git.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Nguyễn Thái Ngọc Duy" <pclouds@gmail.com>
To: git@vger.kernel.org
Cc: "Nguyễn Thái Ngọc Duy" <pclouds@gmail.com>
Subject: [PATCH 0/4] index-pack improvements
Date: Sat, 25 Feb 2012 17:56:12 +0700	[thread overview]
Message-ID: <1330167376-24859-1-git-send-email-pclouds@gmail.com> (raw)

The first two patches are posted already, which help keep memory usage
down in the present of large blobs. The forth patch attempts to
resolve deltas in parallel.

$ time ./git index-pack --threads=1 --verify -v XXXX
Indexing objects: 100% (165375/165375), done.
Resolving deltas: 100% (124749/124749), done.

real    1m15.470s
user    1m14.899s
sys     0m0.552s

$ time ./git index-pack --threads=2 --verify -v XXXX
Indexing objects: 100% (165375/165375), done.
Resolving deltas: 100% (124749/124749), done.

real    0m41.339s
user    1m15.116s
sys     0m0.680s

$ time ./git index-pack --threads=3 --verify -v XXXX
Indexing objects: 100% (165375/165375), done.
Resolving deltas: 100% (124749/124749), done.

real    0m37.008s
user    1m35.742s
sys     0m0.803s

$ time ./git index-pack --verify -v XXXX   # four core machine
Indexing objects: 100% (165375/165375), done.
Resolving deltas: 100% (124749/124749), done.

real    0m33.701s
user    1m51.316s
sys     0m0.768s

$ time ./git index-pack --threads=8 --verify -v XXXX
Indexing objects: 100% (165375/165375), done.
Resolving deltas: 100% (124749/124749), done.

real    0m33.638s
user    1m51.783s
sys     0m0.773s

So there's improvement from user persepective, but overhead is too
high (user time from 1m14 to 1m51). Making threaded_second_pass to
process 512 consecutive objects each iteration to reduce contention on
work_mutex does not help. Any ideas?

Nguyễn Thái Ngọc Duy (4):
  index-pack --verify: skip sha-1 collision test
  index-pack: reduce memory usage when the pack has large blobs
  index-pack: move second pass code into separate function
  index-pack: support multithreaded delta resolving

 Documentation/config.txt         |    4 +
 Documentation/git-index-pack.txt |   10 ++
 Makefile                         |    2 +-
 builtin/index-pack.c             |  298 +++++++++++++++++++++++++++++++-------
 4 files changed, 260 insertions(+), 54 deletions(-)

-- 
1.7.8.36.g69ee2

             reply	other threads:[~2012-02-25 10:54 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-02-25 10:56 Nguyễn Thái Ngọc Duy [this message]
2012-02-25 10:56 ` [PATCH 1/4] index-pack --verify: skip sha-1 collision test Nguyễn Thái Ngọc Duy
2012-02-25 10:56 ` [PATCH 2/4] index-pack: reduce memory usage when the pack has large blobs Nguyễn Thái Ngọc Duy
2012-02-25 10:56 ` [PATCH 3/4] index-pack: move second pass code into separate function Nguyễn Thái Ngọc Duy
2012-02-25 10:56 ` [PATCH 4/4] index-pack: support multithreaded delta resolving Nguyễn Thái Ngọc Duy

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=1330167376-24859-1-git-send-email-pclouds@gmail.com \
    --to=pclouds@gmail.com \
    --cc=git@vger.kernel.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).