1 00:00:00,280 --> 00:00:02,840 Speaker 1: Brought to you by the reinvented two thousand twelve camera. 2 00:00:03,160 --> 00:00:08,800 Speaker 1: It's ready. Are you get in touch with technology? With 3 00:00:08,960 --> 00:00:17,600 Speaker 1: tech Stuff from House Touffi dot com. Hello again, everyone, 4 00:00:17,640 --> 00:00:20,000 Speaker 1: and welcome to tech Stuff. My name is Chris Poulette, 5 00:00:20,040 --> 00:00:22,840 Speaker 1: and I am an editor at how Stuff works dot com. 6 00:00:22,840 --> 00:00:26,320 Speaker 1: Sitting across from me in a cape and tights is 7 00:00:26,440 --> 00:00:32,720 Speaker 1: senior writer Jonathan Strickland. Hey there, citizen. So, so, Chris, 8 00:00:32,880 --> 00:00:34,720 Speaker 1: if I were to ask you, just off the top 9 00:00:34,760 --> 00:00:37,839 Speaker 1: of your head, how would you define a supercomputer? What 10 00:00:37,920 --> 00:00:41,040 Speaker 1: would you say? Well, if I hadn't already made the joke, 11 00:00:41,080 --> 00:00:42,760 Speaker 1: I would have said it was a computer in the 12 00:00:42,800 --> 00:00:45,880 Speaker 1: Cape and tights, But no, I'm honestly I would say 13 00:00:46,400 --> 00:00:50,520 Speaker 1: supercomputer is a computer that can do a lot more 14 00:00:50,840 --> 00:00:54,480 Speaker 1: calculations in a shorter period of time than the machines 15 00:00:54,560 --> 00:00:56,560 Speaker 1: sitting on our desktop. Yeah. I think of it as 16 00:00:56,560 --> 00:00:59,640 Speaker 1: sort of the bleeding edge of what a computer is 17 00:00:59,680 --> 00:01:03,840 Speaker 1: capable of doing. Something that that still feels fills a room, 18 00:01:03,880 --> 00:01:07,040 Speaker 1: even though typical computers these days don't need to fill 19 00:01:07,080 --> 00:01:09,440 Speaker 1: a room because it's that big, it still has that 20 00:01:09,560 --> 00:01:14,080 Speaker 1: much computing power, right right, And the term comes from 21 00:01:14,319 --> 00:01:18,040 Speaker 1: the nineteen sixties and uh. In order to really kind 22 00:01:18,040 --> 00:01:22,479 Speaker 1: of understand the the the span of this, I think 23 00:01:22,800 --> 00:01:24,760 Speaker 1: I was going to talk a little bit about the 24 00:01:25,319 --> 00:01:28,360 Speaker 1: last computer I could find that was a powerful computer 25 00:01:28,880 --> 00:01:33,920 Speaker 1: that existed before people started talking about super computers, which 26 00:01:34,000 --> 00:01:39,120 Speaker 1: was the IBM seventy thirty stretch. Yes, that was the 27 00:01:39,160 --> 00:01:45,080 Speaker 1: one that was made with elastic. Yes, ye gain a 28 00:01:45,080 --> 00:01:47,440 Speaker 1: couple of pounds. Your computer can still you know fit. 29 00:01:47,800 --> 00:01:51,120 Speaker 1: It was Mr. Fantastic of the computer world. No, because 30 00:01:51,120 --> 00:01:53,840 Speaker 1: it was not a super computer. It took up two 31 00:01:53,880 --> 00:01:57,280 Speaker 1: thousand square feet back in the day, this being the 32 00:01:57,920 --> 00:02:01,720 Speaker 1: early sixties, Higger than my two thousand square feet. It 33 00:02:01,760 --> 00:02:05,480 Speaker 1: cost thirteen million dollars, which if you were to translate 34 00:02:05,520 --> 00:02:11,920 Speaker 1: to today's cash, would be ninety one million dollars. It's 35 00:02:11,919 --> 00:02:15,080 Speaker 1: a lot of cash. So that was the fastest computer 36 00:02:15,160 --> 00:02:18,120 Speaker 1: at the time until a fellow named Seymour Roger Craig 37 00:02:18,240 --> 00:02:22,200 Speaker 1: showed up. I s Mr Craze. Yeah. Craig ends up 38 00:02:22,200 --> 00:02:25,960 Speaker 1: being a big name in supercomputers, particularly in the sixties, 39 00:02:26,000 --> 00:02:29,880 Speaker 1: seventies and up to the mid eighties. That was the 40 00:02:30,000 --> 00:02:34,959 Speaker 1: name in supercomputers. And he was working for a company 41 00:02:35,000 --> 00:02:39,120 Speaker 1: called Engineering Research Associates or E r A, which actually 42 00:02:39,200 --> 00:02:43,880 Speaker 1: grew out of a naval operation, um being the U. S. Navy, 43 00:02:43,919 --> 00:02:46,800 Speaker 1: not belly Button. I was gonna you were looking at 44 00:02:46,840 --> 00:02:50,400 Speaker 1: me like joke. No, No, not that it was a 45 00:02:50,520 --> 00:02:53,640 Speaker 1: Navy project. How about that as in as in the 46 00:02:53,680 --> 00:02:57,160 Speaker 1: military force, not the color It was a Navy project 47 00:02:57,400 --> 00:03:00,280 Speaker 1: that was all about code breaking, all right. So there 48 00:03:00,320 --> 00:03:03,080 Speaker 1: was this project about code breaking that eventually kind of 49 00:03:03,120 --> 00:03:06,160 Speaker 1: spun off and became an actual company all on its 50 00:03:06,160 --> 00:03:09,720 Speaker 1: own called Engineering Research Associates, and it branched out beyond 51 00:03:09,800 --> 00:03:12,320 Speaker 1: code breaking, although it took all the code breaking work 52 00:03:12,320 --> 00:03:16,600 Speaker 1: it could get. Yeah, we talked about the Enigma UM 53 00:03:16,680 --> 00:03:20,560 Speaker 1: some episodes back UM and we were talking about the 54 00:03:20,639 --> 00:03:25,359 Speaker 1: bomb UM and yeah, those early uh that was really 55 00:03:25,360 --> 00:03:30,000 Speaker 1: the early application for supercomputers was you know, needing to 56 00:03:30,120 --> 00:03:33,519 Speaker 1: crunch a lot of data very quickly, and there weren't 57 00:03:34,440 --> 00:03:36,440 Speaker 1: There weren't the kind of applications that we have now. 58 00:03:36,480 --> 00:03:38,360 Speaker 1: We'll get into that, I'm sure in a few minutes, 59 00:03:38,400 --> 00:03:41,680 Speaker 1: but but yeah, I mean that was you know, why 60 00:03:41,720 --> 00:03:44,880 Speaker 1: would you need a supercomputer? You know? That was That's 61 00:03:44,920 --> 00:03:47,720 Speaker 1: probably about the only thing I could think of where 62 00:03:47,720 --> 00:03:50,640 Speaker 1: people were needing to crunch that kind of information as 63 00:03:50,720 --> 00:03:56,680 Speaker 1: quickly as possible and defense. Typically, especially with the early supercomputers, 64 00:03:56,720 --> 00:04:01,880 Speaker 1: they were really designed for very specialized suting, so not 65 00:04:01,920 --> 00:04:05,880 Speaker 1: necessarily specialized from the ground up for a one particular 66 00:04:05,920 --> 00:04:08,120 Speaker 1: type of computing, but they were. They were not meant 67 00:04:08,120 --> 00:04:11,640 Speaker 1: to be general computers. They were meant to do tip 68 00:04:12,720 --> 00:04:16,880 Speaker 1: no admiral computers, that's true. Uh No, they were. They 69 00:04:16,920 --> 00:04:19,640 Speaker 1: were meant to do a specific task very very well. 70 00:04:19,720 --> 00:04:22,640 Speaker 1: And that's that's all they were meant to do. Um. Now, 71 00:04:24,160 --> 00:04:27,200 Speaker 1: Craig had an interesting philosophy, he said, and this is 72 00:04:27,240 --> 00:04:29,760 Speaker 1: this is a quote from him. He said, anyone can 73 00:04:29,800 --> 00:04:32,840 Speaker 1: build a fast CPU. The trick is to build a 74 00:04:32,880 --> 00:04:36,919 Speaker 1: fast system. And that was the secret to Craig creating 75 00:04:36,920 --> 00:04:40,640 Speaker 1: the first supercomputer. He realized that if you created a 76 00:04:40,640 --> 00:04:44,559 Speaker 1: processor that was really really fast, that did not matter 77 00:04:44,680 --> 00:04:48,240 Speaker 1: if it couldn't get the data it needed to execute 78 00:04:48,279 --> 00:04:53,200 Speaker 1: operations upon fast enough. So he saw the need to 79 00:04:53,200 --> 00:04:57,120 Speaker 1: create a system that would move data through very very quickly, 80 00:04:57,200 --> 00:04:59,880 Speaker 1: not just processed data, but move it so that means 81 00:05:00,040 --> 00:05:02,280 Speaker 1: needs a lot of memory, It needs a very fast 82 00:05:02,560 --> 00:05:05,680 Speaker 1: pathway from memory to processor. There are a lot of 83 00:05:05,720 --> 00:05:07,599 Speaker 1: pieces that have to be put in place, and he 84 00:05:07,720 --> 00:05:11,760 Speaker 1: saw this very early on, and so using that philosophy, 85 00:05:11,839 --> 00:05:15,000 Speaker 1: he designed a computer back in nineteen sixty two that 86 00:05:15,240 --> 00:05:19,280 Speaker 1: was called the c d C sixty six hundred. Now 87 00:05:19,360 --> 00:05:24,800 Speaker 1: CDC stands for Controlled Data Corporation. Yeah, um, uh e 88 00:05:25,040 --> 00:05:28,679 Speaker 1: R A was taken over by Remington Rand UM and 89 00:05:28,720 --> 00:05:31,680 Speaker 1: that's uh, that's the name I remember because you know, 90 00:05:32,279 --> 00:05:35,120 Speaker 1: UH still remember a lot of those old machine names 91 00:05:35,720 --> 00:05:40,279 Speaker 1: UM from stuff that I found in my uh dad's collection. 