https://www.servethehome.com/intel-announces-it-is-ending-traditional-hpc-platforms/ Refresh (1231 sec) * Forums * AI * Server + Server Systems + Server CPUs + Accelerators + Server Motherboards + Server Chassis + Other Components * 5G Edge * Storage * Networking * Workstation + Workstation Processors + Workstation Motherboards * Software + Operating Systems + Server Applications + Virtualization * Guides + Buyer's Guides + Tips + Top Hardware Components for TrueNAS / FreeNAS NAS Servers + Top Hardware Components for pfSense Appliances + Top Hardware Components for napp-it and Solarish NAS Servers + Top Picks for Windows Server 2016 Essentials Hardware + The DIY WordPress Hosting Server Hardware Guide Search [ ] Friday, March 3, 2023 * News * Storage Reliability + Raid Calculator + RAID Reliability Calculator | Simple MTTDL Model * Contact * About + Editorial and Copyright Policies * Subscribe * STH Forums Sign in Welcome! Log into your account [ ]your username [ ]your password [Login] Forgot your password? Get help Privacy Policy Password recovery Recover your password [ ]your email [Send My Password] A password will be e-mailed to you. STH ServeTheHome Advertisement [INS::INS] STH STH * Forums * AI * Server + Server Systems + Server CPUs + Accelerators + Server Motherboards + Server Chassis + Other Components * 5G Edge * Storage * Networking * Workstation + Workstation Processors + Workstation Motherboards * Software + Operating Systems + Server Applications + Virtualization * Guides + Buyer's Guides + Tips + Top Hardware Components for TrueNAS / FreeNAS NAS Servers + Top Hardware Components for pfSense Appliances + Top Hardware Components for napp-it and Solarish NAS Servers + Top Picks for Windows Server 2016 Essentials Hardware + The DIY WordPress Hosting Server Hardware Guide [ ][Search] Home Server Accelerators Intel Announces it is 3 Years Behind AMD and NVIDIA in XPU... * Server * Accelerators Intel Announces it is 3 Years Behind AMD and NVIDIA in XPU HPC By Patrick Kennedy - March 3, 2023 0 Facebook Twitter Pinterest Linkedin Email Print Intel SC22 Intel Falcon Shores OverviewIntel SC22 Intel Falcon Shores Overview This is a Friday afternoon release if I have ever seen one. In a news brief today, Intel made major changes to its HPC efforts until at least 2026. Or more succinctly, Intel has announced that the current Xeon Max, based on Sapphire Rapids with HBM, and its Ponte Vecchio OAM GPU are effectively the last of that legacy "pre-XPU" HPC platform architecture (Please see the update). Intel Announces it is 3 Years Behind AMD and NVIDIA in XPU HPC Intel did not say the above, or put it in a new roadmap, but parsing the announcements: 1. Rialto Bridge is discontinued 2. The next Max Series XPU/GPU will be Falcon Shores in 2025 What this effectively means is that Intel will not have a new GPU offering between now 2023 and 2025. Update: Intel contacted us to clarify its roadmap. Falcon Shores 2025 will be a GPU architecture, not with CPU cores. Intel told us that the vision for Falcon Shores is still XPU, but 2025 Falcon Shores is GPU focused. The "traditional" part is a GPU-only offering, Falcon Shores will offer different IP blocks as part of an overall XPU offering, but that will not be the 2025 generation. In 2025, we get Falcon Shores which was supposed to co-package CPU cores, GPU cores, memory, and other IP, potentially other accelerators, and even Lightbender silicon photonics. Intel has also shown plans for CPU tile only with co-packaged memory Falcon Shores. Intel SC22 Intel Falcon Shores XPU 2Intel SC22 Intel Falcon Shores XPU 2 Taking stock of what that means for the traditional HPC architecture of CPUs sitting in sockets attached to accelerators like GPUs, that model is effectively discontinued, except at Intel. AMD Instinct MI300 (2023) will have CPU, GPU, and memory co-packaged. Dr Lisa Su AMD ISSCC 2023 AMD Instinct MI300Dr Lisa Su AMD ISSCC 2023 AMD Instinct MI300 NVIDIA Grace Hopper (2023) will have CPU, GPU, and memory co-packaged and the ability to have CPU-only packages. NVIDIA Computex 2022 HGX Grace HGX Grace HopperNVIDIA Computex 2022 HGX Grace HGX Grace Hopper Intel's next-gen HPC part Falcon Shores (2025) was supposed to have CPU, GPU, and memory co-packaged. Intel SC22 Intel Falcon Shores OverviewIntel SC22 Intel Falcon Shores Overview It still may, but not at Intel by 2025. The subtext to today's announcement is that Intel and its customers see the need for this co-packaged architecture, to the point where Rialto Bridge was not set up to be a big winner. Intel is instead focusing on GPU-only Falcon Shores for 2025. That means until 2026 NVIDIA and AMD are ahead in that next-gen architecture. For those wondering about the benefits of co-packaging memory, CPU, and GPU, it is the same that is occurring with the Apple M1/ M2 in many ways. Apple M2 Pro Package And 32GB Of MemoryApple M2 Pro Package And 32GB Of Memory Moving data from CPU memory to GPU memory and vice versa requires potentially duplicating data leading to more wasted memory, but more importantly, is the power. Last week we featured AMD Talks Stacking Compute and DRAM at ISSCC 2023. The power used to move data between sockets and even between CPU sockets and DIMMs while also increasing speeds is a major challenge. Dr Lisa Su AMD ISSCC 2023 Energy Efficiency Of IO InterconnectDr Lisa Su AMD ISSCC 2023 Energy Efficiency Of IO Interconnect As a result, co-packaging CPU, GPU, accelerators, and memory using more advanced methods is seen as a way to increase interconnect speeds at a different power curve than traditional separate components. Dr Lisa Su AMD ISSCC 2023 Bits Per JouleDr Lisa Su AMD ISSCC 2023 Bits Per Joule In large-scale HPC systems, a huge amount of power is used simply moving data around within nodes, and externally. The reason we have AMD, NVIDIA, and now Intel saying next-gen HPC GPU architectures will have co-packaged components is to minimize this "wasted" power use. That in turn allows for more of the power budget to be used for doing computation, rather than moving data around. Final Words The HPC market is sizable, but it is also only a portion of the overall computing market. We still expect there to be PCIe and OAM GPUs and so forth for years to come. It just seems like this trend is accelerating for the HPC space to the point that Intel is going to skip a generation. That will save development resources, but it also means that Intel will be using Ponte Vecchio to compete against the NVIDIA H100, NVIDIA Grace-Hopper, and AMD Instinct MI300 for the next two years until Falcon Shores is hopefully ready in 2025. This announcement also means that it is planning its 2025 Falcon Shores XPU to really be a GPU with the possibility of having customized IP blocks in the future. Intel Aurora Sapphire Rapids And Ponte Vecchio HPE Cray EX Node ISC 2022 2x SPRIntel Aurora Sapphire Rapids And Ponte Vecchio HPE Cray EX Node ISC 2022 2x SPR In addition, Intel announced that it is discontinuing Lancaster Sound, the next-gen Flex series GPU, making the next-generation of record Melville Sound. [INS::INS] * TAGS * amd * CPU Max * GPU Max * Grace * H100 * Hopper * HPC * Instinct * Intel * MI300 * NVIDIA * Ponte Vecchio Previous articleASRock Rack W680D4U-2L2T/G5 Motherboard Review An Intel Core Server Platform Patrick Kennedy https://www.servethehome.com Patrick has been running STH since 2009 and covers a wide variety of SME, SMB, and SOHO IT topics. Patrick is a consultant in the technology industry and has worked with numerous large hardware and storage vendors in the Silicon Valley. The goal of STH is simply to help users find some information about server, storage and networking, building blocks. If you have any helpful information please feel free to post on the forums. RELATED ARTICLESMORE FROM AUTHOR Dr Lisa Su AMD ISSCC 2023 Tighter Integration Of Compute And Memory News AMD Talks Stacking Compute and DRAM at ISSCC 2023 Dynatron L32 1U Liquid Cooler For AMD EPYC Genoa 5 Other Components Dynatron L32 SP5 1U Liquid Cooling for AMD EPYC Genoa and Bergamo AMD EPYC 9004 Genoa De Lidded 4 Server CPUs AMD EPYC 9004 Genoa Under-the-Lid LEAVE A REPLY [ ] [ ] [ ] [ ] [ ] [ ] [ ] [ ] Please enter your comment! [ ] Please enter your name here [ ] You have entered an incorrect email address! Please enter your email address here [ ] [ ]Save my name, email, and website in this browser for the next time I comment. [ ] [ ]Sign me up for the STH newsletter! [Post Comment] [ ] [ ] [ ] [ ] [ ] [ ] [ ] D[ ] This site uses Akismet to reduce spam. Learn how your comment data is processed. [INS::INS] [INS::INS] Newsletter Get the best of STH delivered weekly to your inbox. We are going to curate a selection of the best posts from STH each week and deliver them directly to you. Your email address: [ ] [Sign up] By opting-in you agree to have us send you our newsletter. We are using a third party service to manage subscriptions so you can unsubscribe at any time. Leave this field empty if you're human: [ ] STH ABOUT US ServeTheHome is the IT professional's guide to servers, storage, networking, and high-end workstation hardware, plus great open source projects. Advertise on STH DISCLAIMERS: We are a participant in the Amazon Services LLC Associates Program, an affiliate advertising program designed to provide a means for us to earn fees by linking to Amazon.com and affiliated sites. FOLLOW US (c) 2009-2023 Loyolan Ventures, LLC