SMT Proves Worthwhile Option For 128-Core AMD EPYC "Bergamo" CPUs

Written by Michael Larabel in Processors on 20 July 2023 at 10:57 AM EDT. Page 6 of 8. 22 Comments.

And then the workloads that performed worse or about the same when SMT was enabled, which tended to be those workloads having reliant on system memory bandwidth.

NAS Parallel Benchmarks benchmark with settings of Test / Class: BT.C. EPYC 9754 2P: SMT Off was the fastest.

NAS Parallel Benchmarks benchmark with settings of Test / Class: EP.D. EPYC 9754 2P: SMT Off was the fastest.

NAS Parallel Benchmarks benchmark with settings of Test / Class: FT.C. EPYC 9754 2P: SMT Off was the fastest.

NAS Parallel Benchmarks benchmark with settings of Test / Class: LU.C. EPYC 9754 2P: SMT Off was the fastest.

NAS Parallel Benchmarks benchmark with settings of Test / Class: MG.C. EPYC 9754 2P: SMT Off was the fastest.

NAS Parallel Benchmarks benchmark with settings of Test / Class: SP.C. EPYC 9754 2P: SMT Off was the fastest.

Most of the NPB HPC benchmarks saw small gains in performance when the EPYC 9754 CPUs were running with SMT off.

CloverLeaf benchmark with settings of Lagrangian-Eulerian Hydrodynamics. EPYC 9754 1P: SMT Off was the fastest.

SPECFEM3D benchmark with settings of Model: Mount St. Helens. EPYC 9754 2P: SMT Off was the fastest.

SPECFEM3D benchmark with settings of Model: Tomographic Model. EPYC 9754 2P: SMT Off was the fastest.

SPECFEM3D benchmark with settings of Model: Homogeneous Halfspace. EPYC 9754 2P: SMT Off was the fastest.

SPECFEM3D benchmark with settings of Model: Water-layered Halfspace. EPYC 9754 2P: SMT Off was the fastest.

A number of the usual HPC benchmarks also benefited from running the EPYC 9754 with SMT disabled.

srsRAN

The srsRAN Project 5G software saw a huge hit to performance with SMT enabled.

Timed Godot Game Engine Compilation benchmark with settings of Time To Compile. EPYC 9754 2P: SMT Off was the fastest.

Timed Linux Kernel Compilation benchmark with settings of Build: defconfig. EPYC 9754 2P: SMT Off was the fastest.

Timed Linux Kernel Compilation benchmark with settings of Build: allmodconfig. EPYC 9754 2P: SMT Off was the fastest.

Timed LLVM Compilation benchmark with settings of Build System: Ninja. EPYC 9754 2P: SMT Off was the fastest.

Timed Node.js Compilation benchmark with settings of Time To Compile. EPYC 9754 2P: SMT On was the fastest.

For those considering Bergamo for build farms and compiling one project at a time, disabling SMT tended to provide better performance across a range of open-source projects tested.

22 Comments - Next Page

Page: 1 2 3 4 5 6 7 8 Next Page