1 00:00:00,160 --> 00:00:03,000 Speaker 1: In a bit we'll hear my conversation with Dave Turik, 2 00:00:03,320 --> 00:00:07,640 Speaker 1: who oversees ibm S High Performance Computing Division, to learn 3 00:00:07,880 --> 00:00:10,840 Speaker 1: more about all of that. But before we get to that, 4 00:00:11,200 --> 00:00:12,959 Speaker 1: I thought it would be useful to give a quick 5 00:00:12,960 --> 00:00:19,080 Speaker 1: definition and overview of supercomputers. So what is a supercomputer? 6 00:00:19,920 --> 00:00:22,640 Speaker 1: Generally speaking, it's a computer that can perform at a 7 00:00:22,720 --> 00:00:27,560 Speaker 1: level far beyond the average computer. You know, leap tall 8 00:00:27,640 --> 00:00:31,160 Speaker 1: processes at a single bound. It can be a bit 9 00:00:31,200 --> 00:00:34,560 Speaker 1: of a sliding classification. It's something we apply to an 10 00:00:34,560 --> 00:00:37,720 Speaker 1: elite group of computers that operate at a level above 11 00:00:37,760 --> 00:00:40,680 Speaker 1: and beyond what other machines are capable of at that time. 12 00:00:40,960 --> 00:00:42,559 Speaker 1: And I thought it might be a good idea to 13 00:00:42,560 --> 00:00:45,400 Speaker 1: explain what those are, since otherwise the only impression you'll 14 00:00:45,400 --> 00:00:50,000 Speaker 1: get is that more flops equals more good somehow. So 15 00:00:50,120 --> 00:00:54,600 Speaker 1: let's start with floating point numbers and computing, you might 16 00:00:54,680 --> 00:00:57,960 Speaker 1: deal with integers, and these are whole numbers with no fractions. 17 00:00:58,040 --> 00:01:01,360 Speaker 1: Like the number three. Three is a good number, it's 18 00:01:01,400 --> 00:01:05,880 Speaker 1: an integer. But what about point three? Now we have 19 00:01:05,880 --> 00:01:08,240 Speaker 1: a number that has a decimal in it. This is 20 00:01:08,280 --> 00:01:12,560 Speaker 1: not an integer, but it can be a floating point number. Now, 21 00:01:12,640 --> 00:01:15,920 Speaker 1: computers are really good at working with integers. They can 22 00:01:15,959 --> 00:01:20,480 Speaker 1: calculate processes on integers whippity quick, but floating point numbers 23 00:01:20,880 --> 00:01:23,800 Speaker 1: those can take a bit longer, and speed is a 24 00:01:23,800 --> 00:01:27,520 Speaker 1: big deal in computing. You always want answers quickly. But 25 00:01:27,560 --> 00:01:30,200 Speaker 1: the reason we call them floating point numbers is that 26 00:01:30,280 --> 00:01:34,200 Speaker 1: you can move that decimal around so that point three, well, 27 00:01:34,240 --> 00:01:37,319 Speaker 1: we could represent that as three times ten to the 28 00:01:37,319 --> 00:01:41,480 Speaker 1: power of minus one. This is an example of scientific notation, 29 00:01:41,720 --> 00:01:44,800 Speaker 1: something used in lots of disciplines to help represent very 30 00:01:44,920 --> 00:01:48,000 Speaker 1: large or very small numbers without having to write in 31 00:01:48,040 --> 00:01:50,560 Speaker 1: all those darned zeros. For example, if I wanted to 32 00:01:50,560 --> 00:01:53,400 Speaker 1: write out the number two trillion, I would write the 33 00:01:53,480 --> 00:01:58,040 Speaker 1: numeral two followed by twelve zeros. That's a lot of zeros, 34 00:01:58,120 --> 00:02:00,200 Speaker 1: and honestly, if I wanted to do anything use full 35 00:02:00,240 --> 00:02:02,520 Speaker 1: with that number, it would end up being a real hassle. 36 00:02:02,880 --> 00:02:05,840 Speaker 1: But I could represent the same number as two times 37 00:02:05,880 --> 00:02:08,720 Speaker 1: ten to the twelve power. I wanted to give you 38 00:02:08,760 --> 00:02:12,639 Speaker 1: guys a basic understanding of floating point operations because that's 39 00:02:12,680 --> 00:02:16,480 Speaker 1: going to come into play in my discussion in this episode. 40 00:02:16,560 --> 00:02:18,880 Speaker 1: So now that we've got that out of the way, 41 00:02:19,120 --> 00:02:22,519 Speaker 1: we can move on. Dave Turik, vice president of High 42 00:02:22,560 --> 00:02:26,440 Speaker 1: Performance Computing and Cognitive Systems at IBM, spoke with me 43 00:02:26,560 --> 00:02:31,040 Speaker 1: on Thursday April two, twenty twenty about high performance computing 44 00:02:31,080 --> 00:02:33,960 Speaker 1: in general and how researchers are using it in an 45 00:02:34,000 --> 00:02:38,359 Speaker 1: effort to research the coronavirus and COVID nineteen. And I 46 00:02:38,400 --> 00:02:42,480 Speaker 1: should also add we recorded this call over the Internet, 47 00:02:42,919 --> 00:02:45,600 Speaker 1: and so the quality is not the same as what 48 00:02:45,639 --> 00:02:48,160 Speaker 1: we would usually have in a studio. You're going to 49 00:02:48,240 --> 00:02:53,800 Speaker 1: hear some effects because of the Internet connection. You'll probably 50 00:02:53,800 --> 00:02:56,840 Speaker 1: hear some extraneous noise, and I apologize for that, but 51 00:02:57,080 --> 00:03:00,880 Speaker 1: in these extraordinary circumstances, this was the best we could 52 00:03:01,000 --> 00:03:03,640 Speaker 1: manage in order to have this important conversation. And I 53 00:03:03,680 --> 00:03:07,240 Speaker 1: want to thank Dave for his time and patience in 54 00:03:07,639 --> 00:03:10,239 Speaker 1: setting this up, and I really appreciate it. So let's 55 00:03:10,280 --> 00:03:15,160 Speaker 1: jump into it. Dave, before we go into this incredible 56 00:03:15,200 --> 00:03:19,520 Speaker 1: effort that we're seeing from research institutions using supercomputers to 57 00:03:19,919 --> 00:03:22,920 Speaker 1: research the coronavirus and look at treatments for COVID nineteen, 58 00:03:23,360 --> 00:03:26,400 Speaker 1: can you define in broad terms what is actually meant 59 00:03:26,680 --> 00:03:30,480 Speaker 1: by high performance computing? Well, I think, uh, the way 60 00:03:30,520 --> 00:03:34,440 Speaker 1: to think about high performance computing is in terms of 61 00:03:34,480 --> 00:03:37,760 Speaker 1: the nature of the problem first of all, and then 62 00:03:37,880 --> 00:03:41,480 Speaker 1: the kind of computing the supplied against it. So by 63 00:03:41,560 --> 00:03:44,800 Speaker 1: nature of the problem, I mean that it's fundamentally infused 64 00:03:44,840 --> 00:03:51,200 Speaker 1: with mathematical representations of systems or problem types. And then 65 00:03:51,200 --> 00:03:54,920 Speaker 1: from a computing perspective, the kind of technology that puts 66 00:03:54,920 --> 00:03:59,640 Speaker 1: an emphasis on floating point and very quick communications as 67 00:03:59,680 --> 00:04:03,200 Speaker 1: a coal by which those problems are tackled. That just 68 00:04:03,360 --> 00:04:06,960 Speaker 1: helps one distinguish between somebody saying, well, I can solve 69 00:04:07,000 --> 00:04:09,960 Speaker 1: this problem on a phone, right, that's not what we're 70 00:04:09,960 --> 00:04:12,880 Speaker 1: talking about here. The nature of the mathematics are complex 71 00:04:13,480 --> 00:04:16,160 Speaker 1: and sometimes quite extreme, and the computing we required to 72 00:04:16,240 --> 00:04:21,280 Speaker 1: tackle those have similar kinds of capabilities to overcome that complexity. 73 00:04:21,680 --> 00:04:24,480 Speaker 1: So now that we've kind of got to grasp on that, 74 00:04:24,520 --> 00:04:28,680 Speaker 1: we're looking at a sort of a a massive scale 75 00:04:28,800 --> 00:04:35,080 Speaker 1: form of computing that does very complicated processes very very quickly. 76 00:04:35,520 --> 00:04:39,800 Speaker 1: Can you talk a bit about the Higher Performance Computing Consortium? 77 00:04:39,839 --> 00:04:42,760 Speaker 1: What is what is that organization? How did that come about? 