92 00:05:40,320 --> 00:05:43,279 Speaker 1: Of course, he was, you know, a mechanical engineer UM 93 00:05:43,400 --> 00:05:46,239 Speaker 1: before he retired, and you know, so he was interested 94 00:05:46,240 --> 00:05:48,479 Speaker 1: in all kinds of machines and I didn't know what 95 00:05:48,520 --> 00:05:50,120 Speaker 1: I was looking at at the time, of course, you know, 96 00:05:50,160 --> 00:05:53,480 Speaker 1: but they were all these UM science and computing magazines 97 00:05:53,520 --> 00:05:56,880 Speaker 1: laying around and that name I recognized also UNITIS because 98 00:05:57,120 --> 00:06:01,000 Speaker 1: Remington Ran became Unities UM, and probably a lot more 99 00:06:01,000 --> 00:06:03,960 Speaker 1: of our listeners are familiar with that name. But he 100 00:06:04,480 --> 00:06:07,880 Speaker 1: partnered with William Norris to start Controlled out of Corporation 101 00:06:08,480 --> 00:06:12,359 Speaker 1: UM back in ninety seven. UM. And really at that point, 102 00:06:12,720 --> 00:06:18,240 Speaker 1: the UNIVAC from Remington Rand and IBM were the computing companies. Yeah, 103 00:06:18,360 --> 00:06:20,520 Speaker 1: you know, IBM has been the heavyweight for so long, 104 00:06:21,480 --> 00:06:25,719 Speaker 1: but CDC was the first uh you know, upstart to 105 00:06:25,839 --> 00:06:30,400 Speaker 1: really make a dent in there, uh stranglehold on the industry. 106 00:06:30,480 --> 00:06:33,840 Speaker 1: And and Craig wanted to join CDC fairly early on, 107 00:06:33,960 --> 00:06:38,920 Speaker 1: but apparently he was needed for a project UM that 108 00:06:39,360 --> 00:06:41,919 Speaker 1: would not let him leave exactly what he wanted to. 109 00:06:42,000 --> 00:06:43,839 Speaker 1: So once he did leave, that's when he designed the 110 00:06:43,880 --> 00:06:49,840 Speaker 1: CDC sixty DRED, which was officially announced in nineteen sixty four, 111 00:06:50,240 --> 00:06:53,400 Speaker 1: so designed in sixty two, announced two years later, and 112 00:06:53,400 --> 00:06:57,039 Speaker 1: it was the first commercially successful supercomputer, with a price 113 00:06:57,120 --> 00:07:00,080 Speaker 1: tag of between seven and eight million dollars, sometimes going 114 00:07:00,120 --> 00:07:02,720 Speaker 1: up as high as ten million, depending upon the configuration 115 00:07:02,800 --> 00:07:05,680 Speaker 1: that the customer wanted. UM Now, in today's cash, that 116 00:07:05,720 --> 00:07:10,280 Speaker 1: would equal about sixty million dollars, so thirty one million 117 00:07:10,320 --> 00:07:16,280 Speaker 1: dollars cheaper in today's money than the Stretch computer, and 118 00:07:16,560 --> 00:07:20,560 Speaker 1: it was actually much more powerful. It had four hundred 119 00:07:20,720 --> 00:07:25,800 Speaker 1: thousand transistors and one hundred miles of wiring, and it 120 00:07:25,880 --> 00:07:28,880 Speaker 1: was the size of about four filing cabinets, so it 121 00:07:28,960 --> 00:07:32,600 Speaker 1: was also significantly smaller than the Stretch, didn't take up 122 00:07:32,600 --> 00:07:35,960 Speaker 1: two thousand square feet. The clock speed was around a 123 00:07:36,040 --> 00:07:41,080 Speaker 1: hundred nanoseconds, and it had sixty five thousand sixty bit 124 00:07:41,240 --> 00:07:44,200 Speaker 1: words of memory. So this is kind of an odd 125 00:07:44,200 --> 00:07:49,360 Speaker 1: time in computing. We hadn't really settled on the thirty 126 00:07:49,400 --> 00:07:54,800 Speaker 1: two sixty four bit kind of model. This was before that. UM. 127 00:07:54,800 --> 00:07:57,280 Speaker 1: It also used six high speed drums as sort of 128 00:07:57,320 --> 00:07:59,880 Speaker 1: a temporary storage area. It had a central storage that 129 00:08:00,080 --> 00:08:03,080 Speaker 1: used magnetic tape, and it used the four trans sixty 130 00:08:03,160 --> 00:08:07,800 Speaker 1: six compiler. UM the equivalent to today's machines means that 131 00:08:07,840 --> 00:08:13,200 Speaker 1: it would have about a ten mega hurts processor. Yeah, 132 00:08:13,440 --> 00:08:16,200 Speaker 1: well that could work up to forty mega hurts and speed. Well, 133 00:08:16,280 --> 00:08:22,600 Speaker 1: it could do a three million floating point operations per second. Yeah, 134 00:08:22,640 --> 00:08:25,360 Speaker 1: so three million. That would be a mega flop. Three 135 00:08:25,360 --> 00:08:28,320 Speaker 1: mega flops, right, So we're gonna get into lots of 136 00:08:28,320 --> 00:08:34,439 Speaker 1: different flop terms later as well, and they get incredibly huge. 137 00:08:34,480 --> 00:08:36,120 Speaker 1: H Of course, you have to keep it cool because 138 00:08:36,120 --> 00:08:39,720 Speaker 1: otherwise it breaks out into a flop sweat. That's true. Uh, well, 139 00:08:39,720 --> 00:08:41,240 Speaker 1: not the flop sweat part, but you do have to 140 00:08:41,280 --> 00:08:44,240 Speaker 1: keep it cool. As we know electronics, when you're running 141 00:08:44,240 --> 00:08:47,199 Speaker 1: electricity through them, one of the byproducts is heat, and heat, 142 00:08:47,360 --> 00:08:50,040 Speaker 1: as it turns out, is not a great thing for 143 00:08:50,360 --> 00:08:56,720 Speaker 1: electronic components. It can make stuff expand contacts can lose connections, 144 00:08:56,840 --> 00:09:00,600 Speaker 1: so that stuff starts to malfunction and entire system could 145 00:09:00,600 --> 00:09:05,840 Speaker 1: shut down. So the CDC STRED had a cooling system 146 00:09:05,880 --> 00:09:11,040 Speaker 1: that was provided by a special chemical free on. Really, yeah, 147 00:09:11,080 --> 00:09:13,280 Speaker 1: they used free on to cool the system. In fact, 148 00:09:13,360 --> 00:09:16,000 Speaker 1: it was they would use free on for a while 149 00:09:16,840 --> 00:09:19,800 Speaker 1: before finally having to switch to a different coolant because 150 00:09:20,600 --> 00:09:24,400 Speaker 1: free on just wasn't efficient enough. Eventually, now at the 151 00:09:25,040 --> 00:09:29,560 Speaker 1: hundred it was still doing the job. So Kari was 152 00:09:29,880 --> 00:09:33,079 Speaker 1: also an innovator in another way. UM, the stretch IBM 153 00:09:33,120 --> 00:09:38,320 Speaker 1: stretched the UM was sort of a hybrid machine. They 154 00:09:38,360 --> 00:09:42,600 Speaker 1: had transistors and vacuum tubes in it UM, and that's 155 00:09:42,800 --> 00:09:45,720 Speaker 1: I think why one of the reasons why craze machines 156 00:09:45,800 --> 00:09:49,400 Speaker 1: were smaller. The sixteen O four, which proceeded the sixty 157 00:09:49,480 --> 00:09:53,600 Speaker 1: six hundred UM, was the one of the very first 158 00:09:53,640 --> 00:10:00,120 Speaker 1: to use transistors only. So there was also a transistor 159 00:10:00,160 --> 00:10:02,520 Speaker 1: machine and so it would take up a lot less 160 00:10:02,520 --> 00:10:04,360 Speaker 1: space than the vacuum tubes. And I would imagine that 161 00:10:05,000 --> 00:10:07,439 Speaker 1: based on my knowledge, my personal knowledge of vacuum tubes, 162 00:10:07,640 --> 00:10:10,240 Speaker 1: might have been a little cooler simply because of that. Yeah, 163 00:10:10,240 --> 00:10:12,000 Speaker 1: I would imagine that they wouldn't have had to have 164 00:10:12,280 --> 00:10:15,960 Speaker 1: as dramatic and a c system to keep the the 165 00:10:16,040 --> 00:10:19,560 Speaker 1: room bearable because vacuum tubes do put off a lot 166 00:10:19,600 --> 00:10:25,040 Speaker 1: of heat. Um Another interesting IBM CDC connection here is 167 00:10:25,080 --> 00:10:28,880 Speaker 1: that Thomas Watson Jr. Which was IBM s C. He 168 00:10:29,080 --> 00:10:32,760 Speaker 1: was IBM CEO at the time, wrote a famous memo 169 00:10:33,520 --> 00:10:38,240 Speaker 1: that time two IBM employees, and he said, last week 170 00:10:38,440 --> 00:10:42,760 Speaker 1: Controlled Data announced the sixty system. I understand that in 171 00:10:42,800 --> 00:10:45,960 Speaker 1: the laboratory developing the system, there are only thirty four people, 172 00:10:46,080 --> 00:10:49,760 Speaker 1: including the janitor. Of these fourteen our engineers and for 173 00:10:49,920 --> 00:10:55,319 Speaker 1: our programmers. Contrasting this modest effort with our vast developmental activities, 174 00:10:55,440 --> 00:10:57,840 Speaker 1: I failed to understand why we have lost our industry 175 00:10:57,920 --> 00:11:00,439 Speaker 1: leadership position by letting someone else offer the world's most 176 00:11:00,440 --> 00:11:07,000 Speaker 1: powerful computer. Craize's response was a reportedly, well, there's your problem. Essentially, 177 00:11:07,000 --> 00:11:10,960 Speaker 1: Craig was saying that, you know, perhaps IBMS approach it 178 00:11:11,000 --> 00:11:17,360 Speaker 1: was a little burdened by size that IBM had grown 179 00:11:17,440 --> 00:11:21,840 Speaker 1: so large that to manage a project like this was 180 00:11:22,160 --> 00:11:24,199 Speaker 1: very difficult to do because it was just too big. 181 00:11:25,440 --> 00:11:28,960 Speaker 1: So that's an interesting idea that an organization needed to 182 00:11:29,040 --> 00:11:32,760 Speaker 1: be kind of small and nimble in order to pull 183 00:11:32,800 --> 00:11:36,760 Speaker 1: something off like creating the world's fastest computer. He followed 184 00:11:36,840 --> 00:11:41,320 Speaker 1: up the c d C sundred with which had a 185 00:11:41,440 --> 00:11:44,839 Speaker 1: sixty five thousand, five hundred thirty six sixty bit word 186 00:11:44,920 --> 00:11:47,240 Speaker 1: memory and a clock speed of twenty seven nano