From: dave seddon <dave.seddon.ca@gmail.com>
To: Cake List <cake@lists.bufferbloat.net>
Cc: Tom Herbert <tom@herbertland.com>
Subject: [Cake] flow_hash_from_keys_small?
Date: Mon, 25 May 2026 08:46:37 -0700 [thread overview]
Message-ID: <CANypexQuveja-zHda=N-1OuRZ0P8NvKxfBY4gxCaVwfcdGX3Rw@mail.gmail.com> (raw)
G'day,
Here are a couple of patches that make cake ever so slightly faster.
Longer
For the last few months, I've been working on Tom's xdp2 project, comparing
the performance of the kernel's existing flow dissector versus what xdp2
(c/c++) and a xdp2-rs (rust) can produce.
For real world performance tests the lab has x2 pairs of machines:
- AMD Ryzen 5 PRO 2400G CPU with Intel X710
- AMD Ryzen 5 PRO 2400G CPU with ConnectX-4 Lx
After extensive analysis, I compared the flow metadata between the kernel
and xdp2, and found that simplifying the flow hash yields a small
performance improvement. I have NOT tested this on Intel CPUs.
The kernel used for the performance testing is
[das@hp1:~]$ uname -a Linux hp1 7.1.0-rc4 #1-NixOS SMP PREEMPT_DYNAMIC Tue
Jan 1 00:00:00 UTC 1980 x86_64 GNU/Linux
The draft letter for the kernel network mailing list, which describes the
changes, is here:
https://github.com/randomizedcoder/xdp2/blob/flow-keys-compat-reorder/kernel-patches/series1-flow-hash-small/v1/0000-cover-letter.patch
Documentation patch
https://github.com/randomizedcoder/xdp2/blob/flow-keys-compat-reorder/kernel-patches/series1-flow-hash-small/v1/0001-Documentation-networking-add-flow_dissector-consumer.patch
Patches
https://github.com/randomizedcoder/xdp2/blob/flow-keys-compat-reorder/kernel-patches/series1-flow-hash-small/v1/0002-net-flow_dissector-add-flow_hash_from_keys_small-5-t.patch#L82
https://github.com/randomizedcoder/xdp2/blob/flow-keys-compat-reorder/kernel-patches/series1-flow-hash-small/v1/0003-net-sched-sch_cake-use-flow_hash_from_keys_small-for.patch#L50
I'm interested in any feedback, particularly regarding the naming
"flow_hash_from_keys_small".
Thanks in advance
--
Regards,
Dave Seddon
+1 415 857 5102
next reply other threads:[~2026-05-25 15:46 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-05-25 15:46 dave seddon [this message]
2026-05-25 20:11 ` [Cake] Re: flow_hash_from_keys_small? Tom Herbert
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: https://lists.bufferbloat.net/postorius/lists/cake.lists.bufferbloat.net/
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CANypexQuveja-zHda=N-1OuRZ0P8NvKxfBY4gxCaVwfcdGX3Rw@mail.gmail.com' \
--to=dave.seddon.ca@gmail.com \
--cc=cake@lists.bufferbloat.net \
--cc=tom@herbertland.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox