The First Xe-HPC Deployment: Aurora, with Xe Link

Some of this analysis is taken from our original Aurora Xe news post, but now with Xe Link information.

If you’ve skipped to this final page, it’s worth reading the last few. We’ve covered in detail what we can derive from Intel slides about the Xe architecture, the new Xe-HPC hardware paradigm, the Ponte Vecchio GPU in as much detail as we could figure out, and the new software strategy. All this page talks about is the major commercial deployment of Xe-HPC in a supercomputer: Aurora. If you want the juicy details on the hardware, and what we might see on the graphics side, take a read of the previous pages.

Aurora: Exascale in the USA

Part of the announcement at HPC DevCon was about the first deployment of Ponte Vecchio, to the new Aurora supercomputer for the Department of Energy at Argonne National Laboratories. Intel disclosed that it will be delivering Aurora in 2021, featuring two new future CPUs as well as six GPUs per node.

Intel has been working on the Aurora contract for a while now, and the project has changed scope over time due to market changes and hardware setbacks. Initially announced several years ago as a ‘deliver by 2020’ project between Argonne, Cray, and Intel, the hardware was set to be built around Intel’s Xeon Phi platform, providing high-throughput acceleration though Intel's AVX-512 instructions and the 10nm Knights Hill accelerator. This announcement was made before the recent revolution in AI acceleration, as well as Intel subsequently killing off the Xeon Phi platform after adding AVX-512 to its server processors (with the last breath Knights Mill receiving a very brief lifespan). Intel had to go back to the drawing board, and since Xeon Phi was scrapped, Intel’s audience has been questioning what it will bring to Aurora, and until today the company had only said that it would be built from a combination of Xeon CPUs and XeHPC GPUs.

The two CPUs will be ‘Sapphire Rapids’ Xeon CPUs, which if you follow Intel’s release schedule, is the CPU after the CPU after the CPU after the one currently on the market. It goes Skylake, Cascade Lake (currently out), Cooper Lake (2019/2020), Ice Lake (2H2020), Sapphire Rapids (2021). The promise is that these CPUs will be on 10nm, and judging by the diagram will have 8 channels of memory (likely DDR5, but not confirmed). Sapphire Rapids are predicted to be Intel’s first enterprise CPUs to support the new CXL standard, which would also mean PCIe 5.0 support at a minimum.

With the two CPUs are six Ponte Vecchio GPUs. Judging by the diagram, it must be pretty obvious that these are not PCIe cards, but likely to be OAM or SXM2 modules, or whatever the newest standard will be at the time. Each GPU will be able to communicate directly to each other via CXL, which is an interesting element to the discussion. On these diagrams is a small black chip in the middle, which Intel announced as its new Xe Link chip.

Much like NVIDIA’s NVSwitch, it looks like the Xe Link chip is there to mediate communications between all the GPUs in a node, and by what Intel says on this slide, between nodes as well. It would appear to also manage the unified memory of the system, and like the GPUs, it is CXL based. Part of the announcement on the links between the GPUs is that it should be an all-to-all design, despite the image above showing some GPUs communicating directly to each other and some only through the Link chip. We confirmed that this representation was inaccurate.

The CPUs will be Sapphire Rapids CPUs, Intel’s second generation of 10nm server processors coming after the Ice Lake Xeons. The announcement today reaffirmed that Sapphire Rapids is a 2021 processor; and likely a late 2021 processor, as the company also confirmed that Ice Lake will have its volume ramp through late 2020. Judging from Intel's images, Sapphire Rapids is set to have eight memory channels per processor, with enough I/O to connect to three GPUs. Within an Aurora node, two of these Sapphire Rapids CPUs will be paired together, and support the next generation of Intel Optane DC Persistent Memory (2nd Gen Optane DCPMM). We already know from other sources that Sapphire Rapids is likely to be DDR5 as well, although I don't believe Intel has said that outright at this point.

On the GPU side, all six of the GPUs per node will be Intel’s new 7nm Ponte Vecchio Xe GPU. 

Each Aurora node will have 8 fabric endpoints, giving plenty of topology connectivity options. And with the system being built in part by Cray, connecting the systems will be a version of their Slingshot networking architecture, which is also being used for the other early-2020s US supercomputers. Intel has stated that Slingshot will connect ~200 racks for Aurora, featuring a total of 10 petabytes of Memory and 230 petabytes of Storage.