78 00:04:44,240 --> 00:04:49,839 Speaker 1: The COVID nineteen HPC Consortium came about roughly ten days 79 00:04:49,839 --> 00:04:55,159 Speaker 1: ago um courtesy of the conversation between our director of 80 00:04:55,200 --> 00:04:58,520 Speaker 1: research Dario Gill and people at the White House and 81 00:04:58,560 --> 00:05:02,520 Speaker 1: subsequently the Department Energy to see how we could apply 82 00:05:02,880 --> 00:05:07,720 Speaker 1: high performance computing or super computing, two problems associated with 83 00:05:08,240 --> 00:05:13,360 Speaker 1: uh COVID nineteen and quite quickly the offers were taken 84 00:05:13,440 --> 00:05:15,840 Speaker 1: up and within a matter of a couple of days 85 00:05:15,880 --> 00:05:19,080 Speaker 1: we had a website up and running. They gave the 86 00:05:19,080 --> 00:05:22,840 Speaker 1: broad parameters of the resources that were available on how 87 00:05:22,880 --> 00:05:25,760 Speaker 1: one could make submissions to it, and then with a 88 00:05:25,839 --> 00:05:28,719 Speaker 1: passage of another couple of days, we brought in a 89 00:05:28,800 --> 00:05:32,760 Speaker 1: number of additional partners as well UM to complement the 90 00:05:32,839 --> 00:05:36,560 Speaker 1: capability that we initially we're able to access of via 91 00:05:36,600 --> 00:05:40,000 Speaker 1: IBM and a Department of Energy excellent and one are 92 00:05:40,040 --> 00:05:44,040 Speaker 1: some of the actual technologies that are being used in 93 00:05:44,080 --> 00:05:47,320 Speaker 1: this process. We've mentioned supercomputers, can you talk about any 94 00:05:47,360 --> 00:05:53,280 Speaker 1: specific ones and UH, what about things like artificial intelligence, 95 00:05:53,320 --> 00:05:57,040 Speaker 1: machine learning or what kind of various tech are coming 96 00:05:57,080 --> 00:06:01,480 Speaker 1: together to tackle this this UH this issue. The glib 97 00:06:01,520 --> 00:06:03,600 Speaker 1: answer of course is everything. But let me be a 98 00:06:03,640 --> 00:06:08,000 Speaker 1: little more specific from the perspective of the supercomputers that 99 00:06:08,040 --> 00:06:12,839 Speaker 1: are part of the consortium currently UH. They range from 100 00:06:12,880 --> 00:06:17,000 Speaker 1: the Power nine based supercomputers that one finds at oak 101 00:06:17,120 --> 00:06:21,080 Speaker 1: Ridge and Lawrence Livermore to x AD six based systems 102 00:06:21,120 --> 00:06:25,680 Speaker 1: that you might find a NASA our gone in other places. UM. 103 00:06:25,720 --> 00:06:28,120 Speaker 1: For the most part, most of these systems, but not 104 00:06:28,160 --> 00:06:33,800 Speaker 1: all of them use accelerators UM and UH, and that 105 00:06:33,880 --> 00:06:37,279 Speaker 1: really deals with some of the floating point computations that 106 00:06:37,320 --> 00:06:40,320 Speaker 1: are involved, and in some cases the systems are are 107 00:06:40,360 --> 00:06:46,400 Speaker 1: absent UM UH accelerators, So those are homogeneous systems. So 108 00:06:46,480 --> 00:06:50,480 Speaker 1: that's the hardware characterization when we begin to talk about 109 00:06:50,560 --> 00:06:53,960 Speaker 1: machine learning deep learning in those things, that's a combination 110 00:06:54,120 --> 00:06:58,960 Speaker 1: of software running in sync with particular hardware attributes. So 111 00:06:59,680 --> 00:07:02,880 Speaker 1: from a deep learning from a model training perspective, there's 112 00:07:02,880 --> 00:07:06,880 Speaker 1: a premium placed on the availability of accelerators. So the 113 00:07:06,920 --> 00:07:10,000 Speaker 1: Summit system at oak Ridge, for examples, and fused with 114 00:07:10,040 --> 00:07:15,200 Speaker 1: about accelerators, so it's terrific for helping people train models. 115 00:07:15,680 --> 00:07:18,160 Speaker 1: But then as you begin to do influencing in some 116 00:07:18,320 --> 00:07:23,600 Speaker 1: of the other machine learning techniques, the emphasis UM exclusively 117 00:07:23,600 --> 00:07:27,240 Speaker 1: on accelerators evolves a little bit and you get to 118 00:07:27,480 --> 00:07:31,680 Speaker 1: employ different kinds of architectural approaches to UH to look 119 00:07:31,720 --> 00:07:35,760 Speaker 1: at actually inferencing problems. So it's a combination of software 120 00:07:35,760 --> 00:07:39,520 Speaker 1: and hardware that's meant to be reasonably flexible. Not One 121 00:07:39,560 --> 00:07:42,040 Speaker 1: of the things I'll say, of course, and oak Ridge 122 00:07:42,360 --> 00:07:44,960 Speaker 1: along with IBM, have been a pioneer in this is 123 00:07:45,000 --> 00:07:49,600 Speaker 1: that there's not a sharp dichotomy between AI R at large, 124 00:07:49,680 --> 00:07:53,040 Speaker 1: which includes machine learning, natural language processing, deep learning, and 125 00:07:53,080 --> 00:07:57,240 Speaker 1: so on, and HPC. In fact, that two domains have 126 00:07:57,360 --> 00:08:00,520 Speaker 1: really come together in the last couple of years where 127 00:08:00,760 --> 00:08:04,640 Speaker 1: problems now get decomposed in ways where maybe certain parts 128 00:08:04,640 --> 00:08:09,120 Speaker 1: of the problems are tackled with classic HPC methodologies and 129 00:08:09,160 --> 00:08:12,080 Speaker 1: other parts of the problems are not tackled with more 130 00:08:12,120 --> 00:08:16,560 Speaker 1: current AI approaches. So it's this amalgamation at capabilities that 131 00:08:16,600 --> 00:08:20,160 Speaker 1: are brought together under software control that creates the impact. 132 00:08:20,880 --> 00:08:23,360 Speaker 1: Dave Turik mentioned a few things I feel I should 133 00:08:23,440 --> 00:08:26,200 Speaker 1: unpack here, and let's start with talking about one of 134 00:08:26,240 --> 00:08:30,080 Speaker 1: the supercomputers he alluded to, the Summit Supercomputer at oak 135 00:08:30,160 --> 00:08:33,520 Speaker 1: Ridge National Laboratory. Now, this is just one of the 136 00:08:33,559 --> 00:08:36,600 Speaker 1: supercomputers that are part of this consortium, and it is 137 00:08:36,640 --> 00:08:40,320 Speaker 1: currently the reigning champ of supercomputers, and researchers are using 138 00:08:40,320 --> 00:08:43,640 Speaker 1: it to do everything from understanding how molecular interactions and 139 00:08:43,679 --> 00:08:48,000 Speaker 1: human cells could lead to much more complex traits uh 140 00:08:48,040 --> 00:08:51,000 Speaker 1: to exploring the physics of propulsion systems and an effort 141 00:08:51,040 --> 00:08:54,400 Speaker 1: to make better, more efficient ones in the future. If 142 00:08:54,440 --> 00:08:58,280 Speaker 1: computers were people, Summit would be that amazing overachiever you 143 00:08:58,320 --> 00:09:01,960 Speaker 1: know who tackles any type of olunge with enthusiasm. Someone 144 00:09:02,040 --> 00:09:05,600 Speaker 1: alone can achieve a peak performance of two hundred pedaphlops. 145 00:09:05,600 --> 00:09:10,880 Speaker 1: That's two hundred thousand trillion calculations of floating point operations 146 00:09:10,880 --> 00:09:14,960 Speaker 1: per second. Dave also mentioned inference problems, and that gets 147 00:09:15,000 --> 00:09:18,640 Speaker 1: down to looking at data and inferring probabilities based on 148 00:09:18,679 --> 00:09:22,000 Speaker 1: the data you've gathered, and building probabilistic tables is an 149 00:09:22,040 --> 00:09:25,680 Speaker 1: important part of science and when done properly, can really 150 00:09:25,760 --> 00:09:28,480 Speaker 1: speed things up. You look at which options appear to 151 00:09:28,559 --> 00:09:31,680 Speaker 1: be the most promising, and you focus on those, and 152 00:09:32,000 --> 00:09:34,320 Speaker 1: you might discard all the ones that have a very 153 00:09:34,400 --> 00:09:37,480 Speaker 1: low probability of being helpful, or at least put them 154 00:09:37,559 --> 00:09:40,760 Speaker 1: to the side. If you exhaust all the most promising 155 00:09:40,760 --> 00:09:43,800 Speaker 1: options without a result, then you can revisit some of 156 00:09:43,840 --> 00:09:45,960 Speaker 1: the other ones. But really it's a great way to 157 00:09:46,000 --> 00:09:49,960 Speaker 1: eliminate options, giving you the ability to focus on the 158 00:09:50,000 --> 00:09:53,640 Speaker 1: best chance for success. Let's get back to the interview. 159 00:09:54,320 --> 00:09:58,680 Speaker 1: So with COVID nineteen in particular, what are the ways 160 00:09:58,720 --> 00:10:02,160 Speaker 1: some of the ways that reas searchers are leveraging these 161 00:10:02,200 --> 00:10:07,480 Speaker 1: technologies to specifically look at that crisis. So I think 162 00:10:07,640 --> 00:10:11,400 Speaker 1: the first way to think of it is to just 163 00:10:11,480 --> 00:10:15,960 Speaker 1: take a second and inform your listeners about the modern 164 00:10:16,000 --> 00:10:20,520 Speaker 1: ways in which chemistry, biology and biochemistry are done, because 165 00:10:20,559 --> 00:10:23,240 Speaker 1: I think many lay people have this image from their 166 00:10:23,280 --> 00:10:26,760 Speaker 1: high school or college days of speakers and pipettes and 167 00:10:26,800 --> 00:10:29,680 Speaker 1: things like that, sort of the what I would characterize 168 00:10:29,720 --> 00:10:32,880 Speaker 1: the representation of science in the analog world, what you 169 00:10:33,000 --> 00:10:36,040 Speaker 1: touch and feel and deal with every day. But what's 170 00:10:36,080 --> 00:10:39,679 Speaker 1: transpired over the last several decades is this movement to 171 00:10:40,280 --> 00:10:44,680 Speaker 1: progressively infuse science and the scientific method with more and 172 00:10:44,720 --> 00:10:48,800 Speaker 1: more computational capability. Now, what that comes down to in 173 00:10:48,800 --> 00:10:52,440 Speaker 1: the case of COVID nineteen is one begins to take 174 00:10:52,520 --> 00:10:57,800 Speaker 1: first principles kinds of theories of the way adams are structure, 175 00:10:58,000 --> 00:11:00,720 Speaker 1: molecules of structure, in the way adams be a and 176 00:11:00,760 --> 00:11:05,560 Speaker 1: how they interact with one another, represent that mathematical form, 177 00:11:05,640 --> 00:11:08,959 Speaker 1: and use the computers to explore the behavior from a 178 00:11:09,040 --> 00:11:13,959 Speaker 1: first principle's perspective. Before you ever get to a physical laboratory. 179 00:11:14,160 --> 00:11:17,120 Speaker 1: So what that nets out to is you can now 180 00:11:17,240 --> 00:11:20,760 Speaker 1: use the power of computing to assess thousands and hundreds 181 00:11:20,760 --> 00:11:25,160 Speaker 1: of thousands of molecules in terms of their potential impact 182 00:11:25,200 --> 00:11:28,400 Speaker 1: on the virus and explore the behavior and the and 183 00:11:28,480 --> 00:11:34,320 Speaker 1: the constraints and the amplifications of combinations of molecules digitally 184 00:11:34,840 --> 00:11:37,520 Speaker 1: before you ever have to go to the laboratory to 185 00:11:37,600 --> 00:11:40,520 Speaker 1: try to recreate the results you've seen digitally in the 186 00:11:40,520 --> 00:11:44,720 Speaker 1: analog world. And that's been a tremendous speed up in 187 00:11:44,840 --> 00:11:48,280 Speaker 1: terms of time. You know, pharmaceutical companies today they may 188 00:11:48,320 --> 00:11:51,959 Speaker 1: have at their disposal billions of molecules that they might 189 00:11:52,000 --> 00:11:55,920 Speaker 1: want to look at for particular pharmaceutical impact, and sorting 190 00:11:55,960 --> 00:11:59,560 Speaker 1: through that is just gigantic task. And the ability to 191 00:11:59,600 --> 00:12:02,440 Speaker 1: have co uters to come in and say, look, I 192 00:12:02,480 --> 00:12:05,760 Speaker 1: know you're looking at eight thousand molecules here, which is 193 00:12:05,800 --> 00:12:08,760 Speaker 1: what researchers at oak Ridge did, but I can cut 194 00:12:08,800 --> 00:12:12,319 Speaker 1: that down to seventy seven just by using digital approaches 195 00:12:12,760 --> 00:12:16,160 Speaker 1: and simulation and computation, so that you don't have to 196 00:12:16,200 --> 00:12:18,439 Speaker 1: worry about trying to analyze all eight thousands and the 197 00:12:18,520 --> 00:12:22,400 Speaker 1: laboratory you can focus on seventy seven so that's the 198 00:12:22,480 --> 00:12:26,680 Speaker 1: first big step of what's happening here. No, that's amazing 199 00:12:26,720 --> 00:12:29,400 Speaker 1: because just the idea of cutting out that step in 200 00:12:29,440 --> 00:12:34,679 Speaker 1: the wet lab where you're having to physically uh analyze 201 00:12:34,880 --> 00:12:38,360 Speaker 1: reactions or maybe not even analyze, you're just detecting to 202 00:12:38,360 --> 00:12:41,880 Speaker 1: see if one is happening. Cutting that downs that you 203 00:12:41,880 --> 00:12:48,319 Speaker 1: can really focus on the best uh potential solutions is phenomenal. 204 00:12:48,400 --> 00:12:49,960 Speaker 1: Can we talk a little bit about what is it 205 00:12:50,080 --> 00:12:54,840 Speaker 1: about these simulations that make them so challenging that high 206 00:12:54,840 --> 00:12:59,920 Speaker 1: performance computing is suitable for tackling that kind of thing? Well, 207 00:13:00,080 --> 00:13:03,760 Speaker 1: I think that um. One of the principal methodologies that 208 00:13:03,840 --> 00:13:09,200 Speaker 1: people used in these investigations is molecular dynamics, and what 209 00:13:09,360 --> 00:13:12,800 Speaker 1: that entails is, first of all, the characterization of a 210 00:13:12,880 --> 00:13:18,480 Speaker 1: molecule in atoms, and and then the application of forces 211 00:13:18,520 --> 00:13:21,760 Speaker 1: at the atomic level in terms of how they interact 212 00:13:21,800 --> 00:13:25,720 Speaker 1: with one another. And so those forces are complicated, the 213 00:13:25,840 --> 00:13:29,560 Speaker 1: time steps are extraordinarily small, and yet you want to 214 00:13:29,600 --> 00:13:33,240 Speaker 1: observe how these things interact. Not only yet, let's say 215 00:13:33,320 --> 00:13:36,440 Speaker 1: tend to the minus fifteen seconds, which actually like to 216 00:13:36,480 --> 00:13:40,480 Speaker 1: see how they behave in in in real seconds, in 217 00:13:40,600 --> 00:13:44,720 Speaker 1: minutes and hours and days, and those time scales just 218 00:13:44,840 --> 00:13:49,080 Speaker 1: create a tremendous number of computational steps that one has 219 00:13:49,160 --> 00:13:52,240 Speaker 1: to pursue in the concept of looking at these atomic 220 00:13:52,320 --> 00:13:55,800 Speaker 1: forces that are operating on the target molecules and atoms 221 00:13:55,800 --> 00:13:58,679 Speaker 1: and how they interact with one another. So the mathematics 222 00:13:58,679 --> 00:14:03,600 Speaker 1: is stunningly complex. The time frames are just so extreme 223 00:14:03,720 --> 00:14:06,280 Speaker 1: that it requires tremendous amount of compute power just to 224 00:14:06,360 --> 00:14:10,040 Speaker 1: simulate a handful of seconds. And by virtue of having 225 00:14:10,120 --> 00:14:13,480 Speaker 1: gigantic supercomputers operate on this, we can actually do this 226 00:14:14,200 --> 00:14:16,600 Speaker 1: in a reasonable way and a reasonable amount of wall 227 00:14:16,640 --> 00:14:20,680 Speaker 1: clock time. So let's consider what researchers are doing in 228 00:14:20,720 --> 00:14:23,920 Speaker 1: this case. A virus consists of at least two parts. 229 00:14:23,960 --> 00:14:27,120 Speaker 1: You've got a nucleic acid genome, which contains the material 230 00:14:27,160 --> 00:14:29,440 Speaker 1: the virus needs to make copies of itself once it 231 00:14:29,560 --> 00:14:32,800 Speaker 1: is a proper host cell. And then you've got a 232 00:14:32,880 --> 00:14:37,400 Speaker 1: protein capsid or shell that contains the nucleic acid until 233 00:14:37,400 --> 00:14:40,320 Speaker 1: the virus can attach itself and inject that material into 234 00:14:40,440 --> 00:14:44,920 Speaker 1: the aforementioned host cell. Together, this is called the nucleocapsid. 235 00:14:45,320 --> 00:14:48,640 Speaker 1: Many animal viruses also have a lipid envelope, and that 236 00:14:48,800 --> 00:14:51,000 Speaker 1: is a membrane that has lots of stuff in it, 237 00:14:51,040 --> 00:14:54,360 Speaker 1: including viral y programmed proteins in it. One of the 238 00:14:54,360 --> 00:14:58,359 Speaker 1: purposes of those proteins is to bind with compatible receptors 239 00:14:58,360 --> 00:15:01,040 Speaker 1: on host cells. So can kind of think of it 240 00:15:01,160 --> 00:15:04,080 Speaker 1: as a virus has a special kind of plug and 241 00:15:04,120 --> 00:15:07,040 Speaker 1: it's looking for cells that have a compatible outlet, and 242 00:15:07,080 --> 00:15:09,840 Speaker 1: when it finds such a cell, it can plug in 243 00:15:10,080 --> 00:15:12,960 Speaker 1: connect to that cell, injecting the nucleic acid of the 244 00:15:13,040 --> 00:15:16,000 Speaker 1: virus into the host cell, and then the code in 245 00:15:16,040 --> 00:15:18,920 Speaker 1: that nucleic acid hijacks the host cell turns it into 246 00:15:18,960 --> 00:15:23,760 Speaker 1: a virus replication engine. Scientists need to know how specific 247 00:15:23,800 --> 00:15:28,200 Speaker 1: molecules will interact with each other, the virus, host cells, 248 00:15:28,200 --> 00:15:31,800 Speaker 1: and more. These interactions happen at such a small scale, 249 00:15:31,920 --> 00:15:36,880 Speaker 1: and it's such minute slices or steps of time that 250 00:15:37,000 --> 00:15:39,160 Speaker 1: it is difficult to describe. And this is where the 251 00:15:39,160 --> 00:15:42,400 Speaker 1: speed of high performance computing really comes into play. First, 252 00:15:42,960 --> 00:15:46,840 Speaker 1: breaking down elements of time gets mind boggling. We tend 253 00:15:46,880 --> 00:15:49,400 Speaker 1: to think of it in terms of, as Dave says, 254 00:15:49,680 --> 00:15:53,680 Speaker 1: wall clock time, you know, seconds, minutes, and hours. We 255 00:15:53,680 --> 00:15:56,520 Speaker 1: can get our minds wrapped around shorter slices of time 256 00:15:56,800 --> 00:16:00,320 Speaker 1: because counting on the second might sound like one mrs cippy, 257 00:16:00,800 --> 00:16:04,200 Speaker 1: so we can definitely think of just one right, that's shorter. 258 00:16:04,640 --> 00:16:06,400 Speaker 1: But eventually we hit a point where it's hard for 259 00:16:06,440 --> 00:16:11,760 Speaker 1: us to really understand time at very tiny slices. We 260 00:16:11,800 --> 00:16:15,640 Speaker 1: can always find a way to slice time into smaller increments. 