1 00:00:08,675 --> 00:00:09,542 Hello everyone. 2 00:00:09,542 --> 00:00:11,611 Welcome to the next podcast 3 00:00:11,611 --> 00:00:14,881 From Research to Reality, I have a great honor 4 00:00:14,881 --> 00:00:17,951 and pleasure to host Jason Zeiler. 5 00:00:18,184 --> 00:00:20,253 Hello, Jason, Great to meet you. 6 00:00:20,253 --> 00:00:25,225 Jason is the product manager of Liquid Cooling and Next Generation 7 00:00:25,525 --> 00:00:27,093 Infrastructure. 8 00:00:27,093 --> 00:00:31,398 And this is the first time we're bringing in someone who is product manager. 9 00:00:31,398 --> 00:00:35,168 We've done largely the first part from research. 10 00:00:35,168 --> 00:00:38,038 Now we're getting closer to reality. Excellent. 11 00:00:38,038 --> 00:00:39,906 You need the product in there to where it kind of 12 00:00:39,906 --> 00:00:41,875 gets in the customer's hands is always important. 13 00:00:41,875 --> 00:00:42,842 Yeah, exactly. 14 00:00:42,842 --> 00:00:45,278 So tell us a little bit what that title means. 15 00:00:45,278 --> 00:00:46,079 Sure. 16 00:00:46,079 --> 00:00:48,948 So I kind of work with an HPC group. 17 00:00:48,948 --> 00:00:53,353 Liquid cooling is really an integral part of our infrastructure and planning. 18 00:00:53,620 --> 00:00:57,991 How we're enabling you know, high TDP stuff, new GPUs and CPUs. 19 00:00:58,258 --> 00:01:02,162 So we need a product manager. TDP? Thermal Design Power. 20 00:01:02,162 --> 00:01:05,131 How much power goes into chips and how hot they're going to get. 21 00:01:05,131 --> 00:01:08,868 And the next generation is really what the infrastructure is going to look like. 22 00:01:09,035 --> 00:01:11,471 How big are the racks going to be, how much power do we need, 23 00:01:11,471 --> 00:01:12,572 how much cooling. 24 00:01:12,572 --> 00:01:16,776 So today, liquid cooling product manager, but then also the future next generation. 25 00:01:16,776 --> 00:01:19,379 So that's kind of my hybrid role that I play today. 26 00:01:19,379 --> 00:01:22,982 So Jason, what does it mean to you to be product manager? 27 00:01:23,583 --> 00:01:28,121 So for us and for me personally, product management is always about working 28 00:01:28,121 --> 00:01:29,656 kind of on the left with engineering 29 00:01:29,656 --> 00:01:32,592 and on the right with marketing and the customers understanding 30 00:01:32,592 --> 00:01:36,062 what we can do, what we should do and how we will do it with customers 31 00:01:36,062 --> 00:01:39,399 and making sure that their voice is represented every step of the way, 32 00:01:39,732 --> 00:01:42,902 but also that we represent our internal and kind of corporate intentions. 33 00:01:43,203 --> 00:01:46,106 How much these things should cost, how many can we sell, 34 00:01:46,106 --> 00:01:48,041 how do we actually deploy this in the market? 35 00:01:48,041 --> 00:01:50,743 Product management always plays that middle role to make sure 36 00:01:50,743 --> 00:01:53,880 what we're doing makes sense and actually is going to align with customer needs. 37 00:01:54,214 --> 00:01:55,748 Can we dissect the little bit? 38 00:01:55,748 --> 00:02:00,086 As you said, on one hand, you work with engineers to define it. 39 00:02:00,086 --> 00:02:02,555 So how does that work? Please explain to us. Yeah. 40 00:02:02,555 --> 00:02:06,259 So today a lot of what we do, especially with the liquid cooling, 41 00:02:06,593 --> 00:02:10,029 we're doing a lot of work internally, our own designs, also working 42 00:02:10,029 --> 00:02:11,431 with a lot of third party vendors 43 00:02:11,431 --> 00:02:14,434 to see where should we buy, what should we build ourselves. 44 00:02:14,701 --> 00:02:19,606 But we are pretty knowledgeable of our own competencies, our own abilities. 45 00:02:19,806 --> 00:02:23,910 But we shouldn't build and deploy every technology we think is interesting. 46 00:02:23,910 --> 00:02:24,878 We really have to 47 00:02:24,878 --> 00:02:28,715 first see what we can do, but also what actually makes sense in the market 48 00:02:28,982 --> 00:02:29,983 because we even have to spend 49 00:02:29,983 --> 00:02:33,520 our own R&D dollars, you know, either working with labs really far out 50 00:02:33,786 --> 00:02:36,990 or short term with our own thermal engineering group to decide 51 00:02:36,990 --> 00:02:39,025 what is the actual product we're going to build. 52 00:02:39,025 --> 00:02:41,995 And so we really have to be focused on what we could build. 53 00:02:41,995 --> 00:02:42,996 But does it make sense? 54 00:02:42,996 --> 00:02:45,532 Because we can do a lot of really amazing things, 55 00:02:45,532 --> 00:02:48,401 but should we do it and can we find someone to use it? 56 00:02:48,401 --> 00:02:52,172 And answer to that question you get by talking to customers? 57 00:02:52,172 --> 00:02:53,072 Absolutely. 58 00:02:53,072 --> 00:02:54,674 So how how does that work out? 59 00:02:54,674 --> 00:02:59,345 Yeah, so we do a lot of work surveying, we do a lot of customer site visits. 60 00:02:59,345 --> 00:03:03,283 We come to a lot of events like supercomputing in the U.S., 61 00:03:03,316 --> 00:03:05,185 ISC, often in Germany, 62 00:03:05,185 --> 00:03:08,354 where we actually get a lot of face time with customers and quite frankly, 63 00:03:08,354 --> 00:03:12,392 we do a lot of showing of roadmaps, but making sure that there's time 64 00:03:12,392 --> 00:03:14,160 at the end to say, does this actually make sense? 65 00:03:14,160 --> 00:03:15,828 Do you actually want this? 66 00:03:15,828 --> 00:03:18,531 Because we can show products all day long. 67 00:03:18,531 --> 00:03:19,832 But if customers are kind 68 00:03:19,832 --> 00:03:23,002 of giving us signals at the end saying, yes, this aligns with what we want, 69 00:03:23,603 --> 00:03:26,206 it often causes us to pause and go back to the team 70 00:03:26,206 --> 00:03:29,542 and say we didn't get the reception we wanted with. 71 00:03:29,542 --> 00:03:33,379 I think we need to do a workshop, we need to do a deep dive to really dig into. 72 00:03:33,813 --> 00:03:35,615 Do customers actually want this? 73 00:03:35,615 --> 00:03:39,118 Were they a bit shy during the meeting or what is their long term intention? 