Linux's Proposed Cache Aware Scheduling Benchmarks Show Big Potential On AMD EPYC Turin

Written by Michael Larabel in Software on 23 October 2025 at 01:00 PM EDT. Page 3 of 4. 39 Comments.

Across a variety of different OpenJDK Java workloads there were some nice advantages shown from running the AMD EPYC 9965 2P server on the Linux kernel build with Cache Aware Scheduling enabled.

👁 Renaissance benchmark with settings of Test: Savina Reactors.IO. Cache Aware Scheduling was the fastest.

👁 Renaissance benchmark with settings of Test: Finagle HTTP Requests. Cache Aware Scheduling was the fastest.

👁 Renaissance benchmark with settings of Test: Gaussian Mixture Model. Cache Aware Scheduling was the fastest.

👁 Renaissance benchmark with settings of Test: In-Memory Database Shootout. Cache Aware Scheduling was the fastest.

👁 Renaissance benchmark with settings of Test: Genetic Algorithm Using Jenetics + Futures. Cache Aware Scheduling was the fastest.

For being the first time I got around to testing the Cache Aware Scheduling patches myself, the results were exciting with some nice improvements compared to the stock Linux 6.17 upstream kernel.

39 Comments - Next Page

Page: 1 2 3 4 Next Page

URL: https://www.phoronix.com/review/cache-aware-scheduling-amd-turin/3

⇱ Linux's Proposed Cache Aware Scheduling Benchmarks Show Big Potential On AMD EPYC Turin Review - Phoronix

Linux's Proposed Cache Aware Scheduling Benchmarks Show Big Potential On AMD EPYC Turin