Intel's Rewrite Of Linux MM CID Code Showing Some Nice Gains For AMD

Written by Michael Larabel in Software on 7 November 2025 at 10:25 AM EST. Page 3 of 3. 9 Comments.

👁 ClickHouse benchmark with settings of 100M Rows Hits Dataset, First Run / Cold Cache. rseq-cid was the fastest.

👁 ClickHouse benchmark with settings of 100M Rows Hits Dataset, Third Run. rseq-cid was the fastest.

The ClickHouse database saw very small gains from the rseq-cid kernel run.

👁 PostgreSQL benchmark with settings of Scaling Factor: 100, Clients: 800, Mode: Read Write. rseq-cid was the fastest.

👁 PostgreSQL benchmark with settings of Scaling Factor: 100, Clients: 800, Mode: Read Write, Average Latency. rseq-cid was the fastest.

👁 PostgreSQL benchmark with settings of Scaling Factor: 100, Clients: 1000, Mode: Read Write. rseq-cid was the fastest.

👁 PostgreSQL benchmark with settings of Scaling Factor: 100, Clients: 1000, Mode: Read Write, Average Latency. rseq-cid was the fastest.

With the PostgreSQL database server I did encounter some performance improvements as well with the rseq/cid derived kernel build. Gleixner's PostgreSQL benchmarking noted in his mailing list patch series was done with PostgreSQL + sysbench while my PostgreSQL run was with pgbench.

👁 Schbench benchmark with settings of Message Threads: 4, Cache Footprint: 128 kb, Locking: No. rseq-cid was the fastest.

👁 Schbench benchmark with settings of Message Threads: 4, Cache Footprint: 256 kb, Locking: No. rseq-cid was the fastest.

👁 Schbench benchmark with settings of Message Threads: 8, Cache Footprint: 128 kb, Locking: No. rseq-cid was the fastest.

👁 Schbench benchmark with settings of Message Threads: 16, Cache Footprint: 256 kb, Locking: No. rseq-cid was the fastest.

👁 Schbench benchmark with settings of Message Threads: 32, Cache Footprint: 256 kb, Locking: No. rseq-cid was the fastest.

👁 Schbench benchmark with settings of Message Threads: 4, Cache Footprint: 128 kb, Locking: Yes. rseq-cid was the fastest.

👁 Schbench benchmark with settings of Message Threads: 64, Cache Footprint: 256 kb, Locking: No. rseq-cid was the fastest.

👁 Schbench benchmark with settings of Message Threads: 32, Cache Footprint: 128 kb, Locking: Yes. rseq-cid was the fastest.

👁 Schbench benchmark with settings of Message Threads: 64, Cache Footprint: 128 kb, Locking: Yes. rseq-cid was the fastest.

👁 Schbench benchmark with settings of Message Threads: 128, Cache Footprint: 128 kb, Locking: Yes. rseq-cid was the fastest.

With the synthetic schbench scheduler benchmark there were some improvements found with the rseq/cid code branch. Not to much surprise given these patches are touching the kernel's scheduler code.

Consider me intrigued so far with what I am seeing from this initial benchmarking of the tglx/devel.git rseq/cid Git branch. As mentioned if there is enough reader interest to justify the time needed, I can run more benchmarks of the latest rseq/cid code on more hardware and more benchmarks... At least from this first run on an AMD EPYC server the results are looking quite positive.

9 Comments

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.

Page: 1 2 3

URL: https://www.phoronix.com/review/rseq-cid-benchmarks/3

⇱ Intel's Rewrite Of Linux MM CID Code Showing Some Nice Gains For AMD - Phoronix

Intel's Rewrite Of Linux MM CID Code Showing Some Nice Gains For AMD