seconds 187 00:11:47,600 --> 00:11:51,720 Speaker 1: uh and actually in practice ran about five times faster 188 00:11:51,880 --> 00:11:55,599 Speaker 1: than the six. But then Cray left c d C 189 00:11:56,280 --> 00:12:00,960 Speaker 1: and he formed his own company, Cray Research two and 190 00:12:01,040 --> 00:12:04,760 Speaker 1: in nine six he introduced the Kray one, which if 191 00:12:04,760 --> 00:12:08,040 Speaker 1: you've ever heard the craze supercomputer, that's what this is. 192 00:12:08,120 --> 00:12:10,720 Speaker 1: It's the crazy one was the first of those. It 193 00:12:10,840 --> 00:12:14,360 Speaker 1: had a clock speed of a well, it's processor ran 194 00:12:14,440 --> 00:12:18,800 Speaker 1: at eighty mega hurts and back at this time these 195 00:12:18,800 --> 00:12:23,520 Speaker 1: supercomputers were still using a single CPU, so that was 196 00:12:23,600 --> 00:12:26,440 Speaker 1: kind of interesting to these were single CPU systems. So 197 00:12:26,440 --> 00:12:30,079 Speaker 1: it had eighty mega hurts processor, sixty four bit system, 198 00:12:30,120 --> 00:12:32,720 Speaker 1: it ran at a hundred thirty six mega flops, so 199 00:12:32,800 --> 00:12:36,160 Speaker 1: one or thirty six million floating operations per second, and 200 00:12:36,200 --> 00:12:39,080 Speaker 1: it had one thousand, six hundred sixty two printed circuit 201 00:12:39,080 --> 00:12:43,880 Speaker 1: boards that made up the components of this computer. It 202 00:12:43,960 --> 00:12:46,400 Speaker 1: costs between five and eight million dollars, depending on how 203 00:12:46,440 --> 00:12:48,839 Speaker 1: you wanted it set up, and in today's cash, that's 204 00:12:48,880 --> 00:12:52,240 Speaker 1: about twenty five million dollars. So we see that the 205 00:12:52,360 --> 00:12:56,679 Speaker 1: processor speed is increasing and the price is coming down. 206 00:12:57,080 --> 00:13:00,080 Speaker 1: Often the size of the computer is decreasing as well, 207 00:13:00,120 --> 00:13:04,920 Speaker 1: although that that also flip flops over the years because 208 00:13:05,360 --> 00:13:09,200 Speaker 1: while the solid state electronics definitely brought the size down, 209 00:13:09,880 --> 00:13:16,640 Speaker 1: eventually the way we pack in more speed requires more space. 210 00:13:16,880 --> 00:13:21,319 Speaker 1: But we'll get into that. Okay, So after the Cray 211 00:13:21,400 --> 00:13:26,959 Speaker 1: one came the Cray x MP in Yeah, this is uh, 212 00:13:27,000 --> 00:13:30,640 Speaker 1: this is interesting because Crai realized also in addition to 213 00:13:31,360 --> 00:13:34,480 Speaker 1: the fact that he knew that the components, the all 214 00:13:34,520 --> 00:13:38,719 Speaker 1: of the components, the entire machine was important and not 215 00:13:38,800 --> 00:13:42,679 Speaker 1: just a processor, he also realized that, uh early on, 216 00:13:42,720 --> 00:13:47,480 Speaker 1: that parallel processing could also speed things up. Um. Now 217 00:13:47,840 --> 00:13:50,680 Speaker 1: it's common for us to have multi core processes in 218 00:13:50,720 --> 00:13:54,640 Speaker 1: our desktop machines or laptops, or in fact, now we're 219 00:13:54,640 --> 00:13:57,880 Speaker 1: starting to see them in our mobile devices. Um. But 220 00:13:59,120 --> 00:14:01,720 Speaker 1: you know, at the time in the seventies and eighties, 221 00:14:01,760 --> 00:14:05,920 Speaker 1: this was still something sort of newish, um, and it's 222 00:14:05,920 --> 00:14:09,280 Speaker 1: not something that everybody realized. Uh. So the x MP 223 00:14:09,679 --> 00:14:15,679 Speaker 1: actually was to Cray one computers linked together um and 224 00:14:16,280 --> 00:14:20,920 Speaker 1: using those two machines together in a multiprocessing effort UM 225 00:14:21,000 --> 00:14:26,000 Speaker 1: they could triple the performance of just one Cray one UM, 226 00:14:26,040 --> 00:14:31,160 Speaker 1: which is something interesting to note. And uh, the Create 227 00:14:31,240 --> 00:14:34,440 Speaker 1: two had four processors in the same machine, and that 228 00:14:34,520 --> 00:14:38,040 Speaker 1: was the first to exceed one billion flops as Britainic 229 00:14:38,080 --> 00:14:40,920 Speaker 1: it tells me. Yeah, uh, it actually could have up 230 00:14:40,960 --> 00:14:46,920 Speaker 1: to eight CPUs c um. The these machines often over 231 00:14:47,000 --> 00:14:51,080 Speaker 1: time were upgraded, so the initial step specs you would 232 00:14:51,120 --> 00:14:53,400 Speaker 1: get when they were first released were one thing, and 233 00:14:53,440 --> 00:14:56,440 Speaker 1: then by the end of the run of production they 234 00:14:56,440 --> 00:14:58,680 Speaker 1: would be better. I mean, which makes sense. I mean 235 00:14:58,760 --> 00:15:00,720 Speaker 1: we see that in computers all the time. We definitely 236 00:15:00,760 --> 00:15:03,400 Speaker 1: we tend to call them different model numbers now, but 237 00:15:03,640 --> 00:15:07,000 Speaker 1: the same sort of thing happens. So back in two 238 00:15:07,000 --> 00:15:10,880 Speaker 1: you had this Cray XMP with a hundred five mega 239 00:15:10,920 --> 00:15:16,040 Speaker 1: hurts CPUs running around two hundred megaflops each. Uh, and 240 00:15:16,600 --> 00:15:18,640 Speaker 1: if they had up to four CPUs you could get 241 00:15:18,680 --> 00:15:22,680 Speaker 1: eight hundred megaflops going and that was pretty impressive and 242 00:15:22,760 --> 00:15:25,320 Speaker 1: had the equivalent, by the way of a hundred and 243 00:15:25,320 --> 00:15:29,040 Speaker 1: twenty eight megabytes of RAM. So yeah, you think about 244 00:15:29,080 --> 00:15:33,400 Speaker 1: that one or twenty eight megabytes of RAM in that 245 00:15:33,520 --> 00:15:37,720 Speaker 1: was considered bleeding edge for a supercomputer. UM and the 246 00:15:37,880 --> 00:15:41,080 Speaker 1: storage units for the Cray XMP were the size of 247 00:15:41,120 --> 00:15:43,160 Speaker 1: a file cabinet and they could hold up to twelve 248 00:15:43,160 --> 00:15:46,640 Speaker 1: gigs of storage because they have a flash in my 249 00:15:46,680 --> 00:15:49,960 Speaker 1: bag with me eight gigs. Yeah, and you can find, 250 00:15:50,280 --> 00:15:52,880 Speaker 1: you can find, you can find twenty gig or more 251 00:15:53,560 --> 00:15:55,520 Speaker 1: flash drives, which you know, you think about that, that's 252 00:15:55,560 --> 00:15:57,960 Speaker 1: something that is small enough for you to carry on 253 00:15:58,000 --> 00:16:00,600 Speaker 1: a key chain. While back in nineteen eighty two you 254 00:16:00,640 --> 00:16:03,840 Speaker 1: had a file cabinet sized device that could hold twelve 255 00:16:03,880 --> 00:16:07,920 Speaker 1: gigs and that was considered massive, like a massive amount 256 00:16:07,920 --> 00:16:12,760 Speaker 1: of information. So, yeah, time really does change things, doesn't it. 257 00:16:12,840 --> 00:16:15,760 Speaker 1: So yeah, the Cry two, that's when they switched from 258 00:16:16,200 --> 00:16:21,720 Speaker 1: free On to Flora Nert as their coolant. I'm sorry, 259 00:16:21,720 --> 00:16:24,600 Speaker 1: but that sounds like a made up alien name from 260 00:16:24,640 --> 00:16:27,800 Speaker 1: a from a an animated movie. Technically all names are 261 00:16:27,840 --> 00:16:35,800 Speaker 1: made up. I know, that's I just blew your mind. 262 00:16:35,920 --> 00:16:41,760 Speaker 1: What if there were no hypothetical questions? Turn on the yes. 263 00:16:42,280 --> 00:16:45,320 Speaker 1: So the Floria Nert the reason why they switched was 264 00:16:45,360 --> 00:16:48,200 Speaker 1: because they had at that point packed the components so 265 00:16:48,320 --> 00:16:50,920 Speaker 1: tightly together that free on was not efficient enough to 266 00:16:50,960 --> 00:16:53,360 Speaker 1: cool them, so they switched from free on to Flora 267 00:16:53,400 --> 00:16:58,200 Speaker 1: Nert and it's a little Floria Nert I've had around somewhere. 