Intel Introduces Xeon Max & Data Center GPU Max Series
The other half of Intel's announcement today is introducing the Data Center GPU Max Series. The Intel Data Center GPU Max Series is the branding for what to this point has been known as Ponte Vecchio.
Intel is confirming the Data Center GPU Max Series as offering up to 128 Xe HPC cores, 52 TF peak FP64 throughput, 16 Xe Links for GPU-to-GPU communication, up to 128GB of HBM2e memory, up to 408MB for its Rambo L2 cache, and up to a 64MB L1 cache.
Intel also talked up the Data Center GPU Max Series ray-tracing capabilities during the pre-brief.
Intel's performance claims for the Data Center GPU Max Series is ambitious while we'll see how the pricing and performance ultimately plays out.
With the Intel Data Center GPU Max 1100 product, the card is rated for a 300 Watt TDP and has 56 Xe cores and 48GB of HBM2e memory.
The higher-end Max Series 1350 and 1550 GPUs meanwhile are rated for 450 Watts with 112 Xe cores and 600 Watts with 128 Xe cores, respectively. The Intel Data Center GPU Max Series 1350/1550 are OAM modules.
Intel also teased their upcoming "Rialto Bridge" Data Center GPU that will feature up to 160 Xe HPC cores and up to a 800 Watt power draw per OAM and rely on liquid cooling.
And then there is also Intel's Falcon Shores XPU on the way too.
Those are the highlights for Intel's disclosures today ahead of AMD's EPYC Genoa launch and next week's SC2022 super computing conference. The specs and performance claims are exciting while sadly lacking is the pricing and availability -- hopefully we'll learn more positive news in January. Particularly for the Intel Xeon Max Series CPUs it will be quite exciting to see how well they perform in the lab relative to AMD's EPYC Genoa processors.
If you enjoyed this article consider joining Phoronix Premium to view this site ad-free, multi-page articles on a single page, and other benefits. PayPal or Stripe tips are also graciously accepted. Thanks for your support.