CentOS Stream & Clear Linux Achieve Greater Performance On 4th Gen Xeon Scalable Sapphire Rapids, EPYC Genoa

Written by Michael Larabel in Operating Systems on 3 February 2023 at 08:48 AM EST. Page 2 of 6. 9 Comments.

ONNX Runtime benchmark with settings of Model: yolov4, Device: CPU, Executor: Standard. CentOS Stream 9: Xeon 8490H 2P was the fastest.

ONNX Runtime benchmark with settings of Model: super-resolution-10, Device: CPU, Executor: Standard. CentOS Stream 9: Xeon 8490H 2P was the fastest.

For Microsoft's ONNX Runtime for AI, CentOS Stream 9 was performing very well and generally coming ahead of the others on both the AMD Genoa and Intel Sapphire Rapids servers.

The CPU power consumption wasn't all that different during the ONNX testing.

LeelaChessZero benchmark with settings of Backend: Eigen. Clear Linux: Xeon 8490H 2P was the fastest.

Intel's clear Linux distribution picked up its first win on Sapphire Rapids when it comes to the AI-driven Leela Chess Zero engine.

OpenVKL benchmark with settings of Benchmark: vklBenchmark ISPC. Ubuntu 22.04.1 LTS - perf: EPYC 9654 2P was the fastest.

Intel Open Image Denoise benchmark with settings of Run: RT.hdr_alb_nrm.3840x2160. Clear Linux: Xeon 8490H 2P was the fastest.

When it comes to the Ubuntu results, the benefits of the CPU frequency scaling "performance" governor continue to be quite clear and is unfortunate that Ubuntu Server (and a number of other prominent Linux desktop/server distributions) aren't defaulting to it on such hardware platforms.

OSPRay benchmark with settings of Benchmark: particle_volume/ao/real_time. Ubuntu 22.04.1 LTS: EPYC 9654 2P was the fastest.

In some workloads the AMD schedutil vs. performance governors difference isn't measurable but the default Intel P-State powersave versus performance governors is much more pronounced.

9 Comments - Next Page

Page: 1 2 3 4 5 6 Next Page