Performance on Windows (msysgit) with latest patches

Hi,

After all the performance patches that has been commited to the 
git.git/master branch, I figured it would be interesting to check out 
how it would affect times on Windows. So, I merged the git.git/master 
branch into the 4msysgit.git/next branch and ran the scripts which Moe 
provided. The result you'll find below.

git-status got a 1.6X speed-up, and git-commit on a single file got a 
2.5X speed-up. However, git-add and git-commit of all 100K files got a 
slight slowdown.


Before                        After
performance patches           performance patches
-------------------------     -------------------------
Command: git init             Command: git init
-------------------------     -------------------------

real    0m0.031s              real    0m0.032s
user    0m0.031s              user    0m0.015s
sys     0m0.000s              sys     0m0.015s

-------------------------     -------------------------
Command: git add .            Command: git add .
-------------------------     -------------------------

real    0m19.328s             real    0m20.204s
user    0m0.015s              user    0m0.015s
sys     0m0.015s              sys     0m0.000s

-------------------------     -------------------------
Command: git commit -a...     Command: git commit -a...
-------------------------     -------------------------

real    0m30.937s             real    0m31.328s
user    0m0.015s              user    0m0.015s
sys     0m0.015s              sys     0m0.015s

-------------------------     -------------------------
3x Command: git-status        3x Command: git-status
-------------------------     -------------------------

real    0m19.531s             real    0m12.016s
user    0m0.211s              user    0m0.274s
sys     0m0.136s              sys     0m0.091s

real    0m19.532s             real    0m12.016s
user    0m0.259s              user    0m0.242s
sys     0m0.091s              sys     0m0.137s

real    0m19.593s             real    0m12.000s
user    0m0.211s              user    0m0.227s
sys     0m0.152s              sys     0m0.090s

-------------------------     -------------------------
Command: git commit...        Command: git commit...
          (single file)                 (single file)
-------------------------     -------------------------

real    0m36.688s             real    0m14.640s
user    0m0.031s              user    0m0.015s
sys     0m0.000s              sys     0m0.000s

Total time --------------     Total time --------------
Real time:     2m 25.703s     Real time:     1m 42.358s
- User time:   0m  0.031s     - User time:   0m  0.015s
- Kernel time: 0m  0.031s     - Kernel time: 0m  0.078s
= Idle time:   2m 25.640s     = Idle time:   1m 42.264s
-------------------------     -------------------------

--
.marius

On Tue, 14 Aug 2007, Marius Storm-Olsen wrote:

git-status got a 1.6X speed-up, and git-commit on a single file got a 2.5X
speed-up. However, git-add and git-commit of all 100K files got a slight
slowdown.

The slight slowdown might be noise in the measurements (for example, just 
IO differences due to layout changes). Did you run it multiple times both 
ways to see how stable the numbers are? The changes really shouldn't have 
any visible slowdown, ever.

(Of course, it may be that something else changed, or simply things like 
code layout changed cachelines etc, so small performance differences can 
be due to any number of other factors too).

			Linus