Intel has announced that it is bringing breakthrough memory bandwidth and performance to HPC and AI with its new lineup of Intel Max Series products, the Intel Xeon CPU Max Series (code-named Sapphire Rapids HBM) and Intel Data Center GPU Max Series (code-named Ponte Vecchio). The former is the world’s first (and only) x86-based processor with high bandwidth memory, accelerating HPC workloads without need for code changes, while the latter is Intel’s highest density processor, featuring over 100 billion transistors and up to 128 GB of high bandwidth memory. Both will be featured in the Aurora supercomputer, which is currently under construction at the Argonne National Laboratory and expected to boast over 2 exaflops of peak double-precision compute performance.
Intel Xeon Max CPU highlights:
- 68% less power usage than an AMD Milan-X cluster for the same HPCG performance.
- AMX extensions boost AI performance and deliver 8x peak throughput over AVX-512 for INT8 with INT32 accumulation operations.
- Provides flexibility to run in different HBM and DDR memory configurations.
- Workload benchmarks:
- Climate modeling: 2.4x faster than AMD Milan-X on MPAS-A using only HBM.
- Molecular dynamics: On DeePMD, 2.8x performance improvement against competing products with DDR memory.
Intel Max Series GPU highlights:
- 408MB of L2 cache – the highest in the industry – and 64MB of L1 cache to increase throughput and performance.
- The only HPC/AI GPU with native ray tracing acceleration, designed to speed scientific visualization and animation.
- Workload benchmarks:
- Finance: 2.4x performance gain over NVIDIA’s A100 on Riskfuel credit option pricing.
- Physics: 1.5x improvement over A100 for NekRS virtual reactor simulations.
Max Series GPUs will be available in various form factors:
- Max Series 1100 GPU: A 300-watt double-wide PCIe card with 56 Xe cores and 48GB of HBM2e memory. Multiple cards can be connected via Intel Xe Link bridges.
- Max Series 1350 GPU: A 450-watt OAM module with 112 Xe cores and 96GB of HBM.
- Max Series 1550 GPU: Intel’s maximum performance 600-watt OAM module with 128 Xe cores and 128GB of HBM.
“To ensure no HPC workload is left behind, we need a solution that maximizes bandwidth, maximizes compute, maximizes developer productivity and ultimately maximizes impact. The Intel Max Series product family brings high bandwidth memory to the broader market, along with oneAPI, making it easy to share code between CPUs and GPUs and solve the world’s biggest challenges faster.”
From an Intel press release:
The Max Series products are slated to launch in January 2023. Executing on its commitments to customers, Intel is shipping blades with Max Series GPUs to Argonne National Laboratory to power the Aurora supercomputer and will deliver Xeon Max CPUs to Los Alamos National Laboratory, Kyoto University and other supercomputing sites.
The Xeon Max CPU offers up to 56 performance cores constructed of four tiles and connected using Intel’s embedded multi-die interconnect bridge (EMIB) technology, in a 350-watt envelope. Xeon Max CPUs contain 64GB of high bandwidth in-package memory, as well as PCI Express 5.0 and CXL1.1 I/O. Xeon Max CPUs will provide more than 1GB of high bandwidth memory (HBM) capacity per core, enough to fit most common HPC workloads. The Max Series CPU provides up to 4.8x better performance compared to competition on real-world HPC workloads.
Max Series GPUs deliver up to 128 Xe-HPC cores, the new foundational architecture targeted at the most demanding computing workloads.