261 00:16:15,680 --> 00:16:20,160 Speaker 1: We can continue to make smaller and smaller slices of time. 262 00:16:20,560 --> 00:16:23,640 Speaker 1: For example, there's a femto second. A femto second is 263 00:16:23,680 --> 00:16:27,760 Speaker 1: just one quadrillionth of a second. That's tend to the 264 00:16:27,800 --> 00:16:33,320 Speaker 1: power of minus fifteen. So imagine simulating the interactions between 265 00:16:33,360 --> 00:16:37,720 Speaker 1: molecules in a series of these unimaginably short slices of 266 00:16:37,760 --> 00:16:41,360 Speaker 1: time up to the point that collectively they amount to 267 00:16:41,760 --> 00:16:44,720 Speaker 1: enough that it would reach our perceptible world. So we're 268 00:16:44,720 --> 00:16:48,120 Speaker 1: talking about a quadrillion slices of time to make up 269 00:16:48,160 --> 00:16:51,680 Speaker 1: just one second here, and there could be numerous important 270 00:16:51,720 --> 00:16:55,200 Speaker 1: interactions on the molecular level within that short time frame. 271 00:16:55,320 --> 00:16:58,000 Speaker 1: And this is why supercomputers are necessary for this sort 272 00:16:58,000 --> 00:17:00,560 Speaker 1: of work. It allows for a precision and that we 273 00:17:00,600 --> 00:17:03,960 Speaker 1: otherwise would find impossible. And again it tells us if 274 00:17:03,960 --> 00:17:07,280 Speaker 1: a potential molecule shows promise in our efforts, or if 275 00:17:07,320 --> 00:17:10,080 Speaker 1: it's likely to be a bust. Back to my conversation 276 00:17:10,080 --> 00:17:12,879 Speaker 1: with Dave Turik, vice president of High Performance Computing and 277 00:17:12,920 --> 00:17:16,879 Speaker 1: Cognitive Systems at IBM. With supercomputers being able to tackle 278 00:17:16,920 --> 00:17:20,679 Speaker 1: this kind of thing through their various methodologies, this I 279 00:17:20,680 --> 00:17:22,399 Speaker 1: would imagine would be something that if we were to 280 00:17:22,520 --> 00:17:26,000 Speaker 1: use a classic computer, it could take thousands of years. 281 00:17:26,040 --> 00:17:31,960 Speaker 1: Is that accurate? Yes? Um, And and in some sense 282 00:17:32,000 --> 00:17:37,320 Speaker 1: it wouldn't even be possible because modern supercomputers, which I'll 283 00:17:37,320 --> 00:17:44,280 Speaker 1: declare is roughly the error from UM, really are systems 284 00:17:44,320 --> 00:17:48,200 Speaker 1: that are built on this concept of parallel processing parallel computing, 285 00:17:48,720 --> 00:17:52,880 Speaker 1: which in turn revolves around this idea that you can 286 00:17:52,920 --> 00:17:57,000 Speaker 1: decompose a problem into its component elements, and if you 287 00:17:57,080 --> 00:18:01,159 Speaker 1: have enough elemental computing entities in your supercomputer, you can 288 00:18:01,200 --> 00:18:04,639 Speaker 1: assign each one of those little problem parts to a 289 00:18:04,760 --> 00:18:08,919 Speaker 1: different computing yell element and orchestrate the execution of the 290 00:18:08,920 --> 00:18:13,000 Speaker 1: computation against that and and just radically reduce the amount 291 00:18:13,000 --> 00:18:16,160 Speaker 1: of time required from for computation. So let me put 292 00:18:16,200 --> 00:18:20,680 Speaker 1: it this way, UM, on a standard laptop computer, for example, 293 00:18:20,680 --> 00:18:27,720 Speaker 1: you're gonna be running, principally to some order of magnitude UM, 294 00:18:27,760 --> 00:18:30,800 Speaker 1: a serial kind of process. You know, you're gonna execute 295 00:18:30,800 --> 00:18:34,000 Speaker 1: and solve problem A, which is followed by B, C, 296 00:18:34,320 --> 00:18:37,520 Speaker 1: D and so on. In the parallel world, you'll take A, B, C, 297 00:18:37,720 --> 00:18:39,840 Speaker 1: and D and you'll all run them at the same time, 298 00:18:40,480 --> 00:18:43,560 Speaker 1: but in different parts of the supercomputer, and then through 299 00:18:43,600 --> 00:18:48,280 Speaker 1: software orchestration, you'll sort of coalesce all those outputs and 300 00:18:48,359 --> 00:18:53,080 Speaker 1: render a conclusion based on the set of calculations you've run. Now, 301 00:18:53,119 --> 00:18:56,679 Speaker 1: I gave an example of maybe a decomposition to four pieces, 302 00:18:56,720 --> 00:18:59,199 Speaker 1: but what we really may be talking about maybe a 303 00:18:59,240 --> 00:19:02,800 Speaker 1: hundred thousand or or a million or ten million pieces 304 00:19:03,400 --> 00:19:06,200 Speaker 1: and uh, and it's very complicated to try to orchestrate 305 00:19:06,240 --> 00:19:09,359 Speaker 1: all that activity. So a laptop computer doesn't have the 306 00:19:09,400 --> 00:19:12,680 Speaker 1: ability to do that. And that's why when people think 307 00:19:12,720 --> 00:19:16,640 Speaker 1: about some supercomputers, they sort of render it in terms of, well, 308 00:19:16,640 --> 00:19:19,640 Speaker 1: this is the equivalent of what ten million laptops could 309 00:19:19,640 --> 00:19:22,159 Speaker 1: do or a hundred million laptops could do. But I 310 00:19:22,280 --> 00:19:27,280 Speaker 1: remember laptops or standalone entities. In the supercomputer world, all 311 00:19:27,320 --> 00:19:30,760 Speaker 1: of those computing entities have to be managed, and it 312 00:19:30,880 --> 00:19:33,560 Speaker 1: has to be brain power to orchestrate the way they 313 00:19:33,600 --> 00:19:37,920 Speaker 1: tackle the problem. And the supercomputers are really architected to 314 00:19:38,119 --> 00:19:41,480 Speaker 1: handle that, right, So this is this is a specific, 315 00:19:41,560 --> 00:19:45,119 Speaker 1: purpose built approach to that problem, whereas we've seen things 316 00:19:45,160 --> 00:19:48,920 Speaker 1: like grid computing as sort of an ad hoc approach 317 00:19:49,040 --> 00:19:52,320 Speaker 1: that problem, where it tries to do a similar thing, 318 00:19:52,359 --> 00:19:58,160 Speaker 1: but obviously at exponentially lower levels of processing capability. And 319 00:19:58,240 --> 00:20:01,119 Speaker 1: when we're talking about things like floating point operations. Just 320 00:20:01,200 --> 00:20:03,280 Speaker 1: for you guys out there, you listeners out there, you 321 00:20:03,280 --> 00:20:06,399 Speaker 1: know you might have seen a graphics processing unit that 322 00:20:06,440 --> 00:20:09,000 Speaker 1: talks about things in the Tarraf flop range, which you're 323 00:20:09,000 --> 00:20:12,200 Speaker 1: talking about, you know, a million million floating point operations 324 00:20:12,200 --> 00:20:15,800 Speaker 1: per second. We're looking at Pata flop ranges here, a 325 00:20:15,880 --> 00:20:19,520 Speaker 1: thousand million million floating point operations per second. As I 326 00:20:19,600 --> 00:20:24,359 Speaker 1: understand it, which that's incredible. Uh, it's again for someone 327 00:20:24,400 --> 00:20:26,760 Speaker 1: like me, maybe it's just my limited imagination. I have 328 00:20:26,880 --> 00:20:30,200 Speaker 1: real trouble putting this into a context that I can 329 00:20:30,240 --> 00:20:34,680 Speaker 1: get my hands around. But it's it's an incredibly fascinating thing. 330 00:20:34,680 --> 00:20:37,840 Speaker 1: And this is not this isn't like it's unprecedented. We've 331 00:20:37,840 --> 00:20:43,840 Speaker 1: seen researchers, doctors, scientists use supercomputers to research stuff like 332 00:20:43,960 --> 00:20:48,959 Speaker 1: vaccines for for the flu before as well. Right, well, absolutely, 333 00:20:48,960 --> 00:20:51,919 Speaker 1: in fact, when h one N one came out. I 334 00:20:51,920 --> 00:20:54,920 Speaker 1: guess it was around two thousand and nine. IBM S 335 00:20:54,960 --> 00:20:59,920 Speaker 1: Computational Biology group actually began to model the evolutionary tripe 336 00:21:00,080 --> 00:21:03,400 Speaker 1: victory of the virus because if you think about viruses, 337 00:21:03,600 --> 00:21:06,480 Speaker 1: and I don't want people to be confused that I'm 338 00:21:06,560 --> 00:21:10,200 Speaker 1: equating flu to corona, But if you think about flu 339 00:21:10,359 --> 00:21:14,400 Speaker 1: for a second, the virus is not a static thing. 