74 00:03:39,118 --> 00:03:41,788 Because the worst thing would be us getting our old, 75 00:03:41,788 --> 00:03:45,358 you know, drinking the Kool-Aid, getting really excited and we go to market 76 00:03:45,792 --> 00:03:47,860 and it's crickets and that would be the worst case. 77 00:03:47,860 --> 00:03:50,597 So we need to be validating every step of the way. 78 00:03:50,597 --> 00:03:54,801 So let's now focus on liquid cooling itself, how did it happen 79 00:03:54,801 --> 00:04:00,273 and why. So really liquid cooling has been prevalent in HPC for decades. 80 00:04:00,540 --> 00:04:05,011 Groups like HPE, IBM, Cray, SGI, 81 00:04:05,011 --> 00:04:08,848 they have all been working heavily in the space and built up great expertise. 82 00:04:09,482 --> 00:04:14,621 We have to use liquid cooling today to provide much better energy efficiency, 83 00:04:14,621 --> 00:04:19,859 but also be able to provide much denser racks, smaller racks, smaller 84 00:04:19,859 --> 00:04:24,130 infrastructure overall, but even enabling some high TDP processors. 85 00:04:24,297 --> 00:04:28,501 So today there's really a split like if we painted with broad 86 00:04:28,501 --> 00:04:32,605 strokes, you know, in my opinion, anything over 30 kilowatts. 87 00:04:32,605 --> 00:04:36,509 per rack is prime for liquid cooling based on really the rack 88 00:04:36,509 --> 00:04:38,077 densities, how much stuff is in there. 89 00:04:38,077 --> 00:04:40,513 It becomes quite difficult to air cool. 90 00:04:40,513 --> 00:04:46,019 Anything below that really between, you know, 15, 30 kilowatts is a gray zone. 91 00:04:46,019 --> 00:04:48,921 Sometimes it might not be totally, you know, 92 00:04:48,921 --> 00:04:51,324 worthwhile dollar wise to liquid cool it. 93 00:04:51,324 --> 00:04:54,694 There's more staff, cold plates, rear door heat exchangers 94 00:04:54,894 --> 00:04:57,830 that might not create, you know, the perfect tradeoff scenario. 95 00:04:57,830 --> 00:05:00,833 And then anything below that often doesn't make a lot of sense. 96 00:05:01,067 --> 00:05:05,238 And so HPC has been beyond 30 kilowatts for a very long time. 97 00:05:05,238 --> 00:05:08,708 If you look at any of the, you know, TOP500 systems today, 98 00:05:08,975 --> 00:05:13,446 a very large percentage of them are liquid cooling and anything in the top 99 00:05:13,446 --> 00:05:17,550 50 really firm today and going forward will only be liquid cooled. 100 00:05:17,550 --> 00:05:20,653 There will be almost no existence of purely air cooled systems. 101 00:05:20,653 --> 00:05:23,756 And it's just because of really the density and how hot the chips 102 00:05:23,756 --> 00:05:24,390 are themselves. 103 00:05:24,390 --> 00:05:26,793 They need a more efficient thermal technology. 104 00:05:26,793 --> 00:05:29,696 So it appears to me that while to most 105 00:05:29,696 --> 00:05:32,699 engineers, scaling means how many systems 106 00:05:33,032 --> 00:05:36,936 for you, it's how many systems in a rack you can build or on a..... 107 00:05:36,936 --> 00:05:41,140 whatever your dimensions are is that the .... . Totally, 108 00:05:41,341 --> 00:05:44,377 and it is kind of interesting because when we think about liquid 109 00:05:44,377 --> 00:05:48,715 cooling, there's really a lot of layers at the very micro layer 110 00:05:48,715 --> 00:05:53,386 there is the individual server that has some very specific thermal needs. 111 00:05:53,886 --> 00:05:57,857 There can be very hot CPUs, very hot memory, very hot, you know, 112 00:05:59,225 --> 00:06:02,595 voltage regulators that they themselves that if they're not cooled effectively. 113 00:06:02,595 --> 00:06:05,932 They won’t work, the boards will shut down, will experience thermal throttling. 114 00:06:06,599 --> 00:06:09,469 But even passed that server say you figured out the cold plate 115 00:06:09,469 --> 00:06:13,339 cooling at this level now how do you manage all the heat in a rack? 116 00:06:13,373 --> 00:06:18,711 So now we take this, you know, five 600 watt server, thousand watt server? 117 00:06:18,711 --> 00:06:21,447 And then we put 70 of them, you know, in a rack. 118 00:06:21,447 --> 00:06:25,351 So these, you know, Cray, 2000 style high density servers. 119 00:06:25,651 --> 00:06:28,621 Well, now you need to take all that heat away from the rack. 120 00:06:28,621 --> 00:06:32,291 Do you do CDUs do use do you do rear door heat exchangers, What's the technology 121 00:06:32,291 --> 00:06:33,126 you want to use? 122 00:06:33,126 --> 00:06:36,596 And then it scales out to what's the infrastructure in the building. 123 00:06:36,896 --> 00:06:40,566 So when we look at systems like Frontier, very large exascale 124 00:06:40,566 --> 00:06:44,670 systems, they're doing cold plate cooling, they're doing rack level cooling. 125 00:06:44,670 --> 00:06:47,940 And then the building infrastructure has to manage all of that heat. 126 00:06:47,940 --> 00:06:50,376 That is a tremendous amount of heat that has to be processed. 127 00:06:50,376 --> 00:06:51,878 So there's many layers. 128 00:06:51,878 --> 00:06:54,847 And that's what the kind of liquid cooling group is focusing on. 129 00:06:54,847 --> 00:06:58,384 We can build great cold plates, but if we don't have the proper technology 130 00:06:58,384 --> 00:07:02,121 to take the heat out of the rack itself, the solution isn't going to work. 131 00:07:02,588 --> 00:07:06,025 I don't want to interrupt you, but CDU for our customers is our cooling 132 00:07:06,025 --> 00:07:07,193 distribution unit. 133 00:07:07,193 --> 00:07:11,798 Okay, so when we built Frontier, what were the biggest challenges 134 00:07:11,798 --> 00:07:16,302 to the extent that you can share, this is a proprietary solution for sure. 135 00:07:16,302 --> 00:07:20,573 So so really of any technology using Cray EX in that case, 136 00:07:21,741 --> 00:07:24,944 in almost all cases, when you're building blades 137 00:07:24,977 --> 00:07:28,848 that are not not in watts, they're in the kilowatt range. 138 00:07:28,848 --> 00:07:31,184 There is a lot of density that's built in there. 139 00:07:31,184 --> 00:07:34,420 So you need really two kind of things to be thinking about. 140 00:07:34,420 --> 00:07:38,257 One, how do we take all of that heat out of the individual blades, 141 00:07:38,458 --> 00:07:39,792 bring it back to the racks? 