NVIDIA Vera HPC Performance Benchmarks Take On AMD EPYC, Intel Xeon
While most everyone understandably likes to talk about AI these days, one of the areas of NVIDIA's Vera CPU I have been most eager to test has been for conventional high performance computing (HPC) workloads. HPC was talked up as one of Vera's strengths thanks to its 1.2TB/s LPDDR5x memory bandwidth and custom Olympus cores. Vera is seeing deployments at HPC labs like Los Alamos. With recently…
NVIDIA's Vera CPU has been garnering attention for its performance in conventional high performance computing (HPC) workloads. One of Vera's key strengths is its 1.2TB/s LPDDR5x memory bandwidth and custom Olympus cores. Recent access to a dual NVIDIA Vera server allowed for testing various HPC workloads and performance benchmarks.
The dual Vera server boasted a combined 176 cores and 352 threads, compared to AMD EPYC processors and Intel Xeon processors in the lab. The AMD EPYC 9575F offered 128 cores and 256 threads, while the EPYC 9755 2P had 256 cores and 512 threads. Intel's Xeon 6 Granite Rapids configuration featured two flagship Xeon 6980P processors, delivering a combined 256 cores and 512 threads.
The dual NVIDIA Vera system was equipped with 16 x 96GB LPDDR5-9600 Samsung memory modules, while the AMD EPYC 9575F and EPYC 9755 2P models utilized 24 x 64GB DDR5-6400 Samsung memory modules. The Intel Xeon 6980P setup included 24 x 64GB MRDIMM-8800 Micron memory modules. All systems ran a modified Ubuntu 24.04 LTS stack with a patched Linux 6.17 based kernel and GCC 15.3 compiler, providing a similar configuration for each processor.
Testing measured CPU power consumption and peak CPU frequency in real-time during the HPC workload benchmarks.
Written by urgent.news from Phoronix's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
