News: 0001572480

  ARM Give a man a fire and he's warm for a day, but set fire to him and he's warm for the rest of his life (Terry Pratchett, Jingo)

Linux 6.17 Showing Off Some Nice Gains For 5th Gen AMD EPYC "Turin" Performance

([AMD] 83 Minutes Ago Linux 6.17)


Earlier this week I provided benchmarks looking at [1]the Linux 5.15 LTS through Linux 6.17 Git kernel performance using the latest stable and development kernels compared to the Long Term Support (LTS) kernels over the past four years. For having hardware support back to Linux 5.15 I was using an AMD EPYC Milan-X server. In those benchmarks there were some nice gains and even from Linux 6.16 to 6.17 Git was around a 3% geo mean improvement. So I was curious to run some benchmarks on the latest-generation AMD EPYC 9005 "Turin" processors to see if there was similar uplift there.

The 5th Gen AMD EPYC "Turin" performance is indeed looking nice on the in-development [2]Linux 6.17 kernel with some nice incremental improvements beyond the already-great Linux 6.16 performance.

For some quick benchmarks before closing out the month, I ran some Linux 6.16 vs. 6.17 Git benchmarks on a dual socket AMD EPYC 9965 server. No other changes to this flagship AMD EPYC Turin server configuration besides rebooting after going from Linux 6.16 to 6.17 Git from the Ubuntu Mainline Kernel PPA builds.

In the Stress-NG kernel micro-benchmarks seeing some very healthy improvements to the kernel's futex performance and more with Linux 6.17.

Some small slowdowns in a few areas, however.

In PostgreSQL also seeing some very hearty gains when running on Linux 6.17, similar to some of the gains seen on the older servers too.

In addition to a number of performance wins, there are also [3]many great features with Linux 6.17 . The Linux 6.17 stable release should be out around the end of September.

More benchmarks still being carried out. If you missed the big kernel comparison earlier this week, see [4]Linux 5.15 LTS To 6.17 Benchmarks: Four Years Of Kernel Improvement Net 37% Improvement On AMD EPYC .



[1] https://www.phoronix.com/review/linux-515-617-performance

[2] https://www.phoronix.com/search/Linux+6.17

[3] https://www.phoronix.com/review/linux-617-features

[4] https://www.phoronix.com/review/linux-515-617-performance


phoronix

On the other hand, the TCP camp also has a phrase for OSI people.
There are lots of phrases. My favorite is `nitwit' -- and the rationale
is the Internet philosophy has always been you have extremely bright,
non-partisan researchers look at a topic, do world-class research, do
several competing implementations, have a bake-off, determine what works
best, write it down and make that the standard.
The OSI view is entirely opposite. You take written contributions
from a much larger community, you put the contributions in a room of
committee people with, quite honestly, vast political differences and all
with their own political axes to grind, and four years later you get
something out, usually without it ever having been implemented once.
So the Internet perspective is implement it, make it work well,
then write it down, whereas the OSI perspective is to agree on it, write
it down, circulate it a lot and now we'll see if anyone can implement it
after it's an international standard and every vendor in the world is
committed to it. One of those processes is backwards, and I don't think
it takes a Lucasian professor of physics at Oxford to figure out which.
-- Marshall Rose, "The Pied Piper of OSI"