142 00:07:39,792 --> 00:07:42,795 In this case, we use large row-base CDUs, we're pairing 143 00:07:43,062 --> 00:07:45,932 Big CDUs, big Poms, big heat exchangers 144 00:07:45,932 --> 00:07:49,268 with, you know, between three and four of these large Cray EX cabinets. 145 00:07:49,735 --> 00:07:53,172 But the other really interesting challenge is how do we make it fanless? 146 00:07:53,306 --> 00:07:54,574 And so Frontier, 147 00:07:54,574 --> 00:07:58,277 like all of the large supercomputers that we’ll build, have no fans in them. 148 00:07:58,444 --> 00:08:02,048 They're doing 100% liquid cooling, which means there is a lot of components 149 00:08:02,048 --> 00:08:03,382 at the blade level. 150 00:08:03,382 --> 00:08:06,385 So that's definitely an engineering challenge to work through. 151 00:08:06,452 --> 00:08:09,155 And you wanted to make them fanless 152 00:08:09,155 --> 00:08:12,391 because energy efficiency and density. 153 00:08:12,892 --> 00:08:16,562 And so really when we're trying to liquid cool these components, 154 00:08:16,762 --> 00:08:23,069 anywhere that we have to add fans is also kind of a resiliency issue, reliability. 155 00:08:23,069 --> 00:08:24,003 We have moving parts. 156 00:08:24,003 --> 00:08:26,239 We want to have as little as possible, 157 00:08:26,239 --> 00:08:28,274 but the big thing is really thermal management. 158 00:08:28,274 --> 00:08:30,409 So if we can remove all the fans 159 00:08:30,409 --> 00:08:33,412 and drive everything to the liquid cooled infrastructure, 160 00:08:33,412 --> 00:08:37,717 very, very energy efficient, if you look at the top 500 or top green 161 00:08:38,317 --> 00:08:41,020 kind of infrastructure rankings, 162 00:08:41,020 --> 00:08:45,091 always at the very top are Cray X systems, because there are no fans, very little 163 00:08:45,091 --> 00:08:48,461 electricity is moving the liquid around inside the cold plates. 164 00:08:49,095 --> 00:08:51,264 That's really important to these systems. 165 00:08:51,264 --> 00:08:54,834 So how do we compare to cloud based solutions? 166 00:08:54,834 --> 00:08:57,303 They have just like us, 167 00:08:58,304 --> 00:09:00,673 lots of homogeneity. 168 00:09:00,673 --> 00:09:04,544 I mean, there's heterogeneous components, but, you know, once 169 00:09:04,544 --> 00:09:09,749 you account for GPUs versus CPUs, you have homogeneous solution in a fact. 170 00:09:09,749 --> 00:09:10,516 Yeah. 171 00:09:10,516 --> 00:09:12,018 So one of the interesting things 172 00:09:12,018 --> 00:09:15,755 when we think about cooling is really comes back to density. 173 00:09:15,755 --> 00:09:18,925 So the earlier question kind of why liquid cooling, 174 00:09:19,091 --> 00:09:21,961 you know, how is it evolving in HPC? 175 00:09:21,961 --> 00:09:24,063 The rack densities are so high, it really 176 00:09:24,063 --> 00:09:27,533 it makes easy sense for us to justify the additional infrastructure here. 177 00:09:27,900 --> 00:09:30,703 But for many cloud computing 178 00:09:30,703 --> 00:09:33,773 groups, rack densities are simply not that high. 179 00:09:33,773 --> 00:09:37,610 So they have a lot more of a gray area for where they want to play. 180 00:09:37,877 --> 00:09:40,746 Do they want to add liquid cooling or are they going to invest in their 181 00:09:40,746 --> 00:09:42,148 air cooled infrastructure? 182 00:09:42,148 --> 00:09:45,351 Did they move to a state that has very low electricity costs 183 00:09:45,384 --> 00:09:46,852 and they can buy land cheaply? 184 00:09:46,852 --> 00:09:50,623 We've seen a lot of kind of hyperscalers move out to, well, 185 00:09:50,623 --> 00:09:53,960 you know, very low density density areas of the U.S. 186 00:09:53,960 --> 00:09:57,163 They have cheap electricity and they can just build a new data center. 187 00:09:57,430 --> 00:09:58,598 But if you're building somewhere 188 00:09:58,598 --> 00:10:01,934 like San Francisco, you're going to have kind of a perfect storm. 189 00:10:02,368 --> 00:10:02,835 You know, high 190 00:10:02,835 --> 00:10:06,572 real estate costs, high electricity costs and not a lot of room to build. 191 00:10:06,939 --> 00:10:10,176 That's where high density kind of liquid cooling can really play a strong role. 192 00:10:10,176 --> 00:10:11,944 So it kind of depends 193 00:10:11,944 --> 00:10:15,815 where groups are building and really what their own corporate agenda is. 194 00:10:15,815 --> 00:10:19,919 If they have money to burn and energy efficiency doesn't matter and they have 195 00:10:19,919 --> 00:10:23,456 the space like we talked about, liquid cooling is often not prioritized. 196 00:10:23,456 --> 00:10:25,725 But we see much of the world U.S. 197 00:10:25,725 --> 00:10:30,363 and Europe in particular, that even cloud providers are really trying to figure out 198 00:10:30,363 --> 00:10:33,733 this technology where they can have very high energy efficiency, 199 00:10:34,200 --> 00:10:36,569 but also very high resiliency. 200 00:10:36,569 --> 00:10:37,970 They contain components out. 201 00:10:37,970 --> 00:10:41,340 Yeah, components fit on whatever they want to put in the racks. 202 00:10:41,641 --> 00:10:46,012 And that's one of the challenges of liquid cooling is unlike air cooling 203 00:10:46,012 --> 00:10:49,982 that you have the fans in the server, you can put kind of anything you want 204 00:10:50,082 --> 00:10:53,753 as long as there is effective heat sinks that are built into the servers, 205 00:10:54,420 --> 00:10:55,655 things are going to work. 206 00:10:55,655 --> 00:10:56,455 Liquid cooling. 207 00:10:56,455 --> 00:10:59,792 We need to design the components to fit the components. 208 00:10:59,792 --> 00:11:03,663 So you have to design a cold plate that's going to fit on the latest AMD 209 00:11:04,063 --> 00:11:07,667 CPU or the latest memory or voltage regulators. 210 00:11:07,833 --> 00:11:10,836 So there has to be more foresight and kind of thinking about 211 00:11:10,836 --> 00:11:12,438 how we're going to design products. 212 00:11:12,438 --> 00:11:15,441 We can't kind of just slap a PCIe card in there 213 00:11:16,008 --> 00:11:17,710 and have it be liquid cooled without thinking. 