268 00:16:58,360 --> 00:17:00,760 Speaker 1: Then they also had to figure a new way to 269 00:17:00,880 --> 00:17:04,000 Speaker 1: access the memory on the Create too, because at this 270 00:17:04,040 --> 00:17:06,560 Speaker 1: point they had reached that that point that Krey had 271 00:17:06,560 --> 00:17:10,880 Speaker 1: mentioned earlier about creating a CPU that can process information 272 00:17:10,960 --> 00:17:15,119 Speaker 1: faster than it can pull information in. So they found 273 00:17:15,240 --> 00:17:19,639 Speaker 1: they would actually dedicate processors to getting data from memory 274 00:17:19,680 --> 00:17:26,080 Speaker 1: and funneling it into the central processing units. And UH, 275 00:17:26,280 --> 00:17:28,760 Speaker 1: this was this was really important. It was what kind 276 00:17:28,760 --> 00:17:32,680 Speaker 1: of led the way into threading and and loading memory 277 00:17:33,520 --> 00:17:37,199 Speaker 1: CPUs that have that capability to load information from memory 278 00:17:37,320 --> 00:17:40,600 Speaker 1: preloading things that kind of came out of this work. 279 00:17:40,840 --> 00:17:43,480 Speaker 1: In fact, a lot of the UH, the advances that 280 00:17:43,520 --> 00:17:47,240 Speaker 1: we see in personal computers UM are really possible because 281 00:17:47,240 --> 00:17:49,600 Speaker 1: of the pioneering work that was done in supercomputers. It 282 00:17:49,680 --> 00:17:52,960 Speaker 1: was stuff that that found its way from the engineering 283 00:17:53,040 --> 00:17:59,160 Speaker 1: of supercomputers into personal computers, often a completely different sense 284 00:17:59,200 --> 00:18:03,320 Speaker 1: of scale, but a similar approach. Now after the Create too, 285 00:18:04,160 --> 00:18:09,879 Speaker 1: that's when Japan started to produce some supercomputers that were 286 00:18:10,000 --> 00:18:13,000 Speaker 1: UH that were actually faster than anything that was being 287 00:18:13,000 --> 00:18:16,200 Speaker 1: produced in the United States. So up until this point, 288 00:18:16,280 --> 00:18:19,840 Speaker 1: it was all the US that was they dominated that, 289 00:18:19,840 --> 00:18:24,440 Speaker 1: that country dominated the supercomputer industry. But in n six, 290 00:18:24,840 --> 00:18:28,920 Speaker 1: so this is you know again, the Kray craze, if 291 00:18:28,960 --> 00:18:31,760 Speaker 1: you will, lasted from the sixties all the way into 292 00:18:31,760 --> 00:18:35,480 Speaker 1: the eighties. Well ninety six, Japan introduced the s R 293 00:18:35,680 --> 00:18:40,040 Speaker 1: twenty two oh one, which had two thousand forty eight processors. 294 00:18:40,320 --> 00:18:43,760 Speaker 1: So remember create two. That was up to eight processors. 295 00:18:44,760 --> 00:18:47,520 Speaker 1: The s R two two oh one two thousand forty 296 00:18:47,560 --> 00:18:51,840 Speaker 1: eight processors. I count two thousand forty more processors with 297 00:18:51,880 --> 00:18:55,480 Speaker 1: that computer. Then with the Cray, do my math could 298 00:18:55,480 --> 00:18:58,399 Speaker 1: be off in an English major and it could it 299 00:18:58,400 --> 00:19:02,200 Speaker 1: could have up to six hi flops of processing. That's 300 00:19:02,280 --> 00:19:05,400 Speaker 1: kind of crazy. Um yeah. I also I also feel 301 00:19:05,400 --> 00:19:07,520 Speaker 1: like we would be remissed to mention the efforts of 302 00:19:07,840 --> 00:19:13,200 Speaker 1: Danny Hillis um W. Daniel Hillis was a grad student 303 00:19:13,240 --> 00:19:16,320 Speaker 1: at m i T. Massachusetts Institute of Technology when he 304 00:19:16,800 --> 00:19:20,560 Speaker 1: realized that distributing computing was the way of the future, 305 00:19:20,600 --> 00:19:24,359 Speaker 1: if you will. UM he started thinking Machines Corporation in 306 00:19:26,040 --> 00:19:29,000 Speaker 1: UM and this CM one, which was the first of 307 00:19:29,080 --> 00:19:32,080 Speaker 1: his machines to come out in eight five. Um it 308 00:19:32,200 --> 00:19:40,320 Speaker 1: had sty six one bit processors grouped sixteen to a chip. Interesting, 309 00:19:41,240 --> 00:19:47,320 Speaker 1: that's a that's a really interesting approach tiny tiny processors. Yeah, 310 00:19:48,600 --> 00:19:53,040 Speaker 1: so you know, but yeah, I didn't come across that 311 00:19:53,200 --> 00:19:55,399 Speaker 1: my um my, my research, which is why this is 312 00:19:55,440 --> 00:19:58,400 Speaker 1: actually really like I'm my mind is really as I'm 313 00:19:58,400 --> 00:20:00,880 Speaker 1: thinking about that sort of archetype. Sure, that's really an 314 00:20:00,880 --> 00:20:03,800 Speaker 1: interesting approach. Well, it's interesting too to see how different 315 00:20:04,480 --> 00:20:07,640 Speaker 1: See Jonathan and I do our research separately on purpose 316 00:20:07,720 --> 00:20:10,679 Speaker 1: so that we uh come up with different things on 317 00:20:10,720 --> 00:20:13,560 Speaker 1: the cases. And um, so it's funny that that I 318 00:20:13,560 --> 00:20:16,119 Speaker 1: would have come across that. Also. Well, I think of 319 00:20:16,600 --> 00:20:18,560 Speaker 1: Danny hillis because I've seen his name a lot in 320 00:20:19,119 --> 00:20:22,399 Speaker 1: things like along Now Foundation and people with he he 321 00:20:22,720 --> 00:20:25,640 Speaker 1: hangs out with people like Stewart Brandon, Kevin Kelly, UM, 322 00:20:25,720 --> 00:20:30,200 Speaker 1: fascinating people. But um anyway, yeah, that that's uh, that 323 00:20:30,280 --> 00:20:32,159 Speaker 1: was one of his contributions. And you see that in 324 00:20:32,320 --> 00:20:35,959 Speaker 1: again in today's machines. I mean, we have this, you know, 325 00:20:36,040 --> 00:20:39,160 Speaker 1: with us every day. But you know this is uh, 326 00:20:39,240 --> 00:20:41,160 Speaker 1: this is when we started to realize that you don't 327 00:20:41,200 --> 00:20:44,520 Speaker 1: necessarily have to go buy More's Law and wait until 328 00:20:44,520 --> 00:20:46,879 Speaker 1: next year's chip comes out with twice as many processors 329 00:20:46,880 --> 00:20:50,680 Speaker 1: on it. You can you can do this by dividing 330 00:20:50,720 --> 00:20:53,119 Speaker 1: up the work. Yeah, and and in fact, that's another 331 00:20:53,560 --> 00:20:57,280 Speaker 1: good point about the SR two oh one, the computer 332 00:20:57,320 --> 00:21:00,600 Speaker 1: from Japan, because, uh, in order to who use these 333 00:21:00,640 --> 00:21:04,520 Speaker 1: two thousand forty eight processors, there was a new development 334 00:21:04,560 --> 00:21:08,320 Speaker 1: in computer science which was called multiple instruction multiple data 335 00:21:08,440 --> 00:21:11,400 Speaker 1: or m I m D. Now, this is the idea 336 00:21:11,520 --> 00:21:15,480 Speaker 1: of being able to solve problems by pulling in information 337 00:21:15,640 --> 00:21:19,679 Speaker 1: from from memory and feeding it to different processors that 338 00:21:19,760 --> 00:21:23,680 Speaker 1: are all using different operations on that data to come 339 00:21:23,720 --> 00:21:28,320 Speaker 1: to a single solution, not necessarily a single solution, but 340 00:21:28,400 --> 00:21:30,640 Speaker 1: that's I'm using that as as an example for this 341 00:21:31,280 --> 00:21:35,760 Speaker 1: for this explanation. So this m I m D approach 342 00:21:35,960 --> 00:21:40,520 Speaker 1: is what allowed us to develop multi core processors, because 343 00:21:40,640 --> 00:21:43,760 Speaker 1: in this case we're still talking about single processors that 344 00:21:43,800 --> 00:21:46,800 Speaker 1: are all grouped together. Eventually we will get to the 345 00:21:46,800 --> 00:21:49,399 Speaker 1: point where we have multi core processors where a single 346 00:21:49,440 --> 00:21:52,879 Speaker 1: processor has multiple cores and each core can work on 347 00:21:53,040 --> 00:21:58,200 Speaker 1: part of a problem or separate problems to solve things faster, 348 00:21:58,320 --> 00:22:01,560 Speaker 1: to to get to a inclusion faster than they would 349 00:22:01,880 --> 00:22:05,560 Speaker 1: if it was just one single processor, even if it 350 00:22:05,640 --> 00:22:09,919 Speaker 1: was a really really fast processor working on a series 351 00:22:09,920 --> 00:22:12,639 Speaker 1: of problems. So I always I always use this analogy. 352 00:22:13,080 --> 00:22:18,240 Speaker 1: Imagine that you have one super smart math genius taking 353 00:22:18,240 --> 00:22:22,240 Speaker 1: a math test, and the math genius is going through 354 00:22:22,280 --> 00:22:25,159 Speaker 1: and solving all of these problems, and he or she 355 00:22:25,840 --> 00:22:30,560 Speaker 1: is able to do this flawlessly, able to solve all 356 00:22:30,560 --> 00:22:32,280 Speaker 1: the problems, but it takes a certain amount of time 357 00:22:32,320 --> 00:22:34,760 Speaker 1: to get through the test. Then you get that same 358 00:22:34,800 --> 00:22:39,399 Speaker 1: test to four above average math students. They're not geniuses, 359 00:22:39,440 --> 00:22:42,840 Speaker 1: but they're there. They can hold their own. And you 360 00:22:42,960 --> 00:22:45,320 Speaker 1: divide it up, say all right, you take this this 361 00:22:45,600 --> 00:22:47,840 Speaker 1: one fourth of the test, you take this quarter, you 362 00:22:47,880 --> 00:22:50,119 Speaker 1: take this quarter, and you take that quarter, and the 363 00:22:50,160 --> 00:22:53,199 Speaker 1: four students together start to work. Those four students are 364 00:22:53,280 --> 00:22:55,600 Speaker 1: very likely going to be able to finish the entirety 365 00:22:55,600 --> 00:22:57,800 Speaker 1: of that test much faster, each of them working on 366 00:22:57,840 --> 00:23:00,840 Speaker 1: a quarter of it, rather than the genius who is 367 00:23:00,840 --> 00:23:02,960 Speaker 1: working on the full thing at the same time. Even 368 00:23:02,960 --> 00:23:05,199 Speaker 1: though the genius is smarter and can work faster on 369 00:23:05,240 --> 00:23:09,359 Speaker 1: each individual problem, collectively those four students are going to 370 00:23:09,400 --> 00:23:14,360 Speaker 1: solve that test faster. That's the philosophy behind both grouping 371 00:23:14,640 --> 00:23:19,119 Speaker 1: cores together and making them a parallel processing unit or 372 00:23:19,240 --> 00:23:22,320 Speaker 1: taking a multi core approach to a CPU. Yep, and 373 00:23:22,400 --> 00:23:25,040 Speaker 1: you can. You can thank Danny Hillis for figuring out 374 00:23:25,080 --> 00:23:29,919 Speaker 1: the idea of massively parallel computing UM. But you know 375 00:23:29,960 --> 00:23:33,879 Speaker 1: that that's a problem though too, because instead of having 376 00:23:34,240 --> 00:23:37,800 Speaker 1: two machines running side by side and linked together, now 377 00:23:37,840 --> 00:23:39,960 Speaker 1: you have to figure out how you're going to parse 378 00:23:40,040 --> 00:23:43,720 Speaker 1: all those instructions between all those different processors. So you 379 00:23:43,760 --> 00:23:46,960 Speaker 1: have to have the software or the operating system that 380 00:23:47,119 --> 00:23:52,760 Speaker 1: will give instructions to each of the processors actively and 381 00:23:52,800 --> 00:23:55,840 Speaker 1: direct essentially directing traffic. Yes, this is this is kind 382 00:23:55,840 --> 00:23:57,679 Speaker 1: of like, it's not. It's not a simple thing to 383 00:23:57,680 --> 00:24:01,119 Speaker 1: figure out. It reminds me of Intel's to talk approach 384 00:24:01,240 --> 00:24:05,800 Speaker 1: to developing processors. You think of the TICK being the 385 00:24:05,840 --> 00:24:10,679 Speaker 1: physical machinery that's going to do the processing, and you 386 00:24:10,720 --> 00:24:13,879 Speaker 1: think of the talk as the software that's optimized to 387 00:24:13,960 --> 00:24:17,000 Speaker 1: work on that physical hardware to make it really live 388 00:24:17,080 --> 00:24:19,520 Speaker 1: up to its potential. And then you have another tick 389 00:24:19,640 --> 00:24:22,879 Speaker 1: where you've got an advancement in the physical hardware, but 390 00:24:22,960 --> 00:24:26,119 Speaker 1: perhaps the last generation of software isn't really optimized to 391 00:24:26,160 --> 00:24:28,840 Speaker 1: work on that, so you have to make new software. 392 00:24:29,119 --> 00:24:31,359 Speaker 1: This is a continuation. In fact, that's one of the 393 00:24:31,400 --> 00:24:36,560 Speaker 1: things that people say is a barrier to artificial intelligence 394 00:24:36,560 --> 00:24:38,879 Speaker 1: to the point of having a a computer that has 395 00:24:38,920 --> 00:24:43,200 Speaker 1: self awareness. It's not necessarily that we can't reach the 396 00:24:43,240 --> 00:24:47,679 Speaker 1: physical uh requirements we would need in order to have 397 00:24:47,680 --> 00:24:50,639 Speaker 1: a computer be able to have some form of self awareness. 398 00:24:50,920 --> 00:24:55,000 Speaker 1: It's the idea that we could throw as much horsepower 399 00:24:55,119 --> 00:24:57,480 Speaker 1: at the problem as we wanted to. Without the software, 400 00:24:57,520 --> 00:25:02,400 Speaker 1: it just won't happen anyway. Get back to supercomputers specifically, 401 00:25:02,440 --> 00:25:05,840 Speaker 1: there's one company name we haven't really mentioned yet, and 402 00:25:05,920 --> 00:25:08,240 Speaker 1: it's big. I mean, we talked about a little bit 403 00:25:08,280 --> 00:25:11,480 Speaker 1: just then, but not in the terms of supercomputers. It's 404 00:25:11,520 --> 00:25:14,919 Speaker 1: a big name in computer architecture, but it wasn't a 405 00:25:14,920 --> 00:25:19,040 Speaker 1: really big name in the whole supercomputer story. And that's Intel. Now, 406 00:25:19,080 --> 00:25:23,080 Speaker 1: Intel was not just sitting back during this whole time. Now, granted, 407 00:25:23,119 --> 00:25:29,560 Speaker 1: Intel's main focus is on enterprise and consumer processors, which 408 00:25:30,080 --> 00:25:34,400 Speaker 1: are not completely analogous to what is you you find 409 00:25:34,400 --> 00:25:38,200 Speaker 1: in supercomputers at this time. Right, that would change, but 410 00:25:38,359 --> 00:25:42,800 Speaker 1: not immediately. But Intel did develop something called the Paragon, 411 00:25:43,920 --> 00:25:47,960 Speaker 1: which was supposed to be, you know, another fantastic supercomputer, 412 00:25:48,440 --> 00:25:52,720 Speaker 1: and it could support up to four thousand processors using 413 00:25:52,760 --> 00:25:56,520 Speaker 1: this m I M D architecture. But it did not 414 00:25:56,880 --> 00:25:59,800 Speaker 1: succeed in the market. It just sort of well, it 415 00:26:00,040 --> 00:26:02,760 Speaker 1: lopped in a different way, the other kind of flawed, 416 00:26:03,080 --> 00:26:06,320 Speaker 1: the bad kind, so that didn't really go anywhere, but 417 00:26:06,400 --> 00:26:10,080 Speaker 1: it did again sort of push this trend of parallel 418 00:26:10,119 --> 00:26:13,840 Speaker 1: processing and m I M D. Uh. The Japanese came 419 00:26:13,840 --> 00:26:17,520 Speaker 1: out with a couple of other computers called as Key 420 00:26:17,600 --> 00:26:21,600 Speaker 1: Read and Asky White Until also had an Asky Read. Um. Yeah. 421 00:26:22,640 --> 00:26:25,440 Speaker 1: Well actually this this goes back to the Comprehensive Test 422 00:26:25,480 --> 00:26:31,800 Speaker 1: Band Treaty. Uh. The United States signed UM they needed 423 00:26:31,800 --> 00:26:35,320 Speaker 1: a certification program for the nuclear weapons that they had 424 00:26:35,560 --> 00:26:39,600 Speaker 1: built up and uh so what they started was the 425 00:26:39,640 --> 00:26:44,679 Speaker 1: Accelerated Strategic Computing Initiative. ASKI with only one eye instead 426 00:26:44,720 --> 00:26:49,040 Speaker 1: of asking characters with two eyes, just to clarify, I'm 427 00:26:49,040 --> 00:26:51,400 Speaker 1: glad you did, thank you, uh, and ask you Read 428 00:26:51,440 --> 00:26:54,520 Speaker 1: Yes was built at Sandy and National Laboratories and No Albuquerque, 429 00:26:54,600 --> 00:26:58,479 Speaker 1: New Mexico, UNTIL helped them out with that, and then 430 00:26:58,600 --> 00:27:01,920 Speaker 1: that was the first machine to get a Tarra flop. Yeah, 431 00:27:02,080 --> 00:27:03,960 Speaker 1: and it was the first one to break the Tarra 432 00:27:04,040 --> 00:27:07,000 Speaker 1: flop barrier. It did that with six thousand, two hundred 433 00:27:07,080 --> 00:27:11,880 Speaker 1: mega hurts pentium pro processors, seventy two of them, well 434 00:27:12,040 --> 00:27:14,919 Speaker 1: six thousand at first. It then eventually was upgraded. The 435 00:27:15,000 --> 00:27:18,080 Speaker 1: very first one had six thousand and the very last 436 00:27:18,080 --> 00:27:22,240 Speaker 1: one had nine thousand M two Xeon processors, and it 437 00:27:22,280 --> 00:27:24,960 Speaker 1: actually hit three point one tarra flops at the end 438 00:27:25,040 --> 00:27:28,399 Speaker 1: of its production life. So yeah, like I said, you know, 439 00:27:28,600 --> 00:27:31,480 Speaker 1: when we give these numbers, there are different ones because 440 00:27:31,560 --> 00:27:35,560 Speaker 1: there's a certain amount that was available when the computer 441 00:27:35,600 --> 00:27:39,000 Speaker 1: first premiered, then there was like the average amount during 442 00:27:39,040 --> 00:27:41,080 Speaker 1: the computer's lifetime, and then the amount that was available 443 00:27:41,119 --> 00:27:43,000 Speaker 1: at the very end of its run time. So these 444 00:27:43,080 --> 00:27:47,399 Speaker 1: numbers do change a little bit depending upon which source 445 00:27:47,440 --> 00:27:50,120 Speaker 1: you're reading in which version of the computer they're looking at, 446 00:27:50,119 --> 00:27:53,480 Speaker 1: because again, these computers are they come in a range 447 00:27:53,480 --> 00:27:57,399 Speaker 1: of models, so not all of them are exactly the same. Now, 448 00:27:57,800 --> 00:28:02,399 Speaker 1: while we talk about uh playing games like chess, you 449 00:28:02,440 --> 00:28:08,760 Speaker 1: know that that's one of the big uh consumer UH 450 00:28:09,200 --> 00:28:13,520 Speaker 1: visibility issues with supercomputer You don't see what supercomputers do. 451 00:28:13,720 --> 00:28:16,760 Speaker 1: And that was a way for them, the IBM, and 452 00:28:16,960 --> 00:28:20,600 Speaker 1: in particular to achieve notice, was taking on people like 453 00:28:20,640 --> 00:28:25,520 Speaker 1: Gary Kasparov chess masters worldwide with a supercomputer kind of 454 00:28:25,560 --> 00:28:30,199 Speaker 1: computer outthink quote unquote out think a human. Well, the 455 00:28:30,240 --> 00:28:32,960 Speaker 1: point of ASKI was again one of those behind the 456 00:28:32,960 --> 00:28:35,280 Speaker 1: scenes thing. It was a very military thing. It was 457 00:28:35,320 --> 00:28:41,040 Speaker 1: more like Whopper in more games. Actually, uh actually exactly 458 00:28:41,080 --> 00:28:45,880 Speaker 1: like that. The point was to simulate nuclear tests. Um. 459 00:28:46,200 --> 00:28:49,960 Speaker 1: And that was why they needed a lot of computing power, 460 00:28:50,720 --> 00:28:53,520 Speaker 1: uh and something a machine that could run a lot 461 00:28:53,520 --> 00:28:57,280 Speaker 1: of calculations very quickly, because they wanted to, uh, you know, 462 00:28:57,360 --> 00:28:59,120 Speaker 1: this is not something you want to do. Hey, well 463 00:28:59,200 --> 00:29:03,160 Speaker 1: let's uh, let's test out fifty nuclear warheads. Yeah this, 464 00:29:03,680 --> 00:29:05,560 Speaker 1: you know. They wanted to do this with a computer 465 00:29:05,600 --> 00:29:08,680 Speaker 1: simulation and uh so that's why they started the initiative. 