Guessing Ponte Vecchio Performance

With this information, if we spitball some numbers on performance, here's what we end up with:

  • Going with Intel's number of 200 racks
  • Assume each rack is a standard 42U,
  • Each Aurora node is a standard 2U,
  • We know the system has 200 racks.
  • Take out 6U per rack for networking and support,
  • Take out 1/3 of the racks for storage and other systems
  • We get a rounded value of 2400 total Aurora nodes (2394 based on assumptions).

This means we get just south of 5000 Sapphire Rapids CPUs and 15000 Ponte Vecchio GPUs for the whole of Aurora. Now when calculating the total performance of a supercomputer, the CPU and Accelerator performance often depends on the workload required. If Ponte Vecchio is truly an exascale class GPU, then let’s assume that the GPUs are where the compute is. If we divide 1 ExaFLOPs by 15000 units, we’re looking at 66.6 TeraFLOPs per GPU. Current GPUs will do in the region of 14 TF on FP32, so we could assume that Intel is looking at a ~5x increase in per-GPU performance by 2021/2022 for HPC. Of course, this says nothing about power consumption, and if we were to do the same math at 4U per node, then it would be ~7500 GPUs, and we would need 135 TF per GPU to reach exascale.

Delivering The Hardware

Despite initially being a ‘deliver by 2020’ project, Intel is now saying that Aurora will be delivered in 2021. This means that Intel has to execute on the following:

  • Manufacturing at 10++ in sufficient yield with respect to cores/frequencies
  • Manufacturing at 7nm for chiplets, both in yield and frequencies
  • Transition through DDR3 to DDR4 (and DDR5?) in that time frame
  • Transition through PCIe 3.0 to PCIe 4.0 and PCIe 5.0 in that time frame
  • Release and detail its 2nd generation Optane DC Persistent Memory
  • Provide an SDK to deal with all of the above
  • Teach researchers to use it

On those last two points, Intel has stated that Ponte Vecchio and Aurora will be a primary beneficiary of the company’s new OneAPI SDK. This industry initiative, spearheaded by Intel, is designed to use a singular cross-architecture language called ‘Data Parallel C++’, based on C++ and SYCL, that can pull libraries designed to speak to various elements of Intel’s hardware chain. The idea is that software designs can write the code once, link appropriate libraries, and then cross-compile for different Intel hardware targets. In Intel’s parlance, this is the ‘xPU’ strategy, covering CPU, GPU, FPGA, and other accelerators. Aside from Argonne/Cray, Intel is citing Lenovo and Atos as key partners in the OneAPI strategy.

In order to achieve a single exascale machine, you need several things. First is hardware – the higher performance a single unit of hardware, the fewer you need and the less infrastructure you need. Second is infrastructure, and being able to support such a large machine. Third is the right problem that doesn’t fall afoul of Amdahl’s Law – something that is so embarrassingly parallel that can be spread across all the hardware in the system is a HPC scientist dream. Finally, you need money. Buckets of it.

In the past few years, the top supercomputers in the world have addressed all of these requirements by orders of magnitude: hardware is faster, infrastructure is readily available, traditionally serial problems are being re-architected to be parallel (or lower precision), and governments are spending more than ever for single supercomputers. A $5 million supercomputer used to get a research group into the top echelons of the Top 500 list. Now it’s more like $150+ million. The current leader, Summit, had an estimated build cost of $200 million. And Aurora will handily top that, with the deal valued at more than $500 million.

When I asked before the announcement if Argonne would be the first customer of Ponte Vecchio, the Intel executive heading the briefing dodged the question by instead answering his own, saying ‘we’re not disclosing other customers’. In all my time as a journalist, I don’t think I’ve ever had a question unanswered and dodged in such a way. Unfortunately I had no opportunity in the open Q&A session to follow up.

Ultimately, based on what we know so far, Intel still has a lot of work to do to deliver this by 2021. It may be the case that Aurora is that culmination of all its technologies, and that Intel would be prepared to use the opportunity to help smooth out any issues that might arise. At any rate, some of these technologies might not get in the hands of the rest of us until 2022.