214 00:11:17,710 --> 00:11:20,179 How does it work with the greater infrastructure, 215 00:11:20,179 --> 00:11:24,250 given the tremendous complexity of the system that you just described 216 00:11:24,283 --> 00:11:30,690 and many not moving parts but moving heat paths, if you will, 217 00:11:31,824 --> 00:11:32,291 is there any 218 00:11:32,291 --> 00:11:35,461 opportunity to do co-design between 219 00:11:36,662 --> 00:11:39,665 your products and the whole infrastructure 220 00:11:39,665 --> 00:11:43,969 and then hardware and then even software, both systems software and applications? 221 00:11:44,470 --> 00:11:46,439 Absolutely. 100%, yes. 222 00:11:46,439 --> 00:11:48,641 And so that's where now kind of exciting. 223 00:11:48,641 --> 00:11:51,410 We get to talk about next generation infrastructure. 224 00:11:51,410 --> 00:11:53,345 So today, 225 00:11:53,345 --> 00:11:57,116 liquid cooling, you know, really when we talk about Cray X, 226 00:11:57,550 --> 00:12:00,486 Cray XD, even our Proliant series, 227 00:12:00,486 --> 00:12:03,956 all of those products and platforms have liquid cooling today. 228 00:12:04,657 --> 00:12:06,025 Most of that is steady state. 229 00:12:06,025 --> 00:12:09,028 So often we'll have CDUs with Poms 230 00:12:09,695 --> 00:12:13,065 kind of obviously watching pressure, watching temperature, watching flow. 231 00:12:13,065 --> 00:12:18,704 But they are most often providing pretty stable, constant flow to the rack 232 00:12:18,704 --> 00:12:19,505 of really regardless 233 00:12:19,505 --> 00:12:23,409 of what's happening next generation and what I would expect in the industry 234 00:12:23,409 --> 00:12:28,147 overall is exactly like you said this more, you know, more harmonious 235 00:12:28,481 --> 00:12:32,551 kind of coexistence between power management, you know, 236 00:12:32,551 --> 00:12:35,588 software features and the actual liquid cooling infrastructure. 237 00:12:35,588 --> 00:12:40,092 And so I believe we'll see much more energy efficiency, you know, 238 00:12:40,092 --> 00:12:44,997 kind of backing up, thinking about how server fan tables work today. 239 00:12:45,297 --> 00:12:48,267 When a server ramps up, the fans turn out 240 00:12:48,267 --> 00:12:51,937 when the servers are going into low power state, the fans ramp down. 241 00:12:51,937 --> 00:12:54,173 I expect there's going to be more 242 00:12:54,173 --> 00:12:59,078 versatility in cooling so we can give very hot processors a lot of cooling. 243 00:12:59,311 --> 00:13:01,680 And then for servers that are running an idle, 244 00:13:01,680 --> 00:13:04,984 very low amount of cooling, but that has to be managed through software. 245 00:13:05,217 --> 00:13:07,386 And so I think that is absolutely where we're going. 246 00:13:07,386 --> 00:13:10,623 And how do you manage it, how you make sure that 247 00:13:11,423 --> 00:13:14,960 some CPUs don't go above the threshold. 248 00:13:15,327 --> 00:13:20,166 We've been looking, for example, digital twins, but they're also sophisticated 249 00:13:21,133 --> 00:13:21,901 management 250 00:13:21,901 --> 00:13:25,137 infrastructure, software management, infrastructure to make sure. 251 00:13:25,137 --> 00:13:26,472 How do you do all of that? 252 00:13:26,472 --> 00:13:29,341 Yeah, that is, I think exactly the answer is 253 00:13:29,341 --> 00:13:32,311 the power management software within the rack. 254 00:13:32,344 --> 00:13:36,549 You know, no controller or software and hardware as well. 255 00:13:36,649 --> 00:13:39,718 Those pieces have to be talking with each other. 256 00:13:39,718 --> 00:13:42,721 So CPUs are running their own internal 257 00:13:42,822 --> 00:13:46,659 kind of thermal management so they know what their temperature is. 258 00:13:46,859 --> 00:13:51,096 They're feeding it back to the kind of larger software controlling the nodes. 259 00:13:51,330 --> 00:13:54,900 But that also needs to then communicate with either the CDU 260 00:13:54,900 --> 00:13:57,903 or the cooling infrastructure so we can ramp up and down. 261 00:13:58,103 --> 00:14:00,906 Right now, there's not a lot of that we see in the industry 262 00:14:00,906 --> 00:14:04,610 where there is this kind of beautiful harmony between all the systems, 263 00:14:05,010 --> 00:14:06,145 but that is where we're going to go. 264 00:14:06,145 --> 00:14:08,581 And so that definitely comes from the software. 265 00:14:08,581 --> 00:14:12,785 Speaking of the direction where we are going, what comes after liquid cooling? 266 00:14:12,952 --> 00:14:14,019 That's a very good question. 267 00:14:14,019 --> 00:14:16,322 So it will be more liquid cooling. 268 00:14:16,322 --> 00:14:19,325 You know, when we talk about the future of liquid cooling today, 269 00:14:20,226 --> 00:14:23,329 it is, I would say, the most prevalent technology deployed 270 00:14:23,329 --> 00:14:28,067 today is sky thin, cold plates, Technology that has been around for, you know, 271 00:14:28,067 --> 00:14:32,605 dozens of years is very reliable, very prevalent in the industry. 272 00:14:32,972 --> 00:14:36,942 What will come next, I think is going to be really hybrid technologies. 273 00:14:36,942 --> 00:14:41,847 So we're going to see more two phase cold plate cooling. 274 00:14:41,847 --> 00:14:43,449 So how we're going to see, you know, 275 00:14:43,449 --> 00:14:47,319 within the actual cold plate a phase change between liquid to gas. 276 00:14:47,553 --> 00:14:49,622 I believe we'll see more immersion. 277 00:14:49,622 --> 00:14:52,591 I believe we'll still go see a lot of single phase. 278 00:14:52,591 --> 00:14:55,661 And this is, you know, always a difficult question 279 00:14:55,661 --> 00:14:59,832 to answer because we can see the processor roadmaps. 280 00:14:59,832 --> 00:15:01,600 TDP is going up very quickly. 281 00:15:01,600 --> 00:15:05,371 You know what used to be a high end processor three, four years 282 00:15:05,371 --> 00:15:08,407 ago, 250, 300 watts for CPU. 283 00:15:08,407 --> 00:15:11,377 We're going to see beyond 500 watts very soon. 284 00:15:11,377 --> 00:15:13,646 GPUs are going to be over a thousand watts. 