8 vs. 12 Channel DDR5-6000 Memory Performance With AMD 5th Gen EPYC

Written by Michael Larabel in Memory on 20 November 2024 at 11:40 AM EST. Page 7 of 7. 29 Comments.
TensorFlow benchmark with settings of Device: CPU, Batch Size: 512, Model: ResNet-50. 12c DDR5-6000 was the fastest.

TensorFlow enjoyed a healthy 23% jump in performance out of the four additional memory channels.

OpenVINO benchmark with settings of Model: Person Detection FP16, Device: CPU. 12c DDR5-6000 was the fastest.
OpenVINO benchmark with settings of Model: Person Detection FP16, Device: CPU. 12c DDR5-6000 was the fastest.
OpenVINO benchmark with settings of Model: Road Segmentation ADAS FP16-INT8, Device: CPU. 12c DDR5-6000 was the fastest.
OpenVINO benchmark with settings of Model: Road Segmentation ADAS FP16-INT8, Device: CPU. 12c DDR5-6000 was the fastest.
OpenVINO benchmark with settings of Model: Machine Translation EN To DE FP16, Device: CPU. 12c DDR5-6000 was the fastest.
OpenVINO benchmark with settings of Model: Machine Translation EN To DE FP16, Device: CPU. 12c DDR5-6000 was the fastest.
OpenVINO benchmark with settings of Model: Weld Porosity Detection FP16-INT8, Device: CPU. 8c DDR5-6000 was the fastest.
OpenVINO benchmark with settings of Model: Weld Porosity Detection FP16-INT8, Device: CPU. 8c DDR5-6000 was the fastest.
OpenVINO benchmark with settings of Model: Noise Suppression Poconet-Like FP16, Device: CPU. 12c DDR5-6000 was the fastest.
OpenVINO benchmark with settings of Model: Noise Suppression Poconet-Like FP16, Device: CPU. 12c DDR5-6000 was the fastest.

The OpenVINO AI toolkit also showed off much better performance on this AMD EPYC Supermicro server from utilizing all twelve memory channels compared to eight channel configurations.

Whisperfile benchmark with settings of Model Size: Small. 12c DDR5-6000 was the fastest.
Whisperfile benchmark with settings of Model Size: Medium. 12c DDR5-6000 was the fastest.

Whisperfile as the Whisper.cpp based AI software also enjoyed making use of all twelve DDR5-6000 memory channels on this Supermicro 1P EPYC server.

ONNX Runtime benchmark with settings of Model: GPT-2, Device: CPU, Executor: Standard. 12c DDR5-6000 was the fastest.
ONNX Runtime benchmark with settings of Model: yolov4, Device: CPU, Executor: Standard. 12c DDR5-6000 was the fastest.
ONNX Runtime benchmark with settings of Model: ZFNet-512, Device: CPU, Executor: Standard. 12c DDR5-6000 was the fastest.
ONNX Runtime benchmark with settings of Model: bertsquad-12, Device: CPU, Executor: Standard. 12c DDR5-6000 was the fastest.
ONNX Runtime benchmark with settings of Model: fcn-resnet101-11, Device: CPU, Executor: Standard. 12c DDR5-6000 was the fastest.

The ONNX Runtime improvements out of the 12 channel memory configuration were smaller than in some of the other AI workloads.

OpenVKL benchmark with settings of Benchmark: vklBenchmarkCPU ISPC. 12c DDR5-6000 was the fastest.
7-Zip Compression benchmark with settings of Test: Compression Rating. 12c DDR5-6000 was the fastest.
7-Zip Compression benchmark with settings of Test: Decompression Rating. 12c DDR5-6000 was the fastest.

In total I ran nearly 200 benchmarks for this eight versus twelve DDR5-6000 memory channel comparison on the Supermicro H13SSL-N with AMD EPYC 9655 1P processor. Those wanting to see all the data can do so via this result file if weighing your options between eight and twelve memory channels depending upon the server/motherboard options and/or justifying the cost. Running these benchmarks were also helpful to myself for solidifying my decision to buy the H13SSL-N over some of the other lower-cost AMD EPYC 9004/9005 series motherboard options that only allow for eight memory channels.

If you enjoyed this article consider joining Phoronix Premium to view this site ad-free, multi-page articles on a single page, and other benefits. PayPal or Stripe tips are also graciously accepted. Thanks for your support.

Related Articles
About The Author
Michael Larabel

Michael Larabel is the principal author of Phoronix.com and founded the site in 2004 with a focus on enriching the Linux hardware experience. Michael has written more than 20,000 articles covering the state of Linux hardware support, Linux performance, graphics drivers, and other topics. Michael is also the lead developer of the Phoronix Test Suite, Phoromatic, and OpenBenchmarking.org automated benchmarking software. He can be followed via Twitter, LinkedIn, or contacted via MichaelLarabel.com.