Re: [PATCH 16/15] Add tracing to measure where most of the time is spent

On Wed, Apr 20, 2016 at 7:28 PM, Johannes Schindelin
[off-list ref] wrote:

Hi Duy,

On Wed, 20 Apr 2016, Nguyễn Thái Ngọc Duy wrote:

quoted

All the known heavy code blocks are measured (except object database
access). This should help identify if an optimization is effective or
not. An unoptimized git-status would give something like below (92% of
time is accounted). To sum up the effort of making git scale better:

 - read cache line is being addressed by index-helper
 - preload/refresh index by watchman
 - read packed refs by lmdb backend
 - diff-index by rebuilding cache-tree
 - read directory by untracked cache and watchman
 - write index by split index
 - name hash potientially by index-helper

read-cache.c:2075         performance: 0.004058570 s: read cache .../index
preload-index.c:104       performance: 0.004419864 s: preload index
read-cache.c:1265         performance: 0.000185224 s: refresh index
refs/files-backend.c:1100 performance: 0.001935788 s: read packed refs
diff-lib.c:240            performance: 0.000144132 s: diff-files
diff-lib.c:506            performance: 0.013592000 s: diff-index
name-hash.c:128           performance: 0.000614177 s: initialize name hash
dir.c:2030                performance: 0.015814103 s: read directory
read-cache.c:2565         performance: 0.004052343 s: write index, changed mask = 2
trace.c:420               performance: 0.048365509 s: git command: './git' 'status'

Thank you for doing this! It will be *highly* valuable to get the
performance on Windows where I want it to be, too.

Just to be clear, these are unoptimized numbers, no untracked cache,
no split index, no index helper, and on a tiny repository that is
git.git. It's mostly to show the percentage of time spent in each
phase. We probably can do similar measurement and record it in the
watchman patch (can't wait for lmdb to be here :).
-- 
Duy