285 00:15:13,646 --> 00:15:16,081 And so the future is going to be, 286 00:15:16,081 --> 00:15:19,652 I think, a lot of different technologies for different customer uses. 287 00:15:19,952 --> 00:15:24,423 But to be very honest, single phase cold plates are still 288 00:15:24,423 --> 00:15:28,294 one of the most reliable, most resilient technologies that just work. 289 00:15:28,594 --> 00:15:31,363 And though there is a lot of different stuff happening in the market, 290 00:15:32,464 --> 00:15:33,866 many OEM groups have used 291 00:15:33,866 --> 00:15:37,803 that technology for a long time and it has worked very well. 292 00:15:37,803 --> 00:15:40,739 And so I think we're going to see a lot of different stuff come out. 293 00:15:40,739 --> 00:15:45,311 The winners are going to be the most reliable, resilient. 294 00:15:45,311 --> 00:15:49,648 How easy can we swap out the components, But how do they just work? 295 00:15:49,648 --> 00:15:52,651 And the simplicity is going to be an important part. 296 00:15:52,685 --> 00:15:55,487 You know, we talk about this even with labs. 297 00:15:55,487 --> 00:15:59,191 There are a lot of really amazing technologies that are out there 298 00:15:59,491 --> 00:16:02,328 that from a science perspective 299 00:16:02,328 --> 00:16:04,663 are tremendously cool. 300 00:16:04,663 --> 00:16:08,067 But how do we deploy them at scale and build them reliably 301 00:16:08,067 --> 00:16:11,103 and then put on an HPE warranty and say you're good to go? 302 00:16:11,103 --> 00:16:12,404 Yeah, because that's the problem. 303 00:16:12,404 --> 00:16:14,173 We can build anything we want. 304 00:16:14,173 --> 00:16:14,707 We have some of the 305 00:16:14,707 --> 00:16:18,477 smartest people in the world working on our industry's challenges. 306 00:16:18,844 --> 00:16:22,314 But if you need, you know, your own supercomputer 307 00:16:22,314 --> 00:16:26,218 to manage the cooling infrastructure because it's a very sophisticated system 308 00:16:26,251 --> 00:16:29,288 that's kind of on the thin edge, you know, when we talk about 309 00:16:30,522 --> 00:16:32,257 future cooling 310 00:16:32,257 --> 00:16:35,928 that that need like absolute zero, How do you regulate that? 311 00:16:35,928 --> 00:16:38,497 Is that something any customer can buy? 312 00:16:38,497 --> 00:16:39,732 Maybe not all the time. 313 00:16:39,732 --> 00:16:43,068 So we need stuff that's really reliable that just works. 314 00:16:44,370 --> 00:16:46,071 So you mentioned it has to work 315 00:16:46,071 --> 00:16:49,074 for AMD, for Intel, for GPUs. 316 00:16:49,241 --> 00:16:54,179 There's some underlying implied interoperability 317 00:16:54,780 --> 00:16:58,784 that you need to or supporting multiple components. 318 00:16:59,485 --> 00:17:01,820 Is there any implied standard? 319 00:17:01,820 --> 00:17:04,189 How do you deal with these issues? 320 00:17:04,189 --> 00:17:08,627 Yeah, So, you know, from a just a fit and form function, 321 00:17:09,995 --> 00:17:11,897 you know, between AMD CPU 322 00:17:11,897 --> 00:17:15,100 and an intel CPU, you know, NVIDIA GPU, 323 00:17:15,601 --> 00:17:18,604 as long as we understand kind of the socket geometry, 324 00:17:18,937 --> 00:17:22,908 what its thermal characteristics are, especially around TCase. 325 00:17:22,908 --> 00:17:27,479 So the maximum thermal temperature these CPUs or GPUs can go up to before 326 00:17:27,479 --> 00:17:30,783 they experience thermal issues, we can design around them. 327 00:17:30,783 --> 00:17:34,953 Now the nice thing for the most part is an AMD CPU 328 00:17:35,487 --> 00:17:38,924 within a short period of time from one generation to the other 329 00:17:38,957 --> 00:17:42,961 will most oftenly have the same socket kind of dimensions. 330 00:17:42,961 --> 00:17:46,932 So if we design a cold plate for one and the thermal characteristics 331 00:17:46,932 --> 00:17:50,569 make sense, we can use that same cold plate for the next generation. 332 00:17:51,203 --> 00:17:52,938 As long as we have the kind of 333 00:17:53,972 --> 00:17:57,476 visibility to next generation, we can design very easily for that. 334 00:17:57,476 --> 00:18:00,479 So often it's the geometry of how the CPU is 335 00:18:00,479 --> 00:18:03,482 physically built with the socket and then its thermal characteristics. 336 00:18:04,049 --> 00:18:05,851 We work very closely with all those groups. 337 00:18:05,851 --> 00:18:11,056 So often we are building our solutions with them multiple generations in advance. 338 00:18:11,056 --> 00:18:14,059 And really those are kind of the some of the key criteria we have to look at. 339 00:18:14,326 --> 00:18:17,329 So I'd like to touch a little bit on the business model. 340 00:18:17,996 --> 00:18:21,066 It appears to me that liquid cooling is highly proprietary. 341 00:18:21,066 --> 00:18:25,270 That's our key differentiator for our largest top supercomputers. 342 00:18:26,071 --> 00:18:29,608 Is there any opportunity for open hardware 343 00:18:29,942 --> 00:18:32,945 equivalent for open liquid cooling? 344 00:18:33,011 --> 00:18:36,782 I'm not saying you give it away, but are there 345 00:18:36,782 --> 00:18:40,352 some interfaces that you can expose that others can innovate, 346 00:18:42,421 --> 00:18:43,989 motivating them 347 00:18:43,989 --> 00:18:47,759 to continue using our or maybe perhaps other business model for HPE? 348 00:18:48,060 --> 00:18:53,432 Yeah, this is a very interesting question because a lot of people ask about this, 349 00:18:54,032 --> 00:18:57,102 and the thing I always like to explain 350 00:18:57,102 --> 00:19:00,105 is today, let's say, 351 00:19:00,139 --> 00:19:03,375 you know, no liquid cooling infrastructure at all, just air cooled stuff, 352 00:19:04,643 --> 00:19:08,447 HPE and Dell and Lenovo and all these different groups 353 00:19:09,148 --> 00:19:13,152 are not looking to sell multi vendor rack solutions. 354 00:19:13,152 --> 00:19:13,685 You know, it's 355 00:19:13,685 --> 00:19:18,257 very rare that you would see Dell gear with HPE gear in the same rack. 356 00:19:18,257 --> 00:19:19,725 You know, it's not very common. 357 00:19:19,725 --> 00:19:21,994 And so when we, 358 00:19:21,994 --> 00:19:24,997 you know, accept that as the standard that when customers buy 359 00:19:25,164 --> 00:19:28,167 a rack of stuff, it's going to be primarily HPE. 360 00:19:28,467 --> 00:19:32,538 I believe that kind of gives way that HPE can then control the infrastructure. 