From: zamazan4ik <zamazan4ik@users.noreply.github.com>
To: ml@inbox.vuxu.org
Subject: Re: Expand using Profile-Guided Optimization (PGO) and Post Link Optimization (PLO) in the Void Linux packages
Date: Sun, 03 Dec 2023 02:13:14 +0100 [thread overview]
Message-ID: <20231203011314.YoKYzH4-PI7BPYsdiNUJXILqsa433wIsn7SkHAMG2rs@z> (raw)
In-Reply-To: <gh-mailinglist-notifications-41a7ca26-5023-4802-975b-f1789d68868e-void-packages-47215@inbox.vuxu.org>
[-- Attachment #1: Type: text/plain, Size: 1337 bytes --]
New comment by zamazan4ik on void-packages repository
https://github.com/void-linux/void-packages/issues/47215#issuecomment-1837299561
Comment:
> I suspect the benefits of PGO to be dubious for precompiled packages, because timing results that select between alternatives will be determined by the microarchitecture of the build servers, which will not generally match that of users.
PGO does not depend on the timings. Instrumentation PGO "just" collects counters from the instrumented program, so there is no actual difference if a program finishes the same workload in minutes or hours - the counter values would be the same for almost all applications. The only thing that can differ - is if a program has internal time-dependent logic (like "run this subroutine every N seconds"). In this case, the counters will differ. But from my experience, such software is really rare in real life and such differences do not introduce measurable differences in the optimization results.
Regarding (micro)architecture - PGO does not depend on the architecture features. However, generally [is not recommended](https://discourse.llvm.org/t/profile-guided-optimization-pgo-related-questions-and-suggestions/75232/2) (check the answer for the fifth question) to re-use profiles between different architectures (e.g from x86-64 to ARM).
next prev parent reply other threads:[~2023-12-03 1:13 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-11-13 20:17 [ISSUE] " zamazan4ik
2023-11-15 13:52 ` dmarto
2023-11-15 20:36 ` zamazan4ik
2023-11-15 20:39 ` classabbyamp
2023-12-03 1:05 ` ahesford
2023-12-03 1:13 ` zamazan4ik [this message]
2024-03-02 1:43 ` github-actions
2024-03-16 1:44 ` [ISSUE] [CLOSED] " github-actions
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
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20231203011314.YoKYzH4-PI7BPYsdiNUJXILqsa433wIsn7SkHAMG2rs@z \
--to=zamazan4ik@users.noreply.github.com \
--cc=ml@inbox.vuxu.org \
/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;
as well as URLs for NNTP newsgroup(s).