BIOS Optimizations For AMD 5th Gen EPYC Yield Greater HPC Performance & Power Efficiency

Written by Michael Larabel in Processors on 24 December 2024 at 10:30 AM EST. Page 4 of 5. 4 Comments.

👁 GPAW benchmark with settings of Input: Carbon Nanotube. HPC Tuning Recommendations was the fastest.

The raw performance of GPAW wasn't impacted from the tuning guide but the AMD EPYC server power efficiency came out nicely.

👁 HeFFTe - Highly Efficient FFT for Exascale benchmark with settings of Test: r2c, Backend: FFTW, Precision: float, X Y Z: 128. HPC Tuning Recommendations was the fastest.

👁 HeFFTe - Highly Efficient FFT for Exascale benchmark with settings of Test: r2c, Backend: FFTW, Precision: float, X Y Z: 256. HPC Tuning Recommendations was the fastest.

👁 HeFFTe - Highly Efficient FFT for Exascale benchmark with settings of Test: c2c, Backend: FFTW, Precision: float, X Y Z: 128. HPC Tuning Recommendations was the fastest.

👁 HeFFTe - Highly Efficient FFT for Exascale benchmark with settings of Test: c2c, Backend: FFTW, Precision: float, X Y Z: 256. HPC Tuning Recommendations was the fastest.

With the HeFFTe software for efficient FFT for exascale computing there were some nice performance wins spotted when following the AMD BIOS tuning guide for EPYC 9005 Turin.

👁 XNNPACK benchmark with settings of Model: FP32MobileNetV1. HPC Tuning Recommendations was the fastest.

👁 XNNPACK benchmark with settings of Model: FP32MobileNetV2. HPC Tuning Recommendations was the fastest.

👁 XNNPACK benchmark with settings of Model: FP32MobileNetV3Large. HPC Tuning Recommendations was the fastest.

👁 XNNPACK benchmark with settings of Model: FP32MobileNetV3Small. HPC Tuning Recommendations was the fastest.

👁 XNNPACK benchmark with settings of Model: FP16MobileNetV2. HPC Tuning Recommendations was the fastest.

👁 XNNPACK benchmark with settings of Model: FP16MobileNetV3Large. HPC Tuning Recommendations was the fastest.

👁 XNNPACK benchmark with settings of Model: QS8MobileNetV2. HPC Tuning Recommendations was the fastest.

The XNNPACK software as Google's library for machine learning frameworks saw some very nice wins with the HPC tuning BIOS recommendations from AMD.

👁 libxsmm benchmark with settings of M N K: 256. HPC Tuning Recommendations was the fastest.

The libxsmm library for specialized dense and sparse matrix operations also saw some nice gains from AMD's BIOS tuning guide for HPC.

👁 NWChem benchmark with settings of Input: C240 Buckyball. Default was the fastest.

The NWChem chemistry software benefited from greater power efficiency on EPYC Turin with the AMD public BIOS tuning guide for high performance computing server deployments.

4 Comments - Next Page

Page: 1 2 3 4 5 Next Page

URL: https://www.phoronix.com/review/amd-epyc-9005-hpc-tuning/4

⇱ BIOS Optimizations For AMD 5th Gen EPYC Yield Greater HPC Performance & Power Efficiency - Phoronix

BIOS Optimizations For AMD 5th Gen EPYC Yield Greater HPC Performance & Power Efficiency