361 00:19:32,671 --> 00:19:36,575 Now, how this rack integrates with the rest of the facility 362 00:19:36,575 --> 00:19:39,478 that really matters. We need to use standardized connections. 363 00:19:39,478 --> 00:19:41,880 Now, today that happens, which is very nice. 364 00:19:41,880 --> 00:19:44,883 We use sanitary flanges, tri clamps. 365 00:19:44,883 --> 00:19:48,820 There's a lot of kind of industry standard piping standards that we use. 366 00:19:48,820 --> 00:19:53,192 So it's not very difficult to, you know, build a rack of, you know, HPE 367 00:19:53,192 --> 00:19:57,763 Cray XD stuff and then another vendor can build in the same data center 368 00:19:58,397 --> 00:20:02,334 to meet the cooling you know, componentry 369 00:20:02,935 --> 00:20:04,436 is part of the HPE IP differentiation. 370 00:20:04,436 --> 00:20:05,938 is part of the HPE IP differentiation. 371 00:20:06,138 --> 00:20:09,007 And so it's I think we're at a very interesting 372 00:20:09,007 --> 00:20:12,978 tipping point where today there's a lot of stuff on the market that is 373 00:20:14,012 --> 00:20:15,147 similar. 374 00:20:15,147 --> 00:20:16,682 It uses skived cold plates. 375 00:20:16,682 --> 00:20:19,751 We talked about there's a lot of rear door heat exchanger vendors. 376 00:20:19,985 --> 00:20:22,721 So the components are very similar. 377 00:20:22,721 --> 00:20:25,958 These similar connections, quick disconnects, 378 00:20:25,958 --> 00:20:28,327 you know how the technology works. 379 00:20:28,327 --> 00:20:32,130 But I have always believed that, you know, when we moved to this 380 00:20:32,264 --> 00:20:35,267 this new thermal curve we're seeing for processors, 381 00:20:35,500 --> 00:20:39,137 there are going to be certain companies that can simply cool it and some 382 00:20:39,137 --> 00:20:40,606 they simply cannot. 383 00:20:40,606 --> 00:20:43,108 And there's a there's kind of a perfect storm happening 384 00:20:43,108 --> 00:20:48,213 where we talk to a TCase, so thermal kind of design parameters are going down. 385 00:20:48,213 --> 00:20:51,416 Components want to be cooler before they experience thermal shutdown, 386 00:20:51,650 --> 00:20:54,219 but at the same time, customers want to use the warmer water. 387 00:20:54,219 --> 00:20:55,887 That's all about the energy efficiency. 388 00:20:55,887 --> 00:20:59,191 So how do you when this window starts getting smaller and smaller, 389 00:20:59,458 --> 00:21:02,194 use technology that maybe isn't ready 390 00:21:02,194 --> 00:21:05,197 isn't wasn't designed for that kind of componentry. 391 00:21:05,197 --> 00:21:07,699 So I think we're going to see some companies 392 00:21:07,699 --> 00:21:12,070 really focus on their innovation because their server technology is so amazing. 393 00:21:12,070 --> 00:21:16,141 But now paired with cooling solutions that competitors don't have 394 00:21:17,643 --> 00:21:18,644 as long as we have, 395 00:21:18,644 --> 00:21:21,713 you know, components that can connect to the facility piping, that's important. 396 00:21:21,713 --> 00:21:23,548 But the rest of it, I think there will see 397 00:21:23,548 --> 00:21:26,318 more differentiation happening with groups. 398 00:21:26,318 --> 00:21:29,888 And I don't know if it needs to all be, you know, totally open 399 00:21:29,888 --> 00:21:32,891 because that's going to be part of that group's kind of value offering. 400 00:21:33,091 --> 00:21:36,628 Where I was going more is that there is a flurry 401 00:21:36,628 --> 00:21:39,631 of new accelerators, including very large ones, 402 00:21:40,465 --> 00:21:43,969 different ones, possible different geometries. 403 00:21:44,403 --> 00:21:45,871 Do you have any guidance for them? 404 00:21:45,871 --> 00:21:48,206 Because it doesn't make any sense for them 405 00:21:48,206 --> 00:21:52,177 to build their own liquid cooling because they're belly differentiations? 406 00:21:52,177 --> 00:21:55,814 Are accelerators, possibly compiler software, 407 00:21:55,847 --> 00:21:58,850 not building the whole infrastructure? 408 00:21:58,950 --> 00:22:02,888 Could they somehow leverage our approaches, license, 409 00:22:03,388 --> 00:22:03,989 ABI or a rack and all of that 410 00:22:03,989 --> 00:22:04,623 ABI or a rack and all of that 411 00:22:04,623 --> 00:22:05,891 By power, racks, and all of that 412 00:22:05,891 --> 00:22:08,894 so that it's win win for them and for us? 413 00:22:08,927 --> 00:22:09,628 Definitely. 414 00:22:09,628 --> 00:22:09,928 Yeah. 415 00:22:09,928 --> 00:22:13,432 So I think that is going to be something that HPE does very well. 416 00:22:13,432 --> 00:22:15,534 We know a lot about infrastructure. 417 00:22:15,534 --> 00:22:18,003 We are really a solutions company. 418 00:22:18,003 --> 00:22:20,839 You know, when we look to groups like AMD 419 00:22:20,839 --> 00:22:24,543 and Intel, they are amazing at building fantastic chips, 420 00:22:24,810 --> 00:22:28,213 but they need partners like us that build the rest of the solutions. 421 00:22:28,714 --> 00:22:31,416 Even all the way down to who's going to install these systems. 422 00:22:31,416 --> 00:22:34,019 And so we work very closely with a lot of those groups. 423 00:22:34,019 --> 00:22:36,321 On what will the full solution be, 424 00:22:36,321 --> 00:22:39,658 where we understand the parameters of their chips, how hot they will be. 425 00:22:39,991 --> 00:22:42,961 Like I said, going back to basics, we'll build from the cold plates, to the CDU 426 00:22:42,994 --> 00:22:45,197 to the rack, to the whole infrastructure. 427 00:22:45,197 --> 00:22:48,033 That's the role that we play. So we do a lot of collaboration today. 428 00:22:48,967 --> 00:22:50,235 I was really excited 429 00:22:50,235 --> 00:22:53,805 to see with what energy you talk about all of this. 430 00:22:54,206 --> 00:22:58,710 It reminded me of a startup guy talking not of an enterprise guy. 431 00:22:59,211 --> 00:23:04,149 So you, you are definitely working at the bleeding edge of innovation. 432 00:23:04,549 --> 00:23:05,817 So what's the difference? 