466 00:29:08,680 --> 00:29:12,960 Speaker 1: It was not a game, but a challenge. Hey let's 467 00:29:13,320 --> 00:29:16,200 Speaker 1: you know, let's keep coming up with newer faster machines 468 00:29:16,240 --> 00:29:20,240 Speaker 1: because we need newer faster machines to run nuclear simulations. Yeah. 469 00:29:20,240 --> 00:29:24,600 Speaker 1: And simulations in general were a big part of what 470 00:29:24,720 --> 00:29:28,120 Speaker 1: these supercomputers were put to use for. I mean like 471 00:29:28,400 --> 00:29:33,920 Speaker 1: climatology for example, weather predictions. That was a big requirement 472 00:29:33,960 --> 00:29:36,440 Speaker 1: as well as supercomputers have been put towards that to 473 00:29:36,560 --> 00:29:41,000 Speaker 1: try and help map and predict climate change and just 474 00:29:41,080 --> 00:29:43,680 Speaker 1: weather patterns, not not just climate but weather day to 475 00:29:43,720 --> 00:29:47,480 Speaker 1: day weather, and also other simulations as well. Not to 476 00:29:47,520 --> 00:29:52,800 Speaker 1: mention crunching data from facilities that generate lots and lots 477 00:29:52,800 --> 00:29:58,160 Speaker 1: of information. So um, things like the CET Institute would 478 00:29:59,040 --> 00:30:02,920 Speaker 1: for extraterrestrial intelligence. Yes that they would use very powerful 479 00:30:02,960 --> 00:30:04,920 Speaker 1: computers to try and crunch all the data they would 480 00:30:04,920 --> 00:30:08,000 Speaker 1: get from radio telescopes. You also had things like the 481 00:30:08,120 --> 00:30:11,640 Speaker 1: Large Hadron Collider and other supercolliders that generate lots and 482 00:30:11,680 --> 00:30:14,280 Speaker 1: lots of data and they need these really fast computers 483 00:30:14,280 --> 00:30:16,880 Speaker 1: in order to process the data and make it meaningful. 484 00:30:17,400 --> 00:30:21,200 Speaker 1: So um. Moving on. So right around this time when 485 00:30:21,240 --> 00:30:24,280 Speaker 1: the asky read comes out, Um, that's when there was 486 00:30:24,280 --> 00:30:29,800 Speaker 1: a shift in supercomputing. So before there were all these 487 00:30:29,920 --> 00:30:35,320 Speaker 1: customized uh computers that had their own processors or had 488 00:30:35,520 --> 00:30:40,400 Speaker 1: thousands of processors running together. Uh. But at this point 489 00:30:40,560 --> 00:30:44,320 Speaker 1: it became possible to actually build a supercomputer with off 490 00:30:44,360 --> 00:30:49,800 Speaker 1: the shelf parts. You could actually get enough computers together 491 00:30:49,880 --> 00:30:53,600 Speaker 1: and linked them together to perform as a supercomputer. And 492 00:30:53,600 --> 00:30:56,880 Speaker 1: this was also when there became a shift to using 493 00:30:56,920 --> 00:31:01,600 Speaker 1: the Linux operating system. UH. So Lenox kind of replaces 494 00:31:01,720 --> 00:31:05,600 Speaker 1: Unix as the OS of choice for people who are 495 00:31:05,640 --> 00:31:08,680 Speaker 1: designing supercomputers, which is nice because now you can tell 496 00:31:08,680 --> 00:31:13,920 Speaker 1: the company nurse never mind. In two thousand two, Japan 497 00:31:14,040 --> 00:31:16,440 Speaker 1: comes back with the as Key White, where it's had 498 00:31:16,520 --> 00:31:21,120 Speaker 1: a thirty five terra flops computer. It was the NBC 499 00:31:21,360 --> 00:31:25,680 Speaker 1: Earth Simulator, and it cost a hair under a billion 500 00:31:25,720 --> 00:31:30,240 Speaker 1: dollars million. It's a lot of hairs, actually millions, a 501 00:31:30,240 --> 00:31:31,840 Speaker 1: lot of hairs. If anyone wants to give me a 502 00:31:31,880 --> 00:31:35,720 Speaker 1: hair in that sense, I will take it. Uh. And 503 00:31:35,800 --> 00:31:38,920 Speaker 1: two thousand four, IBM comes out with the Blue Gene 504 00:31:38,960 --> 00:31:43,640 Speaker 1: slash L computer and had sixteen thousand computer nodes and 505 00:31:43,720 --> 00:31:47,120 Speaker 1: each node had two CPUs. I'm gonna be thinking Bowie 506 00:31:47,200 --> 00:31:50,360 Speaker 1: the rest of the day now, So yeah, thirty two 507 00:31:50,360 --> 00:31:54,400 Speaker 1: thousand CPUs. Ultimately, if my math is correct, and then 508 00:31:54,880 --> 00:31:57,480 Speaker 1: that could run it's seventy terra flops, so twice as 509 00:31:57,520 --> 00:32:00,000 Speaker 1: fast as the Askey White, and a two thousand seven 510 00:32:00,160 --> 00:32:03,160 Speaker 1: version of this could actually manage up to six hundred 511 00:32:03,160 --> 00:32:06,760 Speaker 1: tarra flops and it had a hundred thousand computer nodes, 512 00:32:07,080 --> 00:32:10,160 Speaker 1: so two hundred thousand processors. With that starting to get 513 00:32:10,160 --> 00:32:14,520 Speaker 1: into some preretty ridiculous computers from you know if you're 514 00:32:14,520 --> 00:32:16,920 Speaker 1: looking at it as, Hey, I own a computer that's 515 00:32:16,960 --> 00:32:19,800 Speaker 1: got a single processor. This one has a two hundred 516 00:32:19,880 --> 00:32:23,920 Speaker 1: thousand of them. Yeah. Yeah. It also sort of UHUM 517 00:32:24,440 --> 00:32:29,840 Speaker 1: makes apples claim. In the late nineties, UM sort of silly, UM, 518 00:32:29,880 --> 00:32:35,120 Speaker 1: because well, the federal government classified a supercomputer UM. I 519 00:32:35,120 --> 00:32:36,600 Speaker 1: can't remember exactly when it was, it was in the 520 00:32:36,640 --> 00:32:39,240 Speaker 1: late nineties, and UH as as a machine that would 521 00:32:39,320 --> 00:32:43,000 Speaker 1: run a giga flop and UM IBM when they were 522 00:32:43,000 --> 00:32:46,760 Speaker 1: still running on power process power PC processors. UM, there 523 00:32:46,800 --> 00:32:49,640 Speaker 1: was a Mac that they advertised as being a supercomputer 524 00:32:50,600 --> 00:32:53,360 Speaker 1: because it could reach a giga flop. UM. And I 525 00:32:53,440 --> 00:32:55,960 Speaker 1: just thought at the time it was kind of weird 526 00:32:56,040 --> 00:32:58,880 Speaker 1: to think about, UM, But now it's just kind of 527 00:32:58,920 --> 00:33:01,600 Speaker 1: silly when you take it into context and these these 528 00:33:01,640 --> 00:33:06,200 Speaker 1: actual supercomputers at the time. Uh, A gigga flop is good, 529 00:33:06,240 --> 00:33:09,959 Speaker 1: but no, right, So a mega flop is a million 530 00:33:10,080 --> 00:33:13,640 Speaker 1: floating operations per second, A gigga flop is a billion 531 00:33:13,880 --> 00:33:16,760 Speaker 1: floating operations per second, A tarra flop is a trillion 532 00:33:17,040 --> 00:33:23,000 Speaker 1: floating operations per second. Well, and which is a quadrillion 533 00:33:23,400 --> 00:33:27,479 Speaker 1: floating operations per second per second. Yeah, quadrillion and the 534 00:33:27,520 --> 00:33:32,200 Speaker 1: first supercomputer to hit that and break that barrier was 535 00:33:32,320 --> 00:33:37,800 Speaker 1: another IBM machine, the road Runner, and uh it had 536 00:33:37,840 --> 00:33:40,640 Speaker 1: twenty thousand CPUs and it was the first computer to 537 00:33:40,680 --> 00:33:45,520 Speaker 1: break that pedal flop barrier. So one quadrillion floating operations 538 00:33:45,520 --> 00:33:49,600 Speaker 1: per second, it's a serious machine. In twous ten, there 539 00:33:49,680 --> 00:33:53,920 Speaker 1: was an interesting development because China entered the supercomputer FRAY. 540 00:33:54,000 --> 00:33:56,480 Speaker 1: Now at this point it was really a battle down 541 00:33:56,520 --> 00:33:59,000 Speaker 1: between the United States and Japan, and Germany also has 542 00:33:59,040 --> 00:34:02,280 Speaker 1: quite a few supercomputers as well, but but US and 543 00:34:02,360 --> 00:34:04,640 Speaker 1: Japan were the ones that were stealing the record back 544 00:34:04,680 --> 00:34:07,520 Speaker 1: from between each other. And then China came out with 545 00:34:08,160 --> 00:34:11,920 Speaker 1: a computer which I'm sure I'm gonna mispronounced because I 546 00:34:11,920 --> 00:34:14,759 Speaker 1: I don't know how to pronounce Chinese, but tian hey 547 00:34:14,880 --> 00:34:17,640 Speaker 1: is how it would be spelled in English, and and 548 00:34:17,800 --> 00:34:20,239 Speaker 1: someone's probably gonna say it's sheen hey or something like that, 549 00:34:20,760 --> 00:34:23,800 Speaker 1: if you please let us know, yeah, because I don't. 550 00:34:24,320 --> 00:34:26,919 Speaker 1: But it was a computer from China that could run 551 00:34:27,000 --> 00:34:30,160 Speaker 1: at two point five ped falops and uh it had 552 00:34:30,280 --> 00:34:35,080 Speaker 1: fourteen thousand, three hundred thirty six Intel Xeon X five 553 00:34:35,239 --> 00:34:40,240 Speaker 1: six seven zero CPUs and seven thousand, one D sixty 554 00:34:40,320 --> 00:34:44,960 Speaker 1: eight in video Tesla GPUs and so that was, you know, 555 00:34:45,440 --> 00:34:48,719 Speaker 1: a really impressive machine. That was that that stole all 556 00:34:48,760 --> 00:34:53,880 Speaker 1: the titles away in