Gadgetory

64 cores 256 threads running at 1.3 to 1.5 gigahertz with 32 megabytes of cache meet the Knights landing xeon phi one of the most insane CPUs in the world how insane well how does a theoretical two and a half tariff lock sound got no context for that don't stress it's like it's like a lot like like like look how big this thing is I could take over the world with this sort of power but first I need to pay the bills which is why today's video is brought to you by tunnel bear tunnel where makes it easy to privately and securely browse a more open Internet tried tunnel bear for free a tunnel bear comm /l TT so some background information is in order here as this is no ordinary CPU Knights landing is the current generation of Xeon Phi Intel's version of what's referred to as a mini core processor which is exactly what it sounds like it's built around the idea of trading off per thread performance for an obscene number of threads and maximum throughput with the theory being that a simpler scaled-down core design can dramatically scale up in number while still utilizing existing tools so we can actually trace Knights landings lineage back to project Larrabee first unveiled in mid 2008 as a GPU designed around the x86 architecture Intel even announced their intent to release a consumer graphics card version of it back in 2010 but since nobody is talking about dedicated team blue graphics cards today you should probably realize that that never launched so zeal and Phi is what rose from the ashes of that project first appearing in 2012 as a PCI Express add-in card codename Knights corner and it actually did end up getting used in what was the world's fastest supercomputer until June 2016 the newest revision though Knights landing looks a little different it launched last year and unlike Knights corner it works exclusively in the LGA 3647 socket that high-end Xeon Perley CPUs also slot into so adding support for avx-512 and its many extensions gives Knights landing a lot of flexibility in its target market so then who's the target market I'm glad you asked uses for this include protein folding simulation weather prediction AI and neural network research and development and molecular simulation with libraries like tensor flow and the UI you see Zen AMD being able to take advantage of it as well as G you acceleration okay then - so why do you have one I guess because Super Micro pulled a rookie mistake and sent it to me I mean obviously all I was gonna do was game on it and I'm actually kidding about that at one point three gigahertz I can tell you guys right now without firing up a single benchmark what the gaming performance will look like like that as for productivity though yeah right out of the gate it fell flat on its face coming in at under half of korah nines and nearly a third of thread rippers score in 7-zip and Cinebench cpu mark and reel bench not looking much better even blender a conventionally multi-core friendly benchmark sees terrible performance given the 19 hundred dollar price tag for the chip alone so know these results won't do it all xeon phi is clearly unlike anything we've ever tried to benchmark before you'd need like a PhD or something to properly evaluate this thing which is where dr. King horn our friend over at Puget systems comes in as a chemist and mathematician dr. King horn is no stranger to high-level computation like this and agreed to remote into our Knights landing workstation and do some of the testing for us while also running his own 14 core Xeon 2690 v4 workstation through the same suite to give us a point of comparison the results actually pretty surprising in Linpack the higher core clock of the traditional Xeon gives it an edge in smaller problem sizes but beyond 5,000 the tables turn dramatically with the molecular dynamics library NAMD things seem to scale predictably with the number of atoms being simulated and all the tests wind up being far less favorable to the 0 and 5 the greatest gains in this benchmark actually come from dedicated compute GPUs so that is fine for the phi these days since it sits in a socket instead of taking up valuable PCIe slots tensorflow neural networks bring things back though into the Xeon Phi's wheelhouse especially in batch sizes larger than 64 so Xeon Phi takes a significant lead over its traditional Xeon cousin but it should be noted that both of them get creamed by a GPU further highlighting the importance of large banks of compute DP use for tasks like these though that doesn't mean that Xeon Phi is useless it's dramatically better performance than a standard Xeon makes it great for supplementing GPU compute in workloads like these which brings us neatly to the conclusion then not quite as good as a GPU for tasks that can be accelerated far better than a traditional Xeon though in most and finally utterly worthless for just about anything on the Windows desktop I was actually surprised to see Windows even installed we ended up needing a server version of Windows to see all the course but that is sort of irrelevant anyway because this is intended for use in supercomputers we're like massive nodes of these things can all work together to solve complex problems along with the banks of GPUs that accompany of them I mean you're almost certainly not gonna be buying anything like this personally any time soon unless you're like that you like to science for funsies with that said though you yes you are very likely to benefit from the work being done using hardware like this and we're glad we got the opportunity to play with it it was fun speaking of things that are fun it's fun to tell you guys about Squarespace and thank them for sponsoring this video Squarespace has 24/7 support via live chat so anyone can create a beautiful functional website in just a few minutes if all you want is like a one-page online presence that's their cover pages feature and they've got tons of other great features as well including their logo designer the ability to publish in the Apple news format and not to mention that the whole creation tool is cloud-based so it's always up you can access it anywhere in updating your site is as simple as like dragging over some pictures clickety-clacking some text and you are ready to rock Squarespace starts at just 12 bucks a month and you can start a trial with no credit card required by heading to the link in the video description squarespace.com /l tt and when you decide to sign up for Squarespace forever you can use offer code LTT to get 10% off on your first purchase so thanks for watching guys if you dislike this video you can hit that button but if you liked it hit like get subscribed maybe consider checking out where to buy the stuff me featured if you like work at a university or something it's like I'll know that's what happened if like you know 2,000 Xeon Phi's show up on like our Amazon report or whatever anyway the point is also in the description is our merch store as well as our community forum which you shouldn't totally join