AMD EPYC 8534P / EPYC 8534PN Benchmarks - Siena Delivers Incredible Value & Energy Efficiency For Linux Servers

Written by Michael Larabel in Processors on 29 November 2023 at 04:20 PM EST. Page 5 of 6. 9 Comments.
OpenSSL benchmark with settings of Algorithm: SHA256. EPYC 8534P - 225W Power was the fastest.
OpenSSL benchmark with settings of Algorithm: SHA512. EPYC 8534P - 225W Power was the fastest.

Outside of database workloads, the EPYC Siena processors were running great for OpenSSL that needs no introduction.

OpenSSL benchmark with settings of Algorithm: AES-256-GCM. Xeon Platinum 8468 - 6c RAM was the fastest.
OpenSSL benchmark with settings of Algorithm: AES-256-GCM. Xeon Platinum 8468 - 6c RAM was the fastest.
OpenSSL benchmark with settings of Algorithm: AES-256-GCM. Xeon Platinum 8468 - 6c RAM was the fastest.
OpenSSL benchmark with settings of Algorithm: AES-256-GCM. Xeon Platinum 8468 - 6c RAM was the fastest.

The Sapphire Rapids server was doing well for AES-256-GCM until looking at the power metrics and pricing.

OpenSSL benchmark with settings of Algorithm: ChaCha20-Poly1305. EPYC 8534P - 225W Power was the fastest.

For other algorithms the tested Siena CPUs were leading in raw performance and needless to say for power efficiency and value.

Timed Linux Kernel Compilation benchmark with settings of Build: defconfig. Xeon Platinum 8468 was the fastest.

For code compilation workloads it depends upon the size of the code-base and number of files for how much of a difference the higher core/thread counts benefit. In the case of a default configuration of the Linux kernel, the Xeon Platinum 8458 processor led over the EPYC 8534P series.

Timed Linux Kernel Compilation benchmark with settings of Build: allmodconfig. EPYC 8534P - 225W Power was the fastest.

But for a Linux kernel build with all of the kernel modules included, the EPYC 8534P series in power determinism mode were shaving 20 seconds off the build times compared to the Xeon Sapphire Rapids processor.

Timed Linux Kernel Compilation benchmark with settings of Build: allmodconfig. EPYC 8534P - 225W Power was the fastest.

Even in the EPYC power determinism mode, the AMD EPYC CPUs were consuming much less power.

Timed LLVM Compilation benchmark with settings of Build System: Ninja. EPYC 8534P - 225W Power was the fastest.
Timed Node.js Compilation benchmark with settings of Time To Compile. EPYC 8534P - 225W Power was the fastest.

A look at the build times for some of the other popular open-source software projects.

OpenRadioss benchmark with settings of Model: Chrysler Neon 1M. Xeon Platinum 8468 was the fastest.

With Altair's OpenRadioss software and using the Chrysler Neon one million model, the extra two DDR5 memory channels with the Xeon Scalable Sapphire Rapids processor was of significant benefit and allowed the Xeon Platinum 8468 to come out ahead of the EPYC processors tested.

OpenRadioss benchmark with settings of Model: Chrysler Neon 1M. Xeon Platinum 8468 was the fastest.

What was interesting to see here is that the minimum power consumption recorded of the Xeon Platinum 8468 was higher than the peak power consumption found with the EPYC 8534P(N) processors.

Blender benchmark with settings of Blend File: Barbershop, Compute: CPU-Only. EPYC 8534P - 225W Power was the fastest.
Blender benchmark with settings of Blend File: Barbershop, Compute: CPU-Only. EPYC 8534P - 225W Power was the fastest.
Blender benchmark with settings of Blend File: Barbershop, Compute: CPU-Only. EPYC 8534P - 225W Power was the fastest.

Siena also is very interesting for its robustness in range of configurable TDP and performance/power determinism modes if really wanting to maximize value.

LuxCoreRender benchmark with settings of Scene: Orange Juice, Acceleration: CPU. EPYC 8534P - 225W Power was the fastest.
Chaos Group V-RAY benchmark with settings of Mode: CPU. EPYC 8534P - 225W Power was the fastest.

For those dealing with a lot of video encoding/transcoding, the AMD EPYC 8004 series performance was great not only for raw performance but really shine when looking at the power efficiency and TCO savings.


Related Articles