The Incredible Evolution Of AMD EPYC HPC Performance Shown In The Azure Cloud

Written by Michael Larabel in Cloud on 13 November 2025 at 10:44 AM EST. Page 3 of 5. 11 Comments.

👁 HeFFTe - Highly Efficient FFT for Exascale benchmark with settings of Performance Per Core, Test: c2c, Backend: FFTW, Precision: float, X Y Z: 512. HBv4 was the fastest.

👁 HeFFTe - Highly Efficient FFT for Exascale benchmark with settings of Performance Per Core, Test: r2c, Backend: FFTW, Precision: float, X Y Z: 512. HBv4 was the fastest.

👁 HeFFTe - Highly Efficient FFT for Exascale benchmark with settings of Performance Per Core, Test: c2c, Backend: FFTW, Precision: double, X Y Z: 256. HBv4 was the fastest.

👁 HeFFTe - Highly Efficient FFT for Exascale benchmark with settings of Performance Per Core, Test: c2c, Backend: FFTW, Precision: double, X Y Z: 512. HBv5 was the fastest.

It's very fascinating to see how far the AMD EPYC HPC performance and Azure HB-class VMs have come in just over a half decade. Or for AMD EPYC itself there is also the newer Turin core design that is beneficial to the workloads not fully leveraging the HBM3 memory bandwidth. Again, Turin comparison points will be in its own dedicated article

👁 Palabos benchmark with settings of Performance Per Core, Grid Size: 1000. HBv5 was the fastest.

👁 OpenFOAM benchmark with settings of Performance Per Core, Input: drivaerFastback, Large Mesh Size, Execution Time. HBv5 was the fastest.

The latest Azure HBv5 VMs with HBm3 memory can lead to dramatic time savings for workloads like computational fluid dynamics (CFD) with OpenFOAM, especially if you are still relying on any AMD EPYC Rome or Milan class processors.

11 Comments - Next Page

Page: 1 2 3 4 5 Next Page

URL: https://www.phoronix.com/review/amd-epyc-azure-hbv2-hbv5/3

⇱ The Incredible Evolution Of AMD EPYC HPC Performance Shown In The Azure Cloud - Phoronix

The Incredible Evolution Of AMD EPYC HPC Performance Shown In The Azure Cloud