340 00:21:14,640 --> 00:21:17,480 Speaker 1: It will evolve over the course of time, and that's 341 00:21:17,520 --> 00:21:19,880 Speaker 1: why you have a different kind of flu shot every year. 342 00:21:20,119 --> 00:21:23,920 Speaker 1: It's it's an effort to try to create a vaccine 343 00:21:23,960 --> 00:21:28,520 Speaker 1: to intercept the next generation of where this where this 344 00:21:28,800 --> 00:21:32,160 Speaker 1: particular virus has evolved too. And the way you do that, 345 00:21:32,200 --> 00:21:34,840 Speaker 1: the way the industry does that is they use computational 346 00:21:34,880 --> 00:21:38,880 Speaker 1: techniques to kind of predict the evolutionary pathway and they 347 00:21:38,960 --> 00:21:42,800 Speaker 1: build their vaccine to target where the where the virus 348 00:21:42,840 --> 00:21:45,000 Speaker 1: will be in three months as opposed to where it 349 00:21:45,080 --> 00:21:48,400 Speaker 1: is today, because the lead time to design and build 350 00:21:48,440 --> 00:21:50,840 Speaker 1: a virus is, you know, takes a little bit of time. 351 00:21:51,320 --> 00:21:53,680 Speaker 1: You can't wait for the virus to hit. The same 352 00:21:53,760 --> 00:21:56,359 Speaker 1: kind of logic will be applied to the investigation of 353 00:21:56,480 --> 00:22:00,359 Speaker 1: COVID nineteen UM. You know, depending on us on a 354 00:22:00,400 --> 00:22:04,240 Speaker 1: discovery of science in terms of the extent and how 355 00:22:04,440 --> 00:22:06,320 Speaker 1: it will evolve over the course of time. But the 356 00:22:06,359 --> 00:22:10,120 Speaker 1: expectation is it will evolve, and so you'll use computational 357 00:22:10,200 --> 00:22:15,320 Speaker 1: techniques to begin to fathom that infinite possible ways in 358 00:22:15,359 --> 00:22:17,560 Speaker 1: which it could evolve and choose those that are most 359 00:22:17,600 --> 00:22:20,679 Speaker 1: likely to represent where the virus will be in a 360 00:22:20,680 --> 00:22:23,760 Speaker 1: handful of months, and you'll use that to inform the 361 00:22:23,800 --> 00:22:26,680 Speaker 1: way you design your vaccine to intercept it. So we're 362 00:22:26,720 --> 00:22:31,480 Speaker 1: talking about forming probabilistic models to really determine where are 363 00:22:31,760 --> 00:22:36,480 Speaker 1: the most likely pathways that this virus might take evolutionary 364 00:22:36,480 --> 00:22:40,120 Speaker 1: a lee speaking, it's very similar to how I from 365 00:22:40,160 --> 00:22:43,520 Speaker 1: a concept level, It's very similar to how I would 366 00:22:43,560 --> 00:22:48,600 Speaker 1: look at something like IBM Watson when you know everyone 367 00:22:48,760 --> 00:22:53,440 Speaker 1: knows about it competing on Jeopardy. It had probabilistic approaches 368 00:22:53,480 --> 00:22:56,320 Speaker 1: to which answers would be the most accurate, and only 369 00:22:56,359 --> 00:22:59,439 Speaker 1: if it reached a certain threshold of certainty would it 370 00:22:59,680 --> 00:23:02,600 Speaker 1: would buzz in. But of course, obviously now we're talking 371 00:23:02,600 --> 00:23:06,560 Speaker 1: about a much more complex thing and much higher stakes, 372 00:23:06,600 --> 00:23:09,480 Speaker 1: but it's that same sort of approach of where can 373 00:23:09,560 --> 00:23:12,720 Speaker 1: we predict where this is going, how can we get 374 00:23:12,760 --> 00:23:15,159 Speaker 1: ahead of it? Then how can we create you know, 375 00:23:15,200 --> 00:23:18,160 Speaker 1: a dead version or an inert version I should say, 376 00:23:18,200 --> 00:23:21,120 Speaker 1: of the virus to make a vaccine. And then you 377 00:23:21,160 --> 00:23:24,800 Speaker 1: have the other challenges that come in vaccinations, which is 378 00:23:24,840 --> 00:23:29,320 Speaker 1: just you know, the manufacturing process, distribution, that sort of thing. 379 00:23:29,359 --> 00:23:33,879 Speaker 1: But this shortening the pathway to this part to me 380 00:23:34,000 --> 00:23:38,440 Speaker 1: seems like it is Uh, it is absolutely crucial, and 381 00:23:38,800 --> 00:23:40,560 Speaker 1: it's also one of the areas where I would think 382 00:23:40,600 --> 00:23:44,200 Speaker 1: that you would see the longest delay. So seeing the 383 00:23:44,200 --> 00:23:48,400 Speaker 1: the application of supercomputers is really inspiring to me. Are 384 00:23:48,400 --> 00:23:53,800 Speaker 1: there other ways that IBM is contributing to various efforts 385 00:23:53,840 --> 00:23:58,920 Speaker 1: to either track or fight COVID nineteen. Yes, And in fact, 386 00:23:59,040 --> 00:24:03,520 Speaker 1: just last Friday, a IBM released for free on our 387 00:24:03,560 --> 00:24:10,919 Speaker 1: website a UM, an artificial intelligence package that speculatively designs 388 00:24:10,960 --> 00:24:15,360 Speaker 1: new molecules for the treatment of COVID nineteen. So let 389 00:24:15,400 --> 00:24:17,240 Speaker 1: me back up for a second. If you think about 390 00:24:17,280 --> 00:24:19,920 Speaker 1: what's been going on at oak Ridge with Jeremy Smith's 391 00:24:19,920 --> 00:24:23,399 Speaker 1: effort to look at eight thousand compounds and whittle that 392 00:24:23,520 --> 00:24:27,960 Speaker 1: down to seventy seven for further investigation as potential therapies 393 00:24:28,000 --> 00:24:32,080 Speaker 1: to treat COVID nineteen. Well, those eight thousand existed, somebody 394 00:24:32,080 --> 00:24:35,359 Speaker 1: had already built them. Question is, are there new kinds 395 00:24:35,400 --> 00:24:39,280 Speaker 1: of molecules that could be designed that don't exist today 396 00:24:39,320 --> 00:24:42,240 Speaker 1: that could be used to treat COVID nineteen. So the 397 00:24:42,320 --> 00:24:45,520 Speaker 1: artificial intelligence package that IBM put out on Friday lets 398 00:24:45,560 --> 00:24:48,199 Speaker 1: you do that, and it's free and it's open to anyone, 399 00:24:48,280 --> 00:24:51,719 Speaker 1: So anybody can get on the website and begin playing 400 00:24:51,720 --> 00:24:54,320 Speaker 1: with it, and maybe you kick up a new molecule 401 00:24:54,760 --> 00:24:57,280 Speaker 1: which ends up his input to the next generation of 402 00:24:57,320 --> 00:25:00,440 Speaker 1: the work that goes on within the COVID high performance 403 00:25:00,440 --> 00:25:05,679 Speaker 1: computing consortion. So there's innovation at both ends of the process, 404 00:25:05,480 --> 00:25:09,120 Speaker 1: the design and designation new molecules and then of course 405 00:25:09,160 --> 00:25:13,720 Speaker 1: the assessment of existing molecules, including the newly designed or 406 00:25:13,760 --> 00:25:19,000 Speaker 1: invented ones, to assess efficacy against against the COVID nineteen 407 00:25:19,080 --> 00:25:23,040 Speaker 1: virus um. And these ideas need to work in concert, 408 00:25:23,200 --> 00:25:27,399 Speaker 1: and they will. Science is all about us discovering the 409 00:25:27,480 --> 00:25:31,240 Speaker 1: rules of the universe. That sounds grandiose, but it is true. 410 00:25:31,280 --> 00:25:35,239 Speaker 1: The rules exist with or without us. Science is our 411 00:25:35,320 --> 00:25:38,840 Speaker 1: process for figuring out what those rules are and sometimes 412 00:25:38,920 --> 00:25:41,479 Speaker 1: leads to us learning how to take advantage of those rules, 413 00:25:41,640 --> 00:25:44,480 Speaker 1: or to avoid things that might cause us harm or 414 00:25:44,560 --> 00:25:47,800 Speaker 1: pushing back the boundaries of what we see as our limitations. 415 00:25:48,520 --> 00:25:52,919 Speaker 1: Understanding those rules, we can build complicated virtual environments that 416 00:25:53,000 --> 00:25:56,640 Speaker 1: let us play with creating new molecules. The rules are 417 00:25:56,680 --> 00:26:00,440 Speaker 1: the foundation of these virtual environments. The rules include which 418 00:26:00,480 --> 00:26:05,080 Speaker 1: atoms can bond with which other atoms, and under what circumstances. 419 00:26:05,119 --> 00:26:08,240 Speaker 1: So we start off with what is physically possible based 420 00:26:08,280 --> 00:26:13,240 Speaker 1: on how we understand chemistry. Molecules that could exist can 421 00:26:13,280 --> 00:26:17,199 Speaker 1: be fair game. Molecules that cannot exist are a no 422 00:26:17,359 --> 00:26:20,400 Speaker 1: go because it doesn't really help the end cause if 423 00:26:20,400 --> 00:26:24,920 Speaker 1: the solution you propose is physically impossible. After all, as 424 00:26:25,000 --> 00:26:28,960 Speaker 1: Dave mentioned, IBM opened up this artificial intelligence tool to 425 00:26:29,119 --> 00:26:33,720 Speaker 1: anyone who wants to work with it, so chemists, doctors, researchers, 426 00:26:33,720 --> 00:26:36,560 Speaker 1: and others can contribute to the efforts to do so. 427 00:26:36,920 --> 00:26:41,680 Speaker 1: You can visit the website. Here's the address www dot 428 00:26:41,760 --> 00:26:48,399 Speaker 1: research dot IBM dot com, slash COVID nineteen slash deep 429 00:26:48,560 --> 00:26:56,400 Speaker 1: dash search. I'm also curious about other applications of the supercomputers. Obviously, 430 00:26:56,520 --> 00:26:59,960 Speaker 1: right now we're very much focused on the COVID nineteen crisis, 431 00:27:00,000 --> 00:27:02,480 Speaker 1: as we should be. But once we're through this crisis, 432 00:27:02,520 --> 00:27:06,160 Speaker 1: it's not like the work stops for high performance computing. 