AMD Ryzen Threadripper 7980X & 7970X Linux Performance Benchmarks

Written by Michael Larabel in Processors on 20 November 2023 at 09:00 AM EST. Page 12 of 13. 48 Comments.

oneDNN benchmark with settings of Harness: IP Shapes 3D, Data Type: f32, Engine: CPU. Threadripper 7980X - EXPO was the fastest.

oneDNN benchmark with settings of Harness: Recurrent Neural Network Training, Data Type: f32, Engine: CPU. Threadripper 7980X - EXPO was the fastest.

oneDNN benchmark with settings of Harness: Recurrent Neural Network Training, Data Type: bf16bf16bf16, Engine: CPU. Threadripper 7980X - EXPO was the fastest.

oneDNN benchmark with settings of Harness: Recurrent Neural Network Inference, Data Type: bf16bf16bf16, Engine: CPU. Threadripper 7970X was the fastest.

The Zen 4 Threadripper processors with AVX-512 also meant a much better job at handling AI worklaods such as those using the Intel oneDNN neural network software.

TensorFlow benchmark with settings of Device: CPU, Batch Size: 16, Model: ResNet-50. Xeon Platinum 8480+ was the fastest.

TensorFlow benchmark with settings of Device: CPU, Batch Size: 64, Model: ResNet-50. Xeon Platinum 8480+ was the fastest.

For TensorFlow the Intel Sapphire Rapids processors came out ahead thanks to the Advanced Matrix Extensions (AMX) support here. For software ready to make use of AMX, some strong gains can be found on the Sapphire Rapids side. At least generationally the new Threadripper 7000 series is much better off thanks to AVX-512 added with Zen 4.

QuantLib benchmark with settings of Configuration: Multi-Threaded. Threadripper 7980X was the fastest.

For those dealing with quantitative finance and other risk management scenarios or math modeling, the Threadripper 7980X offered excellent performance for the multi-threaded QuantLib calculations.

48 Comments - Next Page

Page: 1 2 3 4 5 6 7 8 9 10 11 12 13 Next Page