433 00:23:05,817 --> 00:23:09,254 What's the difference between your enterprise corporate role 434 00:23:09,588 --> 00:23:11,490 product manager and a startup guy? 435 00:23:11,490 --> 00:23:12,257 Sure, yeah. 436 00:23:12,257 --> 00:23:15,660 So that that's very relevant because my experience is primarily 437 00:23:15,660 --> 00:23:19,231 been working with startups and even kind of coming to HPE. 438 00:23:19,231 --> 00:23:24,202 I worked for a group that worked with HPE and HPC to me really is a startup 439 00:23:24,202 --> 00:23:27,639 and know it's it's kind of cliche, but I think it is really true 440 00:23:27,639 --> 00:23:31,743 when we look at our organizational stack and especially when we look 441 00:23:31,743 --> 00:23:36,748 to our kind of colleagues in Proliant that enterprise level server for them. 442 00:23:37,182 --> 00:23:37,482 You know, 443 00:23:37,482 --> 00:23:42,721 we always kind of laugh at ourselves because we we ask them about run rates 444 00:23:42,721 --> 00:23:45,891 and about kind of predictability of their their deal flow. 445 00:23:46,124 --> 00:23:48,627 And in the enterprise market. 446 00:23:48,627 --> 00:23:50,562 And, you know, it's very it's 447 00:23:50,562 --> 00:23:54,132 not extremely predictable, but it's much more predictable. 448 00:23:54,733 --> 00:23:57,335 In HPC, there's a lot that happens year over year. 449 00:23:57,335 --> 00:24:00,071 You know, there's only so many access scale deals, 450 00:24:00,071 --> 00:24:03,074 there's only so many very large HPC opportunities. 451 00:24:03,241 --> 00:24:06,812 And even at the mid-tier, there's so much that changes in the market. 452 00:24:06,812 --> 00:24:08,680 And so we really have to be adaptable. 453 00:24:08,680 --> 00:24:14,653 And so, you know, in my time at HPE, I have seen such tremendous change 454 00:24:14,653 --> 00:24:15,587 within our teams. 455 00:24:15,587 --> 00:24:19,024 Out of necessity, you know, one year we are focusing on 456 00:24:19,357 --> 00:24:21,326 a lot of exascale systems. 457 00:24:21,326 --> 00:24:22,994 We are looking at HPC with a, 458 00:24:22,994 --> 00:24:26,832 you know, fine tuned vision on what will the software be next year. 459 00:24:26,832 --> 00:24:29,534 Now AI is becoming extremely prevalent. 460 00:24:29,534 --> 00:24:32,537 What is an AI solution? Is it just software? 461 00:24:32,804 --> 00:24:35,106 How does the hardware marriage marry with it? 462 00:24:35,106 --> 00:24:39,544 And so we are constantly scrumming as our own teams kind of cross group. 463 00:24:39,578 --> 00:24:43,648 So right now, even for what we're doing next generation, we are running 464 00:24:43,882 --> 00:24:46,651 software core team, service core team, service 465 00:24:46,651 --> 00:24:49,921 is going to be such an important part of what we do, hardware core teams. 466 00:24:49,921 --> 00:24:52,457 Liquid cooling. But these groups are all talking 467 00:24:52,457 --> 00:24:56,895 because we ourselves are going to get left in the past if we are not, you know, 468 00:24:56,928 --> 00:24:59,931 two years out, we released a product, but software wasn't aware of it. 469 00:25:00,298 --> 00:25:03,134 Well, that's a big problem where software is working on something 470 00:25:03,134 --> 00:25:04,970 that doesn't work without service. 471 00:25:04,970 --> 00:25:06,538 How are we going to marry together? 472 00:25:06,538 --> 00:25:10,141 And so it's very startup mentality where we're not meeting, you know, 473 00:25:10,141 --> 00:25:13,411 quarterly, we're meeting weekly talking about what's happening. 474 00:25:13,411 --> 00:25:15,580 People are very open, you know, 475 00:25:15,580 --> 00:25:20,252 at not quite radical candor, but very transparent intentions. 476 00:25:20,252 --> 00:25:22,053 We don't figure out serviceability. 477 00:25:22,053 --> 00:25:23,522 No one's going to buy the system. 478 00:25:23,522 --> 00:25:25,123 The software doesn't make sense. 479 00:25:25,123 --> 00:25:26,291 How is the hardware going to work? 480 00:25:26,291 --> 00:25:28,827 So we're all working very much in a startup mentality. 481 00:25:28,827 --> 00:25:31,696 So we try to push, push, push that we are an agile group. 482 00:25:31,696 --> 00:25:37,202 I love your definition of HPE as a marriage console for HPC and AI. 483 00:25:37,235 --> 00:25:39,804 Yeah, exactly. Exactly. 484 00:25:39,804 --> 00:25:41,273 So tell me a little bit. 485 00:25:41,273 --> 00:25:43,775 You're coming from Calgary. Is that right? That's right, yeah. 486 00:25:43,775 --> 00:25:48,280 So how is I mean, and by the way, we are now here at SC 23. 487 00:25:48,280 --> 00:25:48,980 Yeah. 488 00:25:48,980 --> 00:25:51,516 Used to be called supercomputing in Denver. 489 00:25:51,516 --> 00:25:55,320 I guess temperature wise it's similar, very similar, but other than that. 490 00:25:56,288 --> 00:25:57,022 Well, so it's 491 00:25:57,022 --> 00:26:00,091 actually funny, when I come to Denver, it actually kind of feels like Calgary. 492 00:26:00,091 --> 00:26:03,595 Sometimes we can see the mountains from our hotel room, just like Calgary. 493 00:26:04,362 --> 00:26:05,897 Still get a lot of snow in the winter. 494 00:26:05,897 --> 00:26:09,801 And what is always very interesting, you know, we talk about Denver, 495 00:26:09,801 --> 00:26:13,371 but then we also talk about Houston with HPE and all the different groups as 496 00:26:14,139 --> 00:26:18,710 there's so many similarities that, you know, even Denver and Houston, 497 00:26:18,710 --> 00:26:20,612 when I tie those two together, 498 00:26:20,612 --> 00:26:24,316 are very kind of oil and gas kind of historical presence. 499 00:26:24,316 --> 00:26:27,252 That's where money was made. That's where a lot of the smart people went. 500 00:26:27,252 --> 00:26:30,255 And what is so interesting, Calgary very much like that is 501 00:26:30,555 --> 00:26:33,358 as oil and gas kind of booms and busts. 502 00:26:33,358 --> 00:26:36,027 So does technology. And I see kind of these 503 00:26:37,295 --> 00:26:39,164 kind of symbiotic curves happening 504 00:26:39,164 --> 00:26:43,435 when oil booms, all the tech kind of groups kind of go into oil. 505 00:26:43,435 --> 00:26:46,237 How can we build oil focused technology 506 00:26:46,237 --> 00:26:49,608 as a very large kind of oil and gas focused kind of sales group? 