433 00:27:06,200 --> 00:27:09,399 Speaker 1: There's so many different applications. Can you talk about some 434 00:27:09,480 --> 00:27:13,760 Speaker 1: of the other purposes that scientists and researchers are putting 435 00:27:14,000 --> 00:27:17,840 Speaker 1: these remarkable machines to, Oh? Absolutely, and and I would 436 00:27:17,880 --> 00:27:22,560 Speaker 1: say the first thing is that you cannot. No person 437 00:27:22,600 --> 00:27:25,760 Speaker 1: on the planet can go through a day without touching 438 00:27:25,960 --> 00:27:29,840 Speaker 1: a product of service or something that's not been impacted 439 00:27:29,880 --> 00:27:33,920 Speaker 1: by the application is supercomputing. Somewhere in the world. They 440 00:27:34,040 --> 00:27:40,080 Speaker 1: used to design automobiles for aerodynamics and fuel efficiency. They 441 00:27:40,200 --> 00:27:43,280 Speaker 1: used to design the kinds of batteries that the electric 442 00:27:43,320 --> 00:27:46,480 Speaker 1: car companies are putting in their cars. They used to 443 00:27:46,520 --> 00:27:50,480 Speaker 1: design air foils and airplanes. Uh. New drugs that we've 444 00:27:50,480 --> 00:27:54,160 Speaker 1: talked about, they they're used for fraud detection. So when 445 00:27:54,200 --> 00:27:57,440 Speaker 1: you get a call on your telephone where your credit 446 00:27:57,480 --> 00:28:01,640 Speaker 1: card company says, by the way, we've signaled potential misuse 447 00:28:01,640 --> 00:28:05,880 Speaker 1: of your car card, that's probably been done by a supercomputer, 448 00:28:06,000 --> 00:28:08,520 Speaker 1: not smaller than the kinds that we're talking about here 449 00:28:08,560 --> 00:28:10,720 Speaker 1: at a place like OK read your Lawrence Livermore or 450 00:28:10,760 --> 00:28:14,119 Speaker 1: are gone, but the same sort of family, this notion 451 00:28:14,160 --> 00:28:20,600 Speaker 1: of parallel computing, floating point analysis, and corporation of AI, etcetera. UM, 452 00:28:20,640 --> 00:28:24,639 Speaker 1: so it's it's extraordinarily widespread. I think. One of the 453 00:28:25,000 --> 00:28:29,680 Speaker 1: really tremendously promising areas for supercomputing, and by the way, 454 00:28:29,720 --> 00:28:32,359 Speaker 1: people have been poking at this for for for quite 455 00:28:32,400 --> 00:28:36,800 Speaker 1: some time, is the area materials science. Materials are used 456 00:28:37,000 --> 00:28:41,680 Speaker 1: in everything. That's sort of um a not very profound 457 00:28:41,720 --> 00:28:46,040 Speaker 1: statement to make, but but the nature materials are quite exotic. 458 00:28:46,720 --> 00:28:50,160 Speaker 1: And when you look, for example, at a designing new 459 00:28:50,160 --> 00:28:54,080 Speaker 1: batteries lithium ion batteries and and so on, and you say, well, 460 00:28:54,080 --> 00:28:56,680 Speaker 1: how do I get more efficiency out of batteries to 461 00:28:56,880 --> 00:29:01,160 Speaker 1: drive electric cars? Well, that's when materials start to come 462 00:29:01,200 --> 00:29:05,080 Speaker 1: into play. Where you you start building battery elements out 463 00:29:05,120 --> 00:29:08,640 Speaker 1: of new combination of alloys that no one previously explored 464 00:29:08,720 --> 00:29:11,680 Speaker 1: or anticipated, especially in the context of the use to 465 00:29:11,720 --> 00:29:15,560 Speaker 1: which the battery will will apply them. So the opportunity 466 00:29:15,560 --> 00:29:19,840 Speaker 1: to explore worlds that don't really exist yet digitally without 467 00:29:19,880 --> 00:29:22,360 Speaker 1: having to incur the expense of creating them in the 468 00:29:22,400 --> 00:29:25,320 Speaker 1: analog sense. You know, you're not building laboratories and things 469 00:29:25,360 --> 00:29:28,200 Speaker 1: like that. And in some cases of course, the nature 470 00:29:28,200 --> 00:29:31,240 Speaker 1: of what you're doing might even be viewed as dangerous. Uh. 471 00:29:31,280 --> 00:29:35,400 Speaker 1: The opportunity to use supercomputing to explore those worlds, explore 472 00:29:35,440 --> 00:29:40,240 Speaker 1: those opportunities, do it safely, do it cost effectively, becomes 473 00:29:40,240 --> 00:29:43,640 Speaker 1: a tremendous boon to the scientific method generally, and I 474 00:29:43,640 --> 00:29:45,960 Speaker 1: would say for the last twenty five years or so, 475 00:29:47,040 --> 00:29:52,600 Speaker 1: when scientists talk about the scientific method, you know, hypothesis, experimentation, 476 00:29:53,240 --> 00:29:56,720 Speaker 1: data and all those things. I think computation is now 477 00:29:56,840 --> 00:29:59,640 Speaker 1: factored in is a key element to the whole scientific 478 00:29:59,680 --> 00:30:04,000 Speaker 1: process us its ability to see things, to explore things 479 00:30:04,040 --> 00:30:07,880 Speaker 1: that you cannot get to with other kinds of scientific 480 00:30:07,960 --> 00:30:12,800 Speaker 1: instruments and tools. Yeah, I I I have been covering 481 00:30:12,880 --> 00:30:15,800 Speaker 1: technology for several years now, and I've talked a lot 482 00:30:15,840 --> 00:30:19,760 Speaker 1: about some of the early scientists physicists who who kind 483 00:30:19,760 --> 00:30:22,600 Speaker 1: of laid the groundwork for the technologies we depend upon today. 484 00:30:22,640 --> 00:30:25,480 Speaker 1: You know, they learned about the science that the technology 485 00:30:25,600 --> 00:30:28,760 Speaker 1: is a physical implementation of and allows us to take 486 00:30:28,760 --> 00:30:31,920 Speaker 1: advantage of that science. And in many cases you're talking 487 00:30:31,920 --> 00:30:35,560 Speaker 1: about people who came across something by accident, you know, 488 00:30:35,560 --> 00:30:38,160 Speaker 1: it was just fortuitous that they observed something and that 489 00:30:38,280 --> 00:30:40,400 Speaker 1: someone else was able to figure out how to make 490 00:30:40,520 --> 00:30:43,800 Speaker 1: use of that. So having a way to virtualize that 491 00:30:44,200 --> 00:30:47,920 Speaker 1: and speed up that process exponentially, to me, what that 492 00:30:48,000 --> 00:30:51,160 Speaker 1: tells me is that we get a chance to enjoy 493 00:30:51,240 --> 00:30:54,680 Speaker 1: the benefits of that science on a time scale that 494 00:30:55,080 --> 00:30:58,760 Speaker 1: is would previously have been impossible. You might have been 495 00:30:58,760 --> 00:31:01,800 Speaker 1: talking about something where, you know what, maybe that discovery 496 00:31:01,840 --> 00:31:05,200 Speaker 1: could be something that impacts my great grandchild. But now 497 00:31:05,200 --> 00:31:09,120 Speaker 1: we're talking about things that could potentially have a physical 498 00:31:09,160 --> 00:31:13,880 Speaker 1: implementation within a decade or less in some cases, and 499 00:31:13,880 --> 00:31:17,760 Speaker 1: in cases where we're actively researching a vaccine much much 500 00:31:17,920 --> 00:31:21,960 Speaker 1: closer to now, which is again incredible. When we're talking 501 00:31:21,960 --> 00:31:24,280 Speaker 1: about computational power, it's not just the speed at which 502 00:31:24,280 --> 00:31:26,920 Speaker 1: we're solving problems, it's the speed at which we're able 503 00:31:26,960 --> 00:31:32,040 Speaker 1: to take advantage of those solutions. So I'm really, well, 504 00:31:32,080 --> 00:31:34,160 Speaker 1: you can tell I'm really jazzed about this conversation. I 505 00:31:34,200 --> 00:31:37,160 Speaker 1: get excited about the weirdest things. This isn't weird, this 506 00:31:37,240 --> 00:31:39,960 Speaker 1: is commonplace. And and you know, one of the things 507 00:31:40,000 --> 00:31:43,040 Speaker 1: that's coming as a result of this is is there's 508 00:31:43,080 --> 00:31:48,280 Speaker 1: a real explosion in growth of knowledge. So, uh, I'll 509 00:31:48,320 --> 00:31:50,840 Speaker 1: go back to material science. If you look at material 510 00:31:50,960 --> 00:31:55,800 Speaker 1: science ten or fifteen years ago, and you said, well, 511 00:31:55,920 --> 00:32:00,000 Speaker 1: how many papers scientific papers are published annually in material 512 00:32:00,000 --> 00:32:03,240 Speaker 1: of science? Um? And I said, guess with that number 513 00:32:03,280 --> 00:32:05,520 Speaker 1: is what would you guess that number to be. Um, 514 00:32:06,240 --> 00:32:09,640 Speaker 1: I'm gonna go with. No matter what number I say, 515 00:32:09,640 --> 00:32:13,520 Speaker 1: it's gonna be wrong. I'm gonna say so that's actually 516 00:32:13,560 --> 00:32:16,720 Speaker 1: an ambitious guess. So the ten or fifteen years ago 517 00:32:16,760 --> 00:32:19,840 Speaker 1: the number was ten thousand, but last