But also another important moment for 557 00:34:54,040 --> 00:34:58,279 Speaker 1: China in that year was that China developed the sun Way, 558 00:34:58,320 --> 00:35:01,640 Speaker 1: which was slow by super comp Peter standards because they 559 00:35:01,640 --> 00:35:05,279 Speaker 1: could only run a pedal flop UM and they had 560 00:35:05,320 --> 00:35:08,120 Speaker 1: already gotten up to two point five pedal flops. Pedaphlop 561 00:35:08,200 --> 00:35:11,160 Speaker 1: is still incredibly fast people, I'm just slow in general 562 00:35:11,239 --> 00:35:14,440 Speaker 1: terms here relative terms. But the cool thing about the Sunway, 563 00:35:14,440 --> 00:35:17,279 Speaker 1: at least from China's perspective, is that it was the 564 00:35:17,320 --> 00:35:23,080 Speaker 1: first supercomputer China had designed with all Chinese processors, so 565 00:35:23,120 --> 00:35:25,920 Speaker 1: they weren't depending upon some other companies process or some 566 00:35:25,960 --> 00:35:29,600 Speaker 1: other country processors. They wanted to be able to be 567 00:35:30,160 --> 00:35:33,200 Speaker 1: self reliant when it came to developing computers. And so 568 00:35:33,280 --> 00:35:39,839 Speaker 1: that China really pushed it's it's computer engineering industry and 569 00:35:40,080 --> 00:35:44,320 Speaker 1: was able to design you know, the Chinese UM engineers 570 00:35:44,320 --> 00:35:48,279 Speaker 1: were able to design this the supercomputer UM. Then you 571 00:35:48,320 --> 00:35:53,080 Speaker 1: had Fujitsu's K supercomputer, which until recently held the record. 572 00:35:53,880 --> 00:35:56,400 Speaker 1: It was capable of running up to ten peda flops 573 00:35:56,480 --> 00:36:01,239 Speaker 1: with eighty eight thousand one spark sixty war processors, and 574 00:36:01,280 --> 00:36:04,520 Speaker 1: each CPU had sixteen gigabytes of local RAM, and it 575 00:36:04,600 --> 00:36:09,200 Speaker 1: had one thousand, three hundred seventy seven terabytes of memory, 576 00:36:10,080 --> 00:36:15,040 Speaker 1: and eventually it got up to seven five thousand process records. Yeah, 577 00:36:15,120 --> 00:36:19,680 Speaker 1: it sits in Japan's REKN Advanced Institute for Computational Science. 578 00:36:19,760 --> 00:36:24,160 Speaker 1: It sits in it thinks what And that's funny. It's 579 00:36:24,239 --> 00:36:26,520 Speaker 1: asky only spelled in different. I mean the letters are 580 00:36:26,520 --> 00:36:30,319 Speaker 1: in different. Um anyway, sorry, I just noticed that as 581 00:36:30,360 --> 00:36:33,280 Speaker 1: I was looking down in my notes. Um, that's actually 582 00:36:33,320 --> 00:36:35,680 Speaker 1: sort of why we decided to do this now, because 583 00:36:35,680 --> 00:36:38,080 Speaker 1: it was just the week that we're recording this that 584 00:36:38,120 --> 00:36:40,799 Speaker 1: we found out about the test. Now, they do these 585 00:36:41,200 --> 00:36:44,520 Speaker 1: tests twice a year. Every six months, they have the 586 00:36:44,600 --> 00:36:49,640 Speaker 1: top five hundred supercomputer sites. Um, so computers from all 587 00:36:49,680 --> 00:36:52,719 Speaker 1: over the world. Uh. They put them on wheels at 588 00:36:52,719 --> 00:36:55,200 Speaker 1: the top of this big hill and push it down 589 00:36:55,200 --> 00:36:59,560 Speaker 1: the hill. It's like a big computer soapbox derby, you know. 590 00:36:59,680 --> 00:37:04,120 Speaker 1: They uh they give them problems to solve and uh 591 00:37:04,400 --> 00:37:07,560 Speaker 1: see who's the fastest the top five hundred supercomputers in 592 00:37:07,600 --> 00:37:10,000 Speaker 1: the world, which, in a way it's kind of silly, 593 00:37:10,200 --> 00:37:12,759 Speaker 1: but at the same time, very very cool and you 594 00:37:12,760 --> 00:37:14,880 Speaker 1: can actually see the results of this if you want to, 595 00:37:14,960 --> 00:37:18,400 Speaker 1: if you go to top five hundred dot org. Um. 596 00:37:18,520 --> 00:37:22,560 Speaker 1: There there are the organizations that put it on uh 597 00:37:22,800 --> 00:37:24,920 Speaker 1: publishers every year and that happens to be the University 598 00:37:24,960 --> 00:37:28,560 Speaker 1: of Mannheim, Lawrence Berkeley National Laboratory, and the University of 599 00:37:28,640 --> 00:37:33,240 Speaker 1: Tennessee actually do this and they are trying to figure 600 00:37:33,239 --> 00:37:36,520 Speaker 1: out the the fastest, and the fastest was just announced. 601 00:37:36,520 --> 00:37:38,800 Speaker 1: The new fastest was just announced this week, and we 602 00:37:38,840 --> 00:37:40,799 Speaker 1: thought that would be a great time to talk about it. 603 00:37:40,880 --> 00:37:44,920 Speaker 1: It's a machine actually name for a tree. Yes, it 604 00:37:45,080 --> 00:37:49,279 Speaker 1: is the IBM Sequoia. And uh when when we say 605 00:37:49,680 --> 00:37:55,080 Speaker 1: recording this week, the date is June. And so the 606 00:37:55,160 --> 00:37:58,520 Speaker 1: Sequoia has taken the title of fastest supercomputer, which means 607 00:37:58,560 --> 00:38:02,200 Speaker 1: that that's from IBMS, means the USA has the title 608 00:38:02,239 --> 00:38:08,200 Speaker 1: once more, at least until the next Supercomputer Olympics. And um, yeah, 609 00:38:08,440 --> 00:38:11,520 Speaker 1: it's a giant gold medal that is stamped on the 610 00:38:11,560 --> 00:38:14,840 Speaker 1: outside of them. So you're you're probably all asking, hey, 611 00:38:14,880 --> 00:38:19,040 Speaker 1: so what are some stats on this? Uh, the Sequoia computer, 612 00:38:19,200 --> 00:38:21,640 Speaker 1: How how fast can it go? And what what's making 613 00:38:21,640 --> 00:38:23,600 Speaker 1: it take? Well? I do want to point out that 614 00:38:23,680 --> 00:38:27,399 Speaker 1: it is owned by the Department of Energy. UM, so 615 00:38:27,440 --> 00:38:33,319 Speaker 1: this isn't really a military machine. UM. It is at 616 00:38:33,360 --> 00:38:39,560 Speaker 1: the Lawrence Livermore National Laboratory. UM. And yes, the specs 617 00:38:39,640 --> 00:38:42,360 Speaker 1: on this are pretty impressive. I mean it uses seven 618 00:38:42,400 --> 00:38:48,919 Speaker 1: thousand kilowatts. Yeah, it's actually fairly efficient for a supercomputer. Yeah. 619 00:38:48,920 --> 00:38:54,040 Speaker 1: It has one million, five hundred seventy two thousand, eight 620 00:38:54,120 --> 00:38:59,080 Speaker 1: hundred sixty four processors and one point six peta bytes 621 00:38:59,200 --> 00:39:03,520 Speaker 1: of memory. It takes up three thousand, four hundred twenty 622 00:39:03,600 --> 00:39:06,680 Speaker 1: two square feet of space, so we've finally gotten back 623 00:39:06,680 --> 00:39:10,160 Speaker 1: to that those enormous computers. Remember the stretch was two 624 00:39:10,200 --> 00:39:14,640 Speaker 1: thousand square feet. Now this one's three thousand foo square feet. Uh. 625 00:39:14,719 --> 00:39:20,280 Speaker 1: And it can run at sixteen point three two pedaphlops, 626 00:39:20,280 --> 00:39:24,000 Speaker 1: so six point three to pedaphalops faster. Well, not even 627 00:39:24,080 --> 00:39:27,960 Speaker 1: quite that much, because the K eventually got up to 628 00:39:28,000 --> 00:39:32,000 Speaker 1: ten point five, but it is significantly faster than the K. 629 00:39:32,920 --> 00:39:37,359 Speaker 1: So IBM now holds the the distinction of having the 630 00:39:37,400 --> 00:39:42,200 Speaker 1: fastest or having designed the fastest supercomputer in the world. Now, 631 00:39:42,480 --> 00:39:45,160 Speaker 1: I thought it'd be kind of fun too to compare 632 00:39:45,200 --> 00:39:49,359 Speaker 1: that to IBM's Watson computer. Because that made headlines last 633 00:39:49,440 --> 00:39:54,600 Speaker 1: year when Watson was designed in part to compete against 634 00:39:54,680 --> 00:39:57,640 Speaker 1: humans in a very human game. Because we've already talked 635 00:39:57,680 --> 00:40:01,399 Speaker 1: about computers playing chess again humans, we've also talked about 636 00:40:01,400 --> 00:40:03,640 Speaker 1: computers playing other games against humans. In fact, we did 637 00:40:03,640 --> 00:40:07,840 Speaker 1: a whole episode about this particular computer. So IBMS Watson 638 00:40:07,880 --> 00:40:10,560 Speaker 1: was designed to play in a game show Let's Make 639 00:40:10,560 --> 00:40:15,240 Speaker 1: a Deal. So they called out Watson and you didn't 640 00:40:15,239 --> 00:40:19,120 Speaker 1: know what was behind Well, it was it did have 641 00:40:19,120 --> 00:40:22,400 Speaker 1: a dress on. No, it wasn't. It wasn't Let's make 642 00:40:22,400 --> 00:40:25,320 Speaker 1: a Deal. It was Jeopardy and uh. And in Jeopardy, 643 00:40:25,360 --> 00:40:27,160 Speaker 1: of course, you are given an answer. You have to 644 00:40:27,200 --> 00:40:30,319 Speaker 1: come up with the appropriate question. And it's it's really 645 00:40:30,320 --> 00:40:32,400 Speaker 1: tricky for a computer to do this because it's not 646 00:40:32,520 --> 00:40:36,680 Speaker 1: just a matching game where you matching an answer to 647 00:40:36,719 --> 00:40:39,560 Speaker 1: a question. You also have to take in context. Sometimes 648 00:40:39,640 --> 00:40:43,160 