was one of his contributions. And you 0:20:59.800 --> 0:21:02.720 see that in again in today's machines. I mean we 0:21:02.760 --> 0:21:06.439 have this, you know, with us every day. But you 0:21:06.480 --> 0:21:08.399 know this is uh, this is when we started to 0:21:08.440 --> 0:21:11.879 realize that you don't necessarily have to go buy More's 0:21:11.960 --> 0:21:13.879 Law and wait until next year's chip comes out with 0:21:13.920 --> 0:21:16.840 twice as many processors on it. You can you can 0:21:16.880 --> 0:21:20.159 do this by dividing up the work. Yeah. And and 0:21:20.200 --> 0:21:23.359 in fact that that's another good point about the s 0:21:23.480 --> 0:21:27.719 R O one, the computer from Japan, because uh, in 0:21:27.840 --> 0:21:31.280 order to use these two thousand forty eight processors, there 0:21:31.400 --> 0:21:34.440 was a new development in computer science which was called 0:21:34.600 --> 0:21:38.720 multiple instruction multiple data or m I M D. Yes. Now, 0:21:38.760 --> 0:21:41.760 this is the idea of being able to solve problems 0:21:41.800 --> 0:21:46.119 by pulling in information from from memory and feeding it 0:21:46.160 --> 0:21:50.239 to different processors that are all using different operations on 0:21:50.280 --> 0:21:55.720 that data to come to a single solution, not necessarily 0:21:55.720 --> 0:21:57.800 a single solution, but that's I'm using that as as 0:21:57.800 --> 0:22:02.919 an example for this for this explanation. So this m 0:22:02.960 --> 0:22:06.120 I M D approach is what allowed us to develop 0:22:06.240 --> 0:22:10.200 multi core processors, because in this case we're still talking 0:22:10.200 --> 0:22:14.359 about single processors that are all grouped together. Eventually we 0:22:14.400 --> 0:22:16.400 will get to the point where we have multi core 0:22:16.480 --> 0:22:19.880 processors where a single processor has multiple cores and each 0:22:20.000 --> 0:22:23.280 core can work on part of a problem or separate 0:22:23.320 --> 0:22:28.000 problems to solve things faster, to to get to a 0:22:28.080 --> 0:22:31.280 conclusion faster than they would if it was just one 0:22:31.400 --> 0:22:35.119 single processor, even if it was a really really fast 0:22:35.200 --> 0:22:39.520 processor working on a series of problems. So I always, 0:22:39.600 --> 0:22:43.800 I always use this analogy. Imagine that you have one 0:22:44.040 --> 0:22:49.040 super smart math genius taking a math test, and the 0:22:49.080 --> 0:22:52.120 math genius is going through and solving all of these problems, 0:22:52.600 --> 0:22:56.600 and he or she is able to do this flawlessly, 0:22:58.000 --> 0:22:59.639 able to solve all the problems, but it takes a 0:22:59.640 --> 0:23:02.520 certain time to get through the test. Then you get 0:23:02.560 --> 0:23:06.800 that same test to four above average math students. They're 0:23:06.800 --> 0:23:09.640 not geniuses, but they're there. They can hold their own. 0:23:10.440 --> 0:23:12.760 And you divide it up, say all right, you take 0:23:12.800 --> 0:23:15.840 this this one fourth of the test. You take this quarter, 0:23:15.960 --> 0:23:18.199 you take this quarter, and you take that quarter and 0:23:18.240 --> 0:23:21.240 the four students together start to work. Those four students 0:23:21.280 --> 0:23:23.240 are very likely going to be able to finish the 0:23:23.400 --> 0:23:25.920 entirety of that test much faster, each of them working 0:23:25.920 --> 0:23:28.920 on a quarter of it, rather than the genius who 0:23:28.960 --> 0:23:31.000 is working on the full thing at the same time. 0:23:31.000 --> 0:23:33.240 Even though the genius is smarter and can work faster 0:23:33.320 --> 0:23:37.280 on each individual problem, collectively, those four students are going 0:23:37.320 --> 0:23:41.600 to solve that test faster. That's the philosophy behind both 0:23:42.119 --> 0:23:46.680 grouping cores together and making them a parallel processing unit 0:23:47.240 --> 0:23:50.320 or taking a multi core approach to a CPU. Yep, 0:23:50.440 --> 0:23:52.920 and you can you can thank Danny Hillis for figuring 0:23:53.000 --> 0:23:58.000 out the idea of massively parallel computing UM. But you 0:23:58.040 --> 0:24:01.760 know that that's a problem though too, because instead of 0:24:01.800 --> 0:24:05.159 having two machines running side by side and linked together, 0:24:05.840 --> 0:24:07.760 now you have to figure out how you're going to 0:24:07.880 --> 0:24:11.800 parse all those instructions between all those different processors. So 0:24:11.840 --> 0:24:15.040 you have to have the software or the operating system 0:24:15.080 --> 0:24:20.679 that will give instructions to each of the processors actively 0:24:20.840 --> 0:24:23.760 and direct essentially directing traffic. Yes, this is this is 0:24:23.840 --> 0:24:25.760 kind of like, it's not It's not a simple thing 0:24:25.800 --> 0:24:28.639 to figure out. It reminds me of Intel's tick talk 0:24:28.800 --> 0:24:33.119 approach to developing processors. You think of the tick being 0:24:33.320 --> 0:24:38.800 the physical machinery that's going to do the processing, and 0:24:38.840 --> 0:24:41.919 you think of the talk as the software that's optimized 0:24:42.000 --> 0:24:45.040 to work on that physical hardware to make it really 0:24:45.080 --> 0:24:47.480 live up to its potential. And then you have another 0:24:47.520 --> 0:24:50.720 tick where you've got an advancement in the physical hardware, 0:24:51.000 --> 0:24:54.280 but perhaps the last generation of software isn't really optimized 0:24:54.280 --> 0:24:57.040