year the number 518 00:32:19,920 --> 00:32:26,480 Speaker 1: was five. Now, now the point is knowledge is growing 519 00:32:26,520 --> 00:32:29,320 Speaker 1: at a rate faster than humans are able to consume 520 00:32:29,360 --> 00:32:33,800 Speaker 1: the knowledge. Because now these are referee papers, so they're 521 00:32:33,800 --> 00:32:36,520 Speaker 1: all serious and people looked at them and you know, 522 00:32:36,600 --> 00:32:39,400 Speaker 1: it's it's contributed to the to the corpus of knowledge 523 00:32:40,000 --> 00:32:43,880 Speaker 1: that that are accessible to human that everybody agrees is true. 524 00:32:44,520 --> 00:32:47,720 Speaker 1: And you think about five thousand papers. That's five thousand 525 00:32:47,800 --> 00:32:52,600 Speaker 1: papers a year. So how do you becount maintain currency 526 00:32:52,600 --> 00:32:55,160 Speaker 1: in the field. And the answer is you can't. So 527 00:32:55,240 --> 00:32:58,720 Speaker 1: now you look at the application is super computing to 528 00:32:58,840 --> 00:33:04,160 Speaker 1: help you grasp, contain, and really model the knowledge that's 529 00:33:04,200 --> 00:33:08,240 Speaker 1: available as as well as generate new knowledge. So these 530 00:33:08,280 --> 00:33:12,640 Speaker 1: ideas embodied him things like Watson couples to supercomputing. Let 531 00:33:12,680 --> 00:33:16,640 Speaker 1: you begin to explore this vast array of scientific knowledge 532 00:33:16,720 --> 00:33:20,600 Speaker 1: in a very coordinated and orchestrated kind of fashion to 533 00:33:20,680 --> 00:33:23,160 Speaker 1: gain insight that you have no way of getting as 534 00:33:24,240 --> 00:33:27,720 Speaker 1: as the conventional way that you know people looked at 535 00:33:27,640 --> 00:33:30,360 Speaker 1: acquiring knowledge fifteen or twenty years ago. You don't go 536 00:33:30,400 --> 00:33:34,000 Speaker 1: to the library read five thousand papers, right, But on 537 00:33:34,040 --> 00:33:37,840 Speaker 1: the other hand, you can use systems equipped with UH infrastructure, 538 00:33:37,880 --> 00:33:40,680 Speaker 1: based on tools like Watson, and you can begin to 539 00:33:40,760 --> 00:33:43,800 Speaker 1: fathom those five thousand papers in the blink of an 540 00:33:43,800 --> 00:33:47,320 Speaker 1: eye and get an understanding of relationships that would have 541 00:33:47,320 --> 00:33:50,920 Speaker 1: never occurred to you naturally, and and to begin to 542 00:33:51,000 --> 00:33:56,440 Speaker 1: give you ideas of new directions to pursue. So my reference, 543 00:33:56,480 --> 00:34:02,400 Speaker 1: for example, to the presentation on Friday of that UM 544 00:34:02,560 --> 00:34:07,560 Speaker 1: software package from IBM using AI to speculatively help you 545 00:34:07,960 --> 00:34:11,880 Speaker 1: look at new molecules for COVID nineteen are based on 546 00:34:11,960 --> 00:34:16,759 Speaker 1: principles like these, harvesting human knowledge at scale that a 547 00:34:16,880 --> 00:34:20,000 Speaker 1: human can't handle and coming up with novel kinds of 548 00:34:20,040 --> 00:34:24,000 Speaker 1: interpretations of the knowledge that gives rise to potentially radically 549 00:34:24,040 --> 00:34:28,680 Speaker 1: new and terrifically important innovations. So this is something that 550 00:34:28,719 --> 00:34:31,719 Speaker 1: people really I don't think you've digested fully yet in 551 00:34:31,840 --> 00:34:35,040 Speaker 1: terms of supercomputing, which they've always fewed as a means 552 00:34:35,080 --> 00:34:40,359 Speaker 1: by which you do the standard scientific calculations faster. Now 553 00:34:40,360 --> 00:34:45,000 Speaker 1: we're looking at this coalescence of approach that spans knowledge 554 00:34:45,040 --> 00:34:47,960 Speaker 1: and data and computation and looking at it all together 555 00:34:48,480 --> 00:34:52,000 Speaker 1: to give rise to insights that previously could never have 556 00:34:52,120 --> 00:34:56,239 Speaker 1: been imagined. Yeah, it's it's been great to see the 557 00:34:56,320 --> 00:34:59,080 Speaker 1: journey of where we were going from a point where 558 00:34:59,080 --> 00:35:02,120 Speaker 1: we were gathering enormous amounts of data, you know, the 559 00:35:02,160 --> 00:35:05,640 Speaker 1: early era of big data, getting a better understanding of 560 00:35:05,680 --> 00:35:10,040 Speaker 1: how to manage and analyze that data to contextualize it. 561 00:35:10,120 --> 00:35:12,120 Speaker 1: And now we're reaching a point or we're at a 562 00:35:12,160 --> 00:35:15,640 Speaker 1: point where we have these incredible systems that are capable 563 00:35:15,800 --> 00:35:19,719 Speaker 1: of of doing that on a human level. If that 564 00:35:19,800 --> 00:35:23,400 Speaker 1: human level, we're you know, every human on the planet 565 00:35:23,440 --> 00:35:26,040 Speaker 1: able to think about this stuff simultaneously and share that 566 00:35:26,080 --> 00:35:29,120 Speaker 1: information in a hive mind. So to me, again, this 567 00:35:29,160 --> 00:35:33,440 Speaker 1: is super exciting stuff and uh, I'm really I'm really 568 00:35:33,440 --> 00:35:37,359 Speaker 1: optimistic about this. I think that this is uh an 569 00:35:37,360 --> 00:35:41,800 Speaker 1: approach that is going to lead to some really actionable solutions, 570 00:35:41,840 --> 00:35:45,960 Speaker 1: and ultimately what that tells me is that you know, 571 00:35:46,000 --> 00:35:47,880 Speaker 1: we can talk about the tech and it's super cool, 572 00:35:48,040 --> 00:35:51,640 Speaker 1: and how advanced it is, and how how complex it 573 00:35:51,719 --> 00:35:53,960 Speaker 1: is and the sort of problems it can it can 574 00:35:54,000 --> 00:35:56,719 Speaker 1: tackle from a very conceptual level. But to me, the 575 00:35:56,719 --> 00:36:00,920 Speaker 1: really inspiring thing is seeing the actual impact on the 576 00:36:00,960 --> 00:36:04,759 Speaker 1: world when we see these solutions enacted in ways that 577 00:36:04,880 --> 00:36:09,160 Speaker 1: make a direct improvement in people's lives. To me, there's 578 00:36:09,239 --> 00:36:13,680 Speaker 1: no greater story of the potential and power of technology 579 00:36:13,719 --> 00:36:18,680 Speaker 1: than that, I would agree, and I think that we're 580 00:36:18,680 --> 00:36:22,800 Speaker 1: a stage now where the application of this technology is 581 00:36:22,840 --> 00:36:26,759 Speaker 1: becoming progressively more and more ubiquitous and accessible. And by 582 00:36:26,800 --> 00:36:31,160 Speaker 1: accessible I mean with the advent of artificial intelligence over 583 00:36:31,160 --> 00:36:34,680 Speaker 1: the last few years. From a commercial perspective. It's not 584 00:36:34,840 --> 00:36:38,640 Speaker 1: accessible to normal humans, right. You don't have to have 585 00:36:39,160 --> 00:36:45,239 Speaker 1: exotic experience in UH in computer science or exotic experience 586 00:36:45,280 --> 00:36:48,319 Speaker 1: in mathematics. You can go on a system like the 587 00:36:48,360 --> 00:36:53,279 Speaker 1: IBM Molecular Forecasting system, and with a little bit of 588 00:36:53,320 --> 00:36:57,560 Speaker 1: knowledge of chemistry, not computers, but chemistry, you can begin 589 00:36:57,719 --> 00:37:03,839 Speaker 1: to explore possibilities that would have been previously inaccessible to you. 590 00:37:04,400 --> 00:37:08,960 Speaker 1: So it's a democratization of supercomputing that's happening as well 591 00:37:09,440 --> 00:37:14,360 Speaker 1: as as these AI methodologies are incorporated and now dramatically 592 00:37:14,400 --> 00:37:19,399 Speaker 1: expands the utility of the technology by virtue of making accessible. 593 00:37:19,480 --> 00:37:23,920 Speaker 1: Thomas everyone fantastic. And this is this is a thread 594 00:37:23,960 --> 00:37:26,439 Speaker 1: that when I've spoken with people at IBM, it has 595 00:37:26,480 --> 00:37:29,120 Speaker 1: come up at time and time again. This not just 596 00:37:29,360 --> 00:37:32,719 Speaker 1: the development of technology and not just the implementation of it, 597 00:37:32,800 --> 00:37:36,600 Speaker 1: but the as you say, the democratization, the making it 598 00:37:36,640 --> 00:37:40,360 Speaker 1: available for people, whether it's call for code where coders 599 00:37:40,440 --> 00:37:43,920 Speaker 1: are building solutions to big problems and they're getting support 600 00:37:44,040 --> 00:37:47,880 Speaker 1: through access to IBM tools, or something along these lines, 601 00:37:48,200 --> 00:37:51,680 Speaker 1: or we talk about not that we should talk about 602 00:37:51,719 --> 00:37:54,480 Speaker 1: this because I'll go down a rabbit hole, but IBM 603 00:37:54,560 --> 00:37:57,560 Speaker 1: developing quantum computers and opening that up for people to 604 00:37:58,120 --> 00:38:01,120 Speaker 1: develop for that so that they can test that out 605 00:38:01,200 --> 00:38:06,560 Speaker 1: sort of the next generation of truly remarkable parallel processing. 