Amazon's New EC2 M7a AMD EPYC "Genoa" Instances Deliver Leading Performance In The AWS Cloud

Written by Michael Larabel in Processors on 17 August 2023 at 11:26 AM EDT. Page 3 of 6. 3 Comments.
LeelaChessZero benchmark with settings of Backend: BLAS. m7i.16xlarge Intel SPR was the fastest.
LeelaChessZero benchmark with settings of Backend: Eigen. m7i.16xlarge Intel SPR was the fastest.
LeelaChessZero benchmark with settings of Backend: Eigen. m7i.16xlarge Intel SPR was the fastest.

With the AI-driven LeeelaChessZero benchmark the Genoa performance was a healthy improvement over the prior-generation Zen 3 instance but the M7i instance did manage to secure a first place finish here thanks to its strong AVX-512 / AMX performance.

Rodinia benchmark with settings of Test: OpenMP Streamcluster. m7i.16xlarge Intel SPR was the fastest.
Rodinia benchmark with settings of Test: OpenMP LavaMD. m7a.16xlarge AMD Zen 4 was the fastest.
HeFFTe - Highly Efficient FFT for Exascale benchmark with settings of Test: c2c, Backend: FFTW, Precision: float, X Y Z: 256. m7a.16xlarge AMD Zen 4 was the fastest.
HeFFTe - Highly Efficient FFT for Exascale benchmark with settings of Test: c2c, Backend: FFTW, Precision: float, X Y Z: 512. m7a.16xlarge AMD Zen 4 was the fastest.

For those running HPC workloads in the cloud, the new M7a instances are very powerful and much better than the prior-generation Zen 3 instances and now out-running the AWS Graviton3 class of processors.

Algebraic Multi-Grid Benchmark benchmark with settings of . m7a.16xlarge AMD Zen 4 was the fastest.
Algebraic Multi-Grid Benchmark benchmark with settings of . m7a.16xlarge AMD Zen 4 was the fastest.
Xcompact3d Incompact3d benchmark with settings of Input: input.i3d 129 Cells Per Direction. m7a.16xlarge AMD Zen 4 was the fastest.
Xcompact3d Incompact3d benchmark with settings of Input: input.i3d 129 Cells Per Direction. m7a.16xlarge AMD Zen 4 was the fastest.
Xcompact3d Incompact3d benchmark with settings of Input: input.i3d 193 Cells Per Direction. m7a.16xlarge AMD Zen 4 was the fastest.
Xcompact3d Incompact3d benchmark with settings of Input: input.i3d 193 Cells Per Direction. m7a.16xlarge AMD Zen 4 was the fastest.

But for some cases the AWS Graviton3 instances do retain better value or at least comparable to the M7a costs.


Related Articles