The Compelling AVX-512 Performance Advantage On AMD EPYC 9005 "Turin"

Written by Michael Larabel in Processors on 31 January 2025 at 10:56 AM EST. Page 4 of 5. 5 Comments.

👁 oneDNN benchmark with settings of Harness: Deconvolution Batch shapes_1d, Engine: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 oneDNN benchmark with settings of Harness: Deconvolution Batch shapes_3d, Engine: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 oneDNN benchmark with settings of Harness: IP Shapes 3D, Engine: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 oneDNN benchmark with settings of Harness: Recurrent Neural Network Training, Engine: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 oneDNN benchmark with settings of Harness: Recurrent Neural Network Inference, Engine: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 PyTorch benchmark with settings of Device: CPU, Batch Size: 256, Model: ResNet-50. EPYC Turin: AVX-512 Enabled was the fastest.

👁 PyTorch benchmark with settings of Device: CPU, Batch Size: 512, Model: ResNet-50. EPYC Turin: AVX-512 Enabled was the fastest.

AVX-512 support continues to serve as a big win for AI workloads...

👁 TensorFlow benchmark with settings of Device: CPU, Batch Size: 512, Model: ResNet-50. EPYC Turin: AVX-512 Enabled was the fastest.

AMD would have had a much tougher time competing in HPC and AI if they hadn't adopted AVX-512 back in Zen 4 and improved it further now with Zen 5.

👁 OpenVINO benchmark with settings of Model: Face Detection FP16-INT8, Device: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 OpenVINO benchmark with settings of Model: Age Gender Recognition Retail 0013 FP16-INT8, Device: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 OpenVINO benchmark with settings of Model: Person Detection FP16, Device: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 OpenVINO benchmark with settings of Model: Weld Porosity Detection FP16, Device: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 OpenVINO benchmark with settings of Model: Machine Translation EN To DE FP16, Device: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 OpenVINO benchmark with settings of Model: Road Segmentation ADAS FP16, Device: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

👁 OpenVINO benchmark with settings of Model: Person Re-Identification Retail FP16, Device: CPU. EPYC Turin: AVX-512 Enabled was the fastest.

OpenVINO is able to leverage Advanced Matrix Extensions (AMX) with newer Intel Xeon processors but even EPYC Turin just with AVX-512 tends to perform competitive here for CPU-based inferencing.

5 Comments - Next Page

Page: 1 2 3 4 5 Next Page

URL: https://www.phoronix.com/review/amd-epyc-turin-avx512/4

⇱ The Compelling AVX-512 Performance Advantage On AMD EPYC 9005 "Turin" - Phoronix

The Compelling AVX-512 Performance Advantage On AMD EPYC 9005 "Turin"