606 00:38:06,680 --> 00:38:08,200 Speaker 1: If you want to talk about that, you go down 607 00:38:08,280 --> 00:38:11,120 Speaker 1: that quantum road. And to me, that's one of those 608 00:38:11,120 --> 00:38:15,239 Speaker 1: really defining features that makes me happy to have these 609 00:38:15,320 --> 00:38:18,960 Speaker 1: kind of conversations because I know that my listeners, if 610 00:38:19,000 --> 00:38:21,719 Speaker 1: they want to, they can actually go out and take 611 00:38:21,760 --> 00:38:24,759 Speaker 1: advantage of these tools themselves. They just have to take 612 00:38:24,800 --> 00:38:28,319 Speaker 1: the step to learn and to go and be part 613 00:38:28,320 --> 00:38:33,120 Speaker 1: of it. And it's not just a a a supercomputer 614 00:38:33,280 --> 00:38:36,600 Speaker 1: that's locked away in a lab or deep underground or 615 00:38:36,640 --> 00:38:40,560 Speaker 1: some sort of Douglas Adams Hitchhiker's Guide deep thought computer. 616 00:38:40,920 --> 00:38:44,840 Speaker 1: It's something that's actually accessible to people. You just have 617 00:38:44,960 --> 00:38:48,520 Speaker 1: to take some pretty simple steps to do it. That's right, 618 00:38:48,680 --> 00:38:52,640 Speaker 1: And our strategies to make more and more of these 619 00:38:52,719 --> 00:38:56,600 Speaker 1: innovative technologies available on the web and free to people 620 00:38:56,719 --> 00:38:59,920 Speaker 1: so that they can play with it. But by virtual 621 00:39:00,040 --> 00:39:02,480 Speaker 1: playing with it infem us about the directions some of 622 00:39:02,480 --> 00:39:05,839 Speaker 1: our innovations should take as well. UM. We've done this 623 00:39:05,920 --> 00:39:09,759 Speaker 1: in chemistry, We've done some biology, we've done it in quantum. 624 00:39:09,800 --> 00:39:13,200 Speaker 1: I think it's um it's a very successful paradigm to 625 00:39:13,400 --> 00:39:17,040 Speaker 1: produce things that are really useful compared to the old 626 00:39:17,080 --> 00:39:20,920 Speaker 1: style way of um, you know, doing it locked away 627 00:39:20,960 --> 00:39:23,560 Speaker 1: in a tower someplace, and then just revealing your innovation 628 00:39:23,640 --> 00:39:26,440 Speaker 1: to the world, hoping for the best. Better to have 629 00:39:26,520 --> 00:39:29,239 Speaker 1: the world along from the very beginning. Yeah, I think 630 00:39:29,280 --> 00:39:33,320 Speaker 1: silos are best left on farms. I also agree with that. Dave. 631 00:39:33,480 --> 00:39:37,120 Speaker 1: Thank you so much for your time and your expertise. 632 00:39:37,560 --> 00:39:39,759 Speaker 1: I wish you and your team all the best as 633 00:39:39,800 --> 00:39:43,719 Speaker 1: you continue to put high performance computing two uses that 634 00:39:43,920 --> 00:39:46,879 Speaker 1: I'm sure I can't even imagine right now. I can't 635 00:39:46,880 --> 00:39:50,040 Speaker 1: wait to see what's next. Me too, and you'll be 636 00:39:50,040 --> 00:39:53,319 Speaker 1: seeing things coming out of the consortium very quickly. For 637 00:39:53,360 --> 00:39:57,040 Speaker 1: people who are following it UM, please go to the 638 00:39:57,080 --> 00:40:02,400 Speaker 1: website COVID nineteen HPC conser Marsha and beginning next week 639 00:40:02,440 --> 00:40:05,719 Speaker 1: we'll start to publish the science it's actually being done 640 00:40:05,760 --> 00:40:09,640 Speaker 1: on the computers. It is encouraging to see IBM take 641 00:40:09,719 --> 00:40:14,200 Speaker 1: an open, inclusive approach towards technological solutions. The company has 642 00:40:14,239 --> 00:40:18,319 Speaker 1: produced lots of complex technologies that have enormous power, but 643 00:40:18,360 --> 00:40:22,279 Speaker 1: IBM also recognizes that innovation and solutions can come from 644 00:40:22,360 --> 00:40:26,560 Speaker 1: any direction, and making these resources easily available speeds up 645 00:40:26,600 --> 00:40:30,960 Speaker 1: the process of arriving at those solutions. In this particular instance, 646 00:40:31,040 --> 00:40:34,360 Speaker 1: we're talking about a dangerous virus and the disease it causes, 647 00:40:34,719 --> 00:40:39,040 Speaker 1: but the underlying philosophy of inclusion extends beyond that. It 648 00:40:39,080 --> 00:40:41,239 Speaker 1: was a pleasure to speak with Dave Turik about high 649 00:40:41,239 --> 00:40:43,880 Speaker 1: performance computing and its role in the response to the 650 00:40:43,920 --> 00:40:48,439 Speaker 1: COVID nineteen crisis. I have no doubt that the complicated 651 00:40:48,560 --> 00:40:51,960 Speaker 1: simulations will allow for much more rapid development, which in 652 00:40:52,000 --> 00:40:55,520 Speaker 1: turn will mean a faster path to effective treatments for 653 00:40:55,640 --> 00:40:59,480 Speaker 1: COVID nineteen. That's something I won't lose sight of. As 654 00:40:59,520 --> 00:41:02,440 Speaker 1: I said to Dave, this technology is really cool, but 655 00:41:02,640 --> 00:41:04,680 Speaker 1: not as cool as the results will see from that 656 00:41:04,760 --> 00:41:09,919 Speaker 1: text application. That's all for today's episode. Before I sign off, 657 00:41:09,960 --> 00:41:12,200 Speaker 1: I want to remind you guys of the Call for 658 00:41:12,320 --> 00:41:16,680 Speaker 1: Code Global Challenge. This is the big coding slash hacking 659 00:41:16,840 --> 00:41:20,040 Speaker 1: challenge IBM sponsors every year. It always takes aim at 660 00:41:20,080 --> 00:41:23,920 Speaker 1: a really big problem and it invites people to submit 661 00:41:24,280 --> 00:41:28,400 Speaker 1: ideas for applications that could address these problems in some way, 662 00:41:28,640 --> 00:41:32,840 Speaker 1: and those applications can tap into the incredible resources of IBM, 663 00:41:32,880 --> 00:41:38,040 Speaker 1: including amazing IBM technologies. This year, there are two tracks 664 00:41:38,400 --> 00:41:42,120 Speaker 1: for the Global Challenge. The first of the two tracks 665 00:41:42,320 --> 00:41:46,680 Speaker 1: specifically focuses on COVID nineteen. If you have an idea 666 00:41:46,880 --> 00:41:50,680 Speaker 1: for an application that could help address the crisis, then 667 00:41:50,719 --> 00:41:54,839 Speaker 1: you need to submit it by April for consideration. By 668 00:41:54,840 --> 00:41:58,520 Speaker 1: May five, they will pick the top three COVID nineteen solutions, 669 00:41:59,000 --> 00:42:02,120 Speaker 1: and then by mayfie teen they start initial deployment of 670 00:42:02,120 --> 00:42:04,960 Speaker 1: those solutions. If you want to submit for the broader 671 00:42:05,000 --> 00:42:09,359 Speaker 1: topic of climate change, then IBM is accepting those applications 672 00:42:09,440 --> 00:42:13,640 Speaker 1: until July one. Now, to be clear, they will be 673 00:42:13,680 --> 00:42:18,560 Speaker 1: accepting COVID nineteen solutions throughout the entirety of the Global Challenge, 674 00:42:18,920 --> 00:42:22,040 Speaker 1: but as I said, the timeline for a consideration for 675 00:42:22,120 --> 00:42:27,640 Speaker 1: those three spots has to be submitted by April. In October, 676 00:42:27,880 --> 00:42:30,920 Speaker 1: the winners of the Call for Code Global Challenge will 677 00:42:30,920 --> 00:42:35,600 Speaker 1: be announced at an award ceremony. So if you have ideas, 678 00:42:35,680 --> 00:42:38,600 Speaker 1: if you're looking for like minded people to work on 679 00:42:38,760 --> 00:42:42,480 Speaker 1: real world solutions that can really change things for people, 680 00:42:43,080 --> 00:42:47,120 Speaker 1: I highly recommend you look at the Call for Code Challenge. 681 00:42:47,280 --> 00:42:50,560 Speaker 1: You can find out more at IBM dot b I 682 00:42:50,680 --> 00:42:55,000 Speaker 1: Z slash Call for Code. In the next Smart Talks 683 00:42:55,040 --> 00:42:58,600 Speaker 1: on tech Stuff, I'll sit down with Grace Sue, VP 684 00:42:58,760 --> 00:43:02,200 Speaker 1: of Education at I b M and Kristen Waznowski, ce 685 00:43:02,239 --> 00:43:05,040 Speaker 1: IO of Design at IBM to talk about how the 686 00:43:05,040 --> 00:43:09,360 Speaker 1: company's technologies are powering remote learning and remote work efforts. 687 00:43:09,840 --> 00:43:18,000 Speaker 1: I'll talk to you again really soon. Text Stuff is 688 00:43:18,000 --> 00:43:21,200 Speaker 1: an I Heart Radio production. For more podcasts from I 689 00:43:21,280 --> 00:43:24,879 Speaker 1: Heart Radio, visit the I Heart Radio app, Apple Podcasts, 690 00:43:25,000 --> 00:43:27,000 Speaker 1: or wherever you listen to your favorite shows