Google Cloud C3D Shows Great Performance With AMD EPYC Genoa

Written by Michael Larabel in Processors on 27 October 2023 at 11:32 AM EDT. Page 4 of 5. 1 Comment.
TensorFlow benchmark with settings of Device: CPU, Batch Size: 16, Model: ResNet-50. c3d-standard-60 AMD Genoa was the fastest.

For those making use of Google Cloud for AI workloads, the C3D VMs are much more capable now thanks to AVX-512.

TensorFlow benchmark with settings of Device: CPU, Batch Size: 64, Model: ResNet-50. c3d-standard-60 AMD Genoa was the fastest.
TensorFlow benchmark with settings of Device: CPU, Batch Size: 64, Model: ResNet-50. c3d-standard-60 AMD Genoa was the fastest.

With the AMD EPYC Genoa VMs there is much greater TensorFlow performance largely due to AVX-512 being implemented with the Zen 4 CPUs.

OpenVINO benchmark with settings of Model: Face Detection FP16, Device: CPU. c3d-standard-60 AMD Genoa was the fastest.
OpenVINO benchmark with settings of Model: Person Detection FP16, Device: CPU. c3d-standard-60 AMD Genoa was the fastest.
OpenVINO benchmark with settings of Model: Person Detection FP16, Device: CPU. c2-standard-60 Intel CXL was the fastest.
OpenVINO benchmark with settings of Model: Person Detection FP32, Device: CPU. c3d-standard-60 AMD Genoa was the fastest.
OpenVINO benchmark with settings of Model: Person Detection FP32, Device: CPU. c2-standard-60 Intel CXL was the fastest.
OpenVINO benchmark with settings of Model: Vehicle Detection FP16, Device: CPU. c3d-standard-60 AMD Genoa was the fastest.
OpenVINO benchmark with settings of Model: Vehicle Detection FP16, Device: CPU. c3d-standard-60 AMD Genoa was the fastest.
OpenVINO benchmark with settings of Model: Face Detection FP16-INT8, Device: CPU. c2-standard-60 Intel CXL was the fastest.
OpenVINO benchmark with settings of Model: Face Detection FP16-INT8, Device: CPU. c2-standard-60 Intel CXL was the fastest.
OpenVINO benchmark with settings of Model: Road Segmentation ADAS FP16, Device: CPU. c3d-standard-60 AMD Genoa was the fastest.
OpenVINO benchmark with settings of Model: Vehicle Detection FP16-INT8, Device: CPU. c3d-standard-60 AMD Genoa was the fastest.
OpenVINO benchmark with settings of Model: Face Detection Retail FP16-INT8, Device: CPU. c3d-standard-60 AMD Genoa was the fastest.
OpenVINO benchmark with settings of Model: Handwritten English Recognition FP16-INT8, Device: CPU. c3d-standard-60 AMD Genoa was the fastest.

The OpenVINO performance is also substantially improved with C3D over C2D and N2D thanks to AVX-512 and other Zen 4 improvements. In a few cases the Intel C2 VMs did deliver better performance, or in most cases was just lower latency -- and the Sapphire Rapids VMs can benefit from Advanced Matrix Extensions (AMX) with OpenVINO. In any event for those making use of OpenVINO on AMD VMs, Genoa really opens the door to it being very competitive.


Related Articles
Ampere Computing 2024 Roadmap Update: 256 Core 3nm CPU In 2025
Benchmarking The First RISC-V Cloud Server: Scaleway EM-RV1 Performance
AMD Announces Ryzen PRO 8840 & PRO 8000G Series CPUs
Intel Announces Gaudi 3 AI Accelerator, Intel Xeon 6 Brand
AMD Zen 4 vs. Zen 4C Performance, Zen 4C Core Scaling With Ryzen 5 8500G
Intel Core i3 14100 / i5 14500 vs. AMD Ryzen 5 8500G / 8600G In 500+ Benchmarks