Re: [StGit PATCH] Convert "sink" to the new infrastructure

["Catalin Marinas" <catalin.marinas@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 2116 bytes --]

2008/9/16 Karl Hasselström <kha@treskal.com>:
> On 2008-09-16 15:59:31 +0100, Catalin Marinas wrote:
>> I think we could merge your experimental branch into master. I gave
>> it a try and seems OK. The only issue I had was that I had an older
>> version of Git and it failed in really weird ways (stg pop still
>> busy-looping after 4 minutes and in another case it failed with
>> broken pipe). Once I pulled the latest Git, it was fine but we
>> should try to be compatible at least with the Git version in the
>> Debian testing distribution. It might be the patch at the top with
>> diff-ing several trees at once but I haven't checked.
>
> There are two patches that depend on new git versions. One needs git
> 1.5.6, which is in testing so I'll be pushing that to you;

OK.

> the other
> needs Junio's master branch, and i won't even consider asking you to
> take it until it's in a released git.

Correct :-)

> I hope to push it out to you tonight or tomorrow, but I have a small
> pet patch I'd like to finish first, so I might be late.

OK, no problem. I won't have much time before the weekend anyway.

>> BTW, I ran some benchmarks on stable/master/kha-experimental
>> branches with 300 patches from the 2.6.27-rc5-mm1 kernel. See
>> attached for the results. Since performance was my worry with the
>> stack log stuff, it turns out that there isn't a big difference with
>> real patches. I think pushing can be made even faster by trying a
>> git-apply first and taking the diff from the saved blobs in the log.
>
> When benchmarking recent StGits, you'll want to try goto as well,
> since push and pop are not yet new-infrastructure-ized (meaning
> they're getting slowdowns from the stack log, but no speedups (if any)
> from the new infrastructure).

Indeed, goto is faster even than the stable branch. I attached the new
figures. I think it could go even faster if pushing attempts a "git
apply" first before the index merge. With the stack log, the patch
diff should be saved already so no need for a "git diff" (as in the
stable branch).

-- 
Catalin

[-- Attachment #2: stack-log-benchmarks.txt --]
[-- Type: text/plain, Size: 1474 bytes --]

CPU: Intel Pentium 4 @ 2.5GHz
Memory: 1GB

2.6.27-rc5-mm1 kernel, 300 patches uncommitted

pop/push ran a few times to heat the caches before running the
benchmarks.


Stable stgit (v0.14.3 + some fixes)

$ time stg pop -a

real	0m1.775s
user	0m0.956s
sys	0m0.724s

$ time stg push -a (fast-forward)

real	0m5.001s
user	0m1.844s
sys	0m2.860s

$ time stg push -a (no fast-forward)

real	1m27.133s
user	0m36.998s
sys	0m34.894s

$ time stg goto top-patch (fast-forward)

real	0m5.314s
user	0m1.920s
sys	0m2.768s

$ time stg goto top-patch (no fast-forward)

real	1m39.040s
user	0m37.022s
sys	0m35.666s


Current stgit master (no stack log):

$ time stg pop -a

real	0m1.621s
user	0m0.820s
sys	0m0.688s

$ time stg push -a (fast-forward)

real	0m27.205s
user	0m8.741s
sys	0m16.849s

$ time stg push -a (no fast-forward)

real	2m8.209s
user	0m46.031s
sys	0m57.260s

$ time stg goto top-patch (fast-forward)

real	0m10.437s
user	0m2.160s
sys	0m2.464s

$ time stg goto top-patch (no fast-forward)

real	1m23.244s
user	0m38.158s
sys	0m36.086s


kha/experimantal stgit (with stack log):

$ time stg pop -a

real	0m2.419s
user	0m1.144s
sys	0m1.132s

$ time stg push -a (fast-forward)

real	0m29.594s
user	0m9.217s
sys	0m17.145s

$ time stg push -a (no fast-forward)

real	2m10.270s
user	0m50.919s
sys	1m2.088s

$ time stg goto top-patch (fast-forward)

real	0m2.170s
user	0m1.084s
sys	0m0.460s

$ time stg goto top-patch (no fast-forward)

real	1m18.271s
user	0m39.026s
sys	0m31.938s