Intel Xeon Max 9480/9468 Show Significant Uplift In HPC & AI Workloads With HBM2e

Written by Michael Larabel in Processors on 28 June 2023 at 05:00 PM EDT. Page 5 of 7. 26 Comments.
JPEG XL Decoding libjxl benchmark with settings of CPU Threads: 1. Xeon Max 9480 2P: HBM Only was the fastest.
GraphicsMagick benchmark with settings of Operation: Noise-Gaussian. Xeon Max 9480 2P: HBM Only was the fastest.
7-Zip Compression benchmark with settings of Test: Decompression Rating. Xeon Max 9480 2P: HBM Only was the fastest.
OpenSSL benchmark with settings of Algorithm: ChaCha20-Poly1305. Xeon Max 9480 2P: HBM Only was the fastest.
Liquid-DSP benchmark with settings of Threads: 128, Buffer Length: 256, Filter Length: 57. Xeon Max 9480 2P: HBM Only was the fastest.

Over a wide range of technical workloads, a number of significant performance improvements could be observed by operating in the HBM-only mode.

ASKAP benchmark with settings of Test: tConvolve MT, Gridding. Xeon Max 9480 2P: HBM Only was the fastest.
ASKAP benchmark with settings of Test: tConvolve OpenMP, Degridding. Xeon Max 9480 2P: HBM Caching was the fastest.
Graph500 benchmark with settings of Scale: 26. Xeon Max 9468 2P: HBM Only was the fastest.
Graph500 benchmark with settings of Scale: 26. Xeon Max 9468 2P: HBM Only was the fastest.
GROMACS benchmark with settings of Implementation: MPI CPU, Input: water_GMX50_bare. Xeon Max 9480 2P: HBM Only was the fastest.
Darmstadt Automotive Parallel Heterogeneous Suite benchmark with settings of Backend: OpenMP, Kernel: NDT Mapping. Xeon Max 9480 2P: HBM Only was the fastest.
GPAW benchmark with settings of Input: Carbon Nanotube. Xeon Max 9468 2P: HBM Only was the fastest.

It was very interesting to see how Xeon Max would perform across a range of HPC workloads and quantifying the difference of HBM-only versus HBM caching modes.


Related Articles