507 00:26:49,608 --> 00:26:52,010 We built some really amazing technologies, 508 00:26:52,010 --> 00:26:55,981 but when oil crashes, we see all these very smart people coming to tech. 509 00:26:55,981 --> 00:26:59,818 And so there are a lot of people we work with that have some time worked in 510 00:26:59,818 --> 00:27:02,821 oil and gas that also work in supercomputing 511 00:27:03,021 --> 00:27:05,724 and then in Calgary also build beautiful breweries. 512 00:27:05,724 --> 00:27:07,359 I often I see that in the U.S., too. 513 00:27:07,359 --> 00:27:10,795 When oil crashes, all these smart people with money use their brains 514 00:27:10,795 --> 00:27:14,232 to build stuff that we love to use, you know, evenings and weekends. 515 00:27:14,466 --> 00:27:17,035 So we see a lot of that kind of boom and bust cycle happening. 516 00:27:17,035 --> 00:27:19,904 Will compare the brewery notes separately. There you go. Yeah. 517 00:27:21,706 --> 00:27:23,308 I think in general 518 00:27:23,308 --> 00:27:27,912 HPC is very humanity friendly because we are solving 519 00:27:27,912 --> 00:27:31,616 not just oil and gas but also weather and many other problems. 520 00:27:31,983 --> 00:27:34,352 So have you ever thought about it? 521 00:27:34,352 --> 00:27:38,123 Did it ever fulfill you more than doing other jobs? 522 00:27:38,123 --> 00:27:38,690 For sure. 523 00:27:38,690 --> 00:27:42,494 I think that that is really, you know, an important 524 00:27:42,494 --> 00:27:45,563 part of her basic kind of hiring strategy. 525 00:27:45,597 --> 00:27:49,267 Our basic kind of corporate culture is we are not just selling computers. 526 00:27:49,434 --> 00:27:53,071 We're not selling kind of laptops to fill quota. 527 00:27:53,505 --> 00:27:55,840 What we're doing is really pushing humanity forward. 528 00:27:55,840 --> 00:27:59,978 So when we look at Frontier, it's such an easy flagship thing to talk about. 529 00:28:00,345 --> 00:28:03,481 It is working on some of humanity's greatest problems 530 00:28:03,815 --> 00:28:07,419 because it is a combination of high powered hardware, 531 00:28:07,419 --> 00:28:11,623 software, fabrics, all kind of the machine that makes it happen. 532 00:28:12,657 --> 00:28:14,192 That is something definitely to feel good about. 533 00:28:14,192 --> 00:28:18,396 And you can bet at every interview that we run for new grads 534 00:28:18,396 --> 00:28:21,533 and for folks kind of moving industries, we lead with that. 535 00:28:21,533 --> 00:28:23,868 We're not just building computers or laptops. 536 00:28:23,868 --> 00:28:26,871 We're building kind of technology that moves humanity forward. 537 00:28:27,806 --> 00:28:29,374 Definitely. That's what I tell my family I do. 538 00:28:29,374 --> 00:28:31,509 It makes me feel good about it. So well. 539 00:28:31,509 --> 00:28:35,880 I've done future workforce evaluation and the current generation 540 00:28:35,880 --> 00:28:37,682 wants to be fulfilled. 541 00:28:37,682 --> 00:28:41,086 They they're not caring only about salary, about, 542 00:28:41,619 --> 00:28:46,424 you know, all kinds of old thinking, but they want to know 543 00:28:46,558 --> 00:28:49,561 that their company is doing something good. 544 00:28:49,994 --> 00:28:53,865 I usually end up this discussion with recommendation for the book. 545 00:28:54,032 --> 00:28:57,001 Did you have any good one that you can recommend 546 00:28:57,535 --> 00:28:59,537 if you even have the time to read? 547 00:28:59,537 --> 00:29:01,906 Yeah. So? So. 548 00:29:01,906 --> 00:29:05,944 So one of the books that I actually use quite often 549 00:29:05,944 --> 00:29:09,514 I talk with my new team members on. 550 00:29:09,514 --> 00:29:14,319 It's called the first 90 Days and I 100% was recommended it 551 00:29:14,319 --> 00:29:18,423 by a friend in the industry before I joined HPE, read it and used it 552 00:29:18,423 --> 00:29:22,026 and I literally just recommended it to some new staff that joined last week. 553 00:29:22,560 --> 00:29:25,997 The main kind of takeaway for me, one of the important nuggets is really 554 00:29:26,364 --> 00:29:31,202 what are you going to do in the first 90 days to really let people know you exist? 555 00:29:31,536 --> 00:29:34,472 But to really clarify, what does the group think 556 00:29:34,472 --> 00:29:37,509 that you're going to be doing and what's your own kind of path forward? 557 00:29:37,509 --> 00:29:42,080 And though that's kind of generic advice, I think that any organization you work in 558 00:29:43,081 --> 00:29:45,049 being the new person can be difficult. 559 00:29:45,049 --> 00:29:47,085 Especially a group like HPE were so large. 560 00:29:47,085 --> 00:29:49,954 We've so many different kind of departments and groups 561 00:29:49,954 --> 00:29:53,057 that in the first 90 days you really want to understand 562 00:29:53,057 --> 00:29:56,561 what's the culture of the organization, how can you contribute to it? 563 00:29:56,795 --> 00:29:59,497 But really, what do people think that you're going to do? 564 00:29:59,497 --> 00:30:03,868 And so when I joined, I introduced myself as kind of the liquid cooling guy. 565 00:30:03,868 --> 00:30:05,970 Here's what I think I'm going to be doing. 566 00:30:05,970 --> 00:30:07,172 Do you think that makes sense? 567 00:30:07,172 --> 00:30:10,842 Because there's nothing more genuine in candor 568 00:30:11,042 --> 00:30:14,579 than the people working on the ground floor saying you were hired to do what? 569 00:30:14,579 --> 00:30:17,315 No, no, no, no. We have some big problems over here. 570 00:30:17,315 --> 00:30:18,483 You have time to tackle that. 571 00:30:18,483 --> 00:30:20,351 That would be tremendously helpful. 572 00:30:20,351 --> 00:30:22,253 And so you can kind of map that out. 573 00:30:22,253 --> 00:30:23,721 Here's what my job title says. 574 00:30:23,721 --> 00:30:27,892 But here's some low lying fruit that makes sense that I can accomplish easily. 575 00:30:28,126 --> 00:30:30,495 But move our group forward. And so I like that book. 576 00:30:30,495 --> 00:30:33,498 It's just filled with very like, straightforward advice, 577 00:30:33,531 --> 00:30:36,901 nothing HPC related, but just kind of how to get in there quick 578 00:30:36,901 --> 00:30:38,336 and get your hands dirty. 579 00:30:38,336 --> 00:30:42,340 Jason, you had so many advices for me, for our audience. 580 00:30:42,674 --> 00:30:44,943 Thank you very much. Thanks. I really appreciate it. 581 00:30:44,943 --> 00:30:45,343 Thank you.