Igt-dev Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: Ville Syrjala <ville.syrjala@linux.intel.com>
To: igt-dev@lists.freedesktop.org
Subject: [igt-dev] [PATCH i-g-t 0/3] lib/i915/perf: Speed up the build
Date: Mon,  3 Apr 2023 20:47:57 +0300	[thread overview]
Message-ID: <20230403174800.19621-1-ville.syrjala@linux.intel.com> (raw)

From: Ville Syrjälä <ville.syrjala@linux.intel.com>

The perf xml stuff generates terrible C code that is giving
gcc a big headache and leading to disgustingly long build 
times. Try to generate a bit more sensible code to speed
up the build.

Total build time change for the series:
$ ninja -Cbuild  clean
$ time ninja -Cbuild 
 ADL (4+8 cores):
  - real        1m18,244s
  + real        0m36,695s
 VLV (2 cores):
  - real        17m29.100s
  + real        9m44.268s

For comparison this is what I get if I skip all the perf xml
files apart from hsw:
 ADL:
  real  0m20,080s
 VLV
  real  7m3.045s

Thinking of further ways to reduce this... There are about a
thousand "Ext" metric sets. Are those *actually* useful or
could we just skip them? That would give us a total build 
time of:
 ADL:
  real  0m24,593s
 VLV:
  real  8m15.977s

Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>

Ville Syrjälä (3):
  lib/i915/perf: Stop generating silly C code
  lib/i915/perf: Stop making copies of the registers
  lib/i915/perf: Convert the metric counters to an array as well

 .../perf-configs/perf-metricset-codegen.py    | 81 ++++++++++++++-----
 .../perf-configs/perf-registers-codegen.py    | 30 +++----
 lib/i915/perf.h                               |  8 +-
 3 files changed, 78 insertions(+), 41 deletions(-)

-- 
2.39.2

             reply	other threads:[~2023-04-03 17:48 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-04-03 17:47 Ville Syrjala [this message]
2023-04-03 17:47 ` [igt-dev] [PATCH i-g-t 1/3] lib/i915/perf: Stop generating silly C code Ville Syrjala
2023-04-03 17:47 ` [igt-dev] [PATCH i-g-t 2/3] lib/i915/perf: Stop making copies of the registers Ville Syrjala
2023-04-03 17:48 ` [igt-dev] [PATCH i-g-t 3/3] lib/i915/perf: Convert the metric counters to an array as well Ville Syrjala
2023-04-03 19:07 ` [igt-dev] ✓ Fi.CI.BAT: success for lib/i915/perf: Speed up the build Patchwork
2023-04-04  1:27 ` [igt-dev] ✓ Fi.CI.IGT: " Patchwork
2023-04-04  7:59 ` [igt-dev] [PATCH i-g-t 0/3] " Lionel Landwerlin

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=20230403174800.19621-1-ville.syrjala@linux.intel.com \
    --to=ville.syrjala@linux.intel.com \
    --cc=igt-dev@lists.freedesktop.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