Hopefully, when Aurora is up and running, we’ll get a chance to take a look. Watch this space.

oneAPI: Intel’s Solution to Software
Comments Locked

47 Comments

View All Comments

  • martinw - Tuesday, December 24, 2019 - link

    > we’re looking at 66.6 TeraFLOPs per GPU. Current GPUs will do in the region of 14 TF on FP32, so we could assume that Intel is looking at a ~5x increase in per-GPU performance by 2021/2022 for HPC.

    But HPC ExaFLOPs are traditionally measured using FP64, so that means a ~10x increase.
  • Santoval - Tuesday, December 24, 2019 - link

    If Intel manage to deliver ~67 TFLOPs of *double* precision in a single GPU package -even if it consists of multiple GPU chiplets- I will eat the hat I don't have. ~67 TFLOPs of single precision in a single GPU package might be possible (at a 480 - 500W TDP) due to Intel's new GPU design and its 7nm node, which should be quite more power efficient than their 10nm & 14nm nodes, assuming Intel can fab it at a tolerable yield that is.

    The use of Foveros and EMIB also reduce the power budget and increase performance further, because they alleviate I/O power draw and, along with that "Rambo cache", mitigate the memory bottleneck. The graphics memory will also probably be HBM3, with quite a higher performance and energy efficiency.

    So a ~5x performance at roughly x2 the TDP of the RTX 2080 Ti might be doable. It is ~2.5 times the performance per watt, which is high but not excessive. To double that performance further though is impossible. Intel are a semiconductor company, they are not wizards.
  • nft76 - Wednesday, December 25, 2019 - link

    I'm guessing the number of nodes (and GPUs) will be at least two, probably more like three to four times larger than estimated in the article. I'm guessing that the ~200 racks is without storage included and there will be more nodes per rack. If I'm not mistaken, Cray Shasta high-density racks are much larger than standard.
  • eastcoast_pete - Tuesday, December 24, 2019 - link

    Thanks Ian, Happy Holidays to All at AT and here in "Comments"!
    My first thought was, boy, that lower case/upper case in oneAPI is really necessary; reading the subheading, I almost thought it's about an unusual Irish name (O'NEAPI), w/o the apostrophe.
    On a more serious note, this also shows how important the programming ecosystem is; IMO, a key reason why NVIDIA remains the market leader in graphics and HPC.
  • UltraWide - Tuesday, December 24, 2019 - link

    Nvidia recognized this more than 10 years ago, everyone else is playing catch up.
  • JayNor - Tuesday, December 24, 2019 - link

    Intel is extending Sycl for FPGA config using data flow pipes. They've mentioned previously that Agilex will have the first implementation of pcie5 and CXL. Perhaps OneAPI will do something to simplify FPGA design.

    https://github.com/intel/llvm/blob/sycl/sycl/doc/e...
  • JayNor - Tuesday, December 24, 2019 - link

    Intel's current NNP chips don't have PCIE5 or CXL, and I recall some discussion about it being a feature that the NNP-I chips did manual memory management.

    Is Intel enthusiastically pushing shared memory for the GPU high performance programming, or is this just a convenience during development to get CPU solutions working on GPU quickly?
  • ksec - Tuesday, December 24, 2019 - link

    >The promise with Xe-HPC is a 40x increate in FP64 compute performance.

    Increase

    One or two other spelling mistakes but I can no longer find it.

    >The CPUs will be Sapphire Rapids CPUs, Intel’s second generation of 10nm server processors coming after the Ice Lake Xeons

    First time I heard SR will be an 10nm++ CPU, always thought it was destined for 7nm. Possibly Another Roadmap shift.

    Other than that Great Article. But as with anything Recent Intel, I will believe it when I see it. They are ( intentionally or not ) leaking lots of benchmarks and roadmaps, and lots more "communication" on the ( far ) future as some sort of distraction against AMD.

    I have my doubt on their GPU Drivers, not entirely sure their 10nm yield and cost could compete against NV and AMD without lowering margin. But at least in terms of GPGPU it will bring some competition to Nvidia's ridiculously expensive solution.
  • Alexvrb - Tuesday, December 24, 2019 - link

    Yeah they'll probably be more competitive in HPC in the short term. For gaming... we'll see. I suspect they'll get murdered in the short term unless they are really aggressive with pricing. If they go this route most likely they'll do CPU+GPU bundle deals with OEMs to force their way into the "gaming" market.
  • Spunjji - Friday, December 27, 2019 - link

    That approach seems highly likely.

Log in

Don't have an account? Sign up now