Speaker 1: there's word play, sometimes there's a riddle. Um, it's a 649 00:40:43,239 --> 00:40:46,200 Speaker 1: it's a lot more complicated than just question answer. Yeah, 650 00:40:46,280 --> 00:40:50,840 Speaker 1: they they specifically wanted it to play a human game. 651 00:40:51,000 --> 00:40:55,160 Speaker 1: They didn't alter the clues. They're actually clues on this show. 652 00:40:55,160 --> 00:40:57,040 Speaker 1: If you've never seen it, Um, they give you the 653 00:40:57,040 --> 00:40:59,400 Speaker 1: answer and they you are supposed to supply the question, 654 00:40:59,760 --> 00:41:03,680 Speaker 1: and they use wordplay and and things in these clues, 655 00:41:04,239 --> 00:41:07,640 Speaker 1: and they specifically want the IBM engineers specifically wanted it 656 00:41:07,680 --> 00:41:10,160 Speaker 1: to play a human game to to test its natural 657 00:41:10,239 --> 00:41:14,200 Speaker 1: language processing ability. Can it figure out what from context 658 00:41:14,520 --> 00:41:17,640 Speaker 1: what it is you're talking about? And it did very well. Yeah, 659 00:41:17,840 --> 00:41:20,680 Speaker 1: So what was powering the Watson if you want to 660 00:41:20,680 --> 00:41:23,400 Speaker 1: compare it to say the Sequoia, Well, it had a 661 00:41:23,760 --> 00:41:27,719 Speaker 1: it was using ninety IBM power, seven fifty servers in 662 00:41:27,880 --> 00:41:33,200 Speaker 1: ten server racks, and it had sixteen terabytes of memory 663 00:41:33,640 --> 00:41:38,759 Speaker 1: and two thousand eight D eight processors um so or 664 00:41:38,840 --> 00:41:42,000 Speaker 1: processor cores I should say, not just processors, uh, and 665 00:41:42,080 --> 00:41:44,359 Speaker 1: so two thousand that sounds like a lot. But then 666 00:41:44,400 --> 00:41:48,640 Speaker 1: you compare that to the one million, five hundred sixty 667 00:41:48,680 --> 00:41:52,000 Speaker 1: four processors that the Sequoia has and you realize that Watson, 668 00:41:52,320 --> 00:41:57,160 Speaker 1: as far as supercomputers go, doesn't merit mention. It's that 669 00:41:57,600 --> 00:42:01,400 Speaker 1: which again, Watson was this I'm for a very specific purpose, 670 00:42:01,480 --> 00:42:04,200 Speaker 1: this whole natural language being able to recognize that, being 671 00:42:04,239 --> 00:42:09,520 Speaker 1: able to come up with information. That's a very specialized computer. 672 00:42:09,840 --> 00:42:13,799 Speaker 1: So it doesn't necessarily have to have this incredible by 673 00:42:13,880 --> 00:42:18,960 Speaker 1: comparison processing speed and number crunching ability, which might be 674 00:42:19,080 --> 00:42:24,560 Speaker 1: used for other very intensive tasks, so things like very 675 00:42:24,680 --> 00:42:28,239 Speaker 1: very realistic simulations that kind of thing, and predictions. So 676 00:42:28,920 --> 00:42:30,960 Speaker 1: I just wanted to compare that so that people could 677 00:42:31,000 --> 00:42:33,000 Speaker 1: understand because Watson's one of those words that we've heard 678 00:42:33,040 --> 00:42:35,319 Speaker 1: a lot about and we think of that as like 679 00:42:35,360 --> 00:42:38,880 Speaker 1: a supercomputer. But really, if we define supercomputer as a 680 00:42:38,880 --> 00:42:41,360 Speaker 1: computer that has is on that bleeding edge of what 681 00:42:41,400 --> 00:42:44,000 Speaker 1: a computer is capable of doing, it does not it 682 00:42:44,040 --> 00:42:47,799 Speaker 1: doesn't measure up. But when you talk about comparing the 683 00:42:47,800 --> 00:42:51,200 Speaker 1: top five hundred or putting a computer in a chess 684 00:42:51,239 --> 00:42:55,160 Speaker 1: match or in a game of jeopardy, Um, you know 685 00:42:55,200 --> 00:42:56,759 Speaker 1: I was. I made the joke that it was a 686 00:42:56,760 --> 00:42:59,000 Speaker 1: little silly, and yeah, you could. You could say that 687 00:42:59,080 --> 00:43:01,359 Speaker 1: you're using a computer. You could be using it for 688 00:43:01,360 --> 00:43:04,240 Speaker 1: scientific purposes or doing something, and instead you're you're taking 689 00:43:04,280 --> 00:43:07,640 Speaker 1: time off to do something else. But really, um, it's 690 00:43:07,760 --> 00:43:10,640 Speaker 1: nice that for one thing, people understand what it is 691 00:43:10,680 --> 00:43:15,320 Speaker 1: it's a supercomputer is and can do. And also it's uh, 692 00:43:15,400 --> 00:43:18,319 Speaker 1: it's a way to test out these machines and make 693 00:43:18,360 --> 00:43:21,799 Speaker 1: them better. Um, you know, even like I was talking 694 00:43:21,840 --> 00:43:26,440 Speaker 1: about the h the power used by the Sequoia machine, 695 00:43:26,800 --> 00:43:31,960 Speaker 1: it's considerably more efficient than the K computer. UM. The 696 00:43:32,120 --> 00:43:40,359 Speaker 1: seven seven thousand watts beats K's twelve thousand swats. So 697 00:43:40,640 --> 00:43:43,120 Speaker 1: with every every time that they come out with a 698 00:43:43,200 --> 00:43:48,120 Speaker 1: new supercomputer, it's more efficient. They find better ways to 699 00:43:48,200 --> 00:43:51,719 Speaker 1: route instructions, UM, you know, and and they can make 700 00:43:51,760 --> 00:43:54,919 Speaker 1: things smaller than than before. So you really do see 701 00:43:54,920 --> 00:43:57,960 Speaker 1: the implications in in our our everyday computers because now 702 00:43:58,040 --> 00:44:04,040 Speaker 1: we have multi corps processors in um, these everyday devices 703 00:44:04,080 --> 00:44:07,360 Speaker 1: that we use. UM. You don't necessarily need that to 704 00:44:07,400 --> 00:44:11,040 Speaker 1: write a letter or surf the internet, but it does 705 00:44:11,120 --> 00:44:14,920 Speaker 1: make things faster and more efficient. Uh, computers are are 706 00:44:14,960 --> 00:44:19,120 Speaker 1: more reliable. You see advances in operating systems that we 707 00:44:19,200 --> 00:44:23,720 Speaker 1: use every day because UM, the things that they've found out, 708 00:44:24,239 --> 00:44:27,960 Speaker 1: UM in the process of making these supercomputers. They find 709 00:44:28,000 --> 00:44:31,640 Speaker 1: better ways to route instructions in a simpler computer. UM. 710 00:44:32,040 --> 00:44:35,040 Speaker 1: And so it's really worth it to do these these 711 00:44:35,080 --> 00:44:39,040 Speaker 1: tests and UH find out just what a computer can do. So, 712 00:44:39,239 --> 00:44:41,640 Speaker 1: you know, having a challenge just for the fun of it. 713 00:44:41,920 --> 00:44:45,200 Speaker 1: You know, I don't see that necessarily as a bad thing, UM, 714 00:44:45,239 --> 00:44:47,680 Speaker 1: you know, especially when we can we can make advances 715 00:44:47,719 --> 00:44:50,280 Speaker 1: and build on those for the next generation of machines. 716 00:44:51,080 --> 00:44:53,439 Speaker 1: And just to kind of sum this up, I thought 717 00:44:53,440 --> 00:44:55,759 Speaker 1: I would just kind of a fun fact. If you 718 00:44:55,800 --> 00:44:58,840 Speaker 1: look at the top ten fastest supercomputers in the world, 719 00:44:59,760 --> 00:45:03,279 Speaker 1: three of them are in the United States, two of 720 00:45:03,320 --> 00:45:06,480 Speaker 1: them are in Germany, two of them are in China, 721 00:45:07,040 --> 00:45:10,280 Speaker 1: and the other three are in Japan, Italy, and France. 722 00:45:11,040 --> 00:45:14,440 Speaker 1: Uh So that's where you could find these these supercomputers. 723 00:45:14,480 --> 00:45:17,520 Speaker 1: And what I think is even more interesting is, let's 724 00:45:17,520 --> 00:45:22,839 Speaker 1: see one, two, three, four, five of them were made 725 00:45:22,880 --> 00:45:26,160 Speaker 1: by IBM, and only one of them was made by Craig. 726 00:45:27,480 --> 00:45:31,000 Speaker 1: Uh that one being the Jaguar or Jaguar would you prefer, 727 00:45:31,520 --> 00:45:33,320 Speaker 1: which is another wide of the ones in the United States. 728 00:45:33,520 --> 00:45:37,359 Speaker 1: So IBM is definitely dominating the supercomputer space now, even 729 00:45:37,400 --> 00:45:40,439 Speaker 1: though not all of those computers are in the United States, 730 00:45:40,480 --> 00:45:45,520 Speaker 1: but IBM developed five of them, so that's pretty impressive. Uh. 731 00:45:45,880 --> 00:45:49,680 Speaker 1: I guess that kind of sums up our conversation around supercomputers. 732 00:45:49,680 --> 00:45:52,880 Speaker 1: And guys, if you have any ideas for episodes that 733 00:45:52,920 --> 00:45:56,439 Speaker 1: we should cover in the future, let us know. Send 734 00:45:56,520 --> 00:45:59,960 Speaker 1: us an email. Our address is tech stuff at Discovery 735 00:46:00,040 --> 00:46:02,640 Speaker 1: dot com, or let us know on Facebook or Twitter 736 00:46:02,960 --> 00:46:05,320 Speaker 1: or handled. There's text uff hs W and Chris and 737 00:46:05,360 --> 00:46:09,560 Speaker 1: I will talk to you again really soon for more 738 00:46:09,600 --> 00:46:11,880 Speaker 1: on this and thousands of other topics. Is it how 739 00:46:11,920 --> 00:46:19,359 Speaker 1: staff works dot com brought to you by the reinvented 740 00:46:19,400 --> 00:46:22,080 Speaker 1: two thousand twelve camera. It's ready, are you