- Intel has officially introduced its upcoming Sapphire Rapids CPU, which is included in the Xeon Max series.
- Alongside the new Xeon Max CPU, the company also announced its upcoming GPUs that offer up to 128 cores and native ray tracing acceleration.
- Both the CPU and the GPUs will be used in various high-performance computing applications, starting with some supercomputers in laboratories and universities.
Some details of Intel‘s Xeon Sapphire Rapids CPU were leaked online in April this year and they were expected to be available in the third quarter. However, they were not ready and delayed. Now, the chip giant officially introduces its Xeon Max Sapphire Rapids CPU and GPUs, and they will be available in January 2023.
Up to 56 cores
Intel Xeon Max Sapphire Rapids CPU, which is designed to be used on HPC (High-Performance Computing) applications, offers 56 cores and 64 GB of high bandwidth HBM2e memory in the package. It delivers PCI Express 5.0 and CX 1.1 I/O support as well as new AMX extensions to boost the performance of some AI-related tasks. Intel claims that it can deliver 2.4x performance of AMD Milan-X CPU on MPAS-A using only HBM memory. This CPU will be used in Los Alamos National Laboratory, Kyoto University, and some other supercomputing sites.
Intel Xeon Max series GPUs can offer up to 128 X-HPC cores, coming in three different options as follows:
- Max Series 1100 GPU: A 300-watt double-wide PCIe card with 56 Xe cores and 48 GB 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 96 GB of HBM.
- Max Series 1550 GPU: Intel’s maximum performance 600-watt OAM module with 128 Xe cores and 128 GB of HBM.
Intel states that those are currently the only HPC/AI GPUs with native ray tracing acceleration support. The company claims a 2.4x performance gain over Nvidia A100 on Riskfuel credit option pricing. Those GPUs will power Argonne National Laboratory’s Aurora Supercomputer.