Intel Linux Optimizations Help AMD EPYC "Genoa" Improve Scaling To 384 Threads

Written by Michael Larabel in Software on 6 April 2023 at 04:00 PM EDT. Page 4 of 5. 13 Comments.

PostgreSQL benchmark with settings of Scaling Factor: 1000, Clients: 500, Mode: Read Only. Clear Linux: 48 Threads was the fastest.

With the PostgreSQL 15 database server, Intel's Clear Linux with all of the company's latest kernel patches continued to show better scaling than CentOS Stream and Ubuntu as the core counts increased. The advantages were smaller here but consistent at 12 cores/threads and above with the exception of the 96 thread configuration hitting similar performance between distributions.

PostgreSQL benchmark with settings of Scaling Factor: 1000, Clients: 500, Mode: Read Write. CentOS Stream: 96 Threads was the fastest.

PostgreSQL benchmark with settings of Scaling Factor: 1000, Clients: 500, Mode: Read Write, Average Latency. CentOS Stream: 96 Threads was the fastest.

When switching from just a database read scenario to a mix of reads and writes, Clear Linux presented much greater performance for PostgreSQL at 192 cores and 192 cores + SMT.

PostgreSQL benchmark with settings of Scaling Factor: 1000, Clients: 800, Mode: Read Write. Clear Linux: 96 Threads was the fastest.

PostgreSQL benchmark with settings of Scaling Factor: 1000, Clients: 800, Mode: Read Write, Average Latency. Clear Linux: 96 Threads was the fastest.

As the number of concurrent clients increased, Intel's Linux platform continued to show much greater performance at 192 cores (with/without SMT) over CentOS Stream and Ubuntu.

13 Comments - Next Page

Page: 1 2 3 4 5 Next Page