From: Junio C Hamano <gitster@pobox.com>
To: Jeff King <peff@peff.net>
Cc: Taylor Blau <me@ttaylorr.com>, git@vger.kernel.org
Subject: Re: [PATCH 4/4] t5326: test propagating hashcache values
Date: Mon, 13 Sep 2021 22:49:22 -0700 [thread overview]
Message-ID: <xmqqilz34sgt.fsf@gitster.g> (raw)
In-Reply-To: <YUAxxPcd7RHC2AH+@coredump.intra.peff.net> (Jeff King's message of "Tue, 14 Sep 2021 01:23:16 -0400")
Jeff King <peff@peff.net> writes:
> On Mon, Sep 13, 2021 at 07:05:32PM -0700, Junio C Hamano wrote:
>
>> Taylor Blau <me@ttaylorr.com> writes:
>>
>> > Alas, there they are. They are basically no different than having the
>> > name-hash for single pack bitmaps, it's just now we don't throw them
>> > away when generating a MIDX bitmap from a state where the repository
>> > already has a single-pack bitmap.
>>
>> I actually wasn't expecting any CPU/time difference.
>
> I was, for the same reason we saw an improvement there in ae4f07fbcc
> (pack-bitmap: implement optional name_hash cache, 2013-12-21): without a
> name-hash, we try a bunch of fruitless deltas before we find a decent
> one.
Nice. We learn new things every once in a while ;-)
>> I hope that we are talking about the same name-hash, which is used
>> to sort the blobs so that when pack-objects try to find a good delta
>> base, the blobs from the same path will sit close to each other and
>> hopefully fit in the pack window.
>
> Yes, exactly. We spend less time finding the good ones if the likely
> candidates are close together. We may _also_ find better ones overall,
> depending on the number of candidates and the window size.
It is a pleasant surprise that in a real history like linux.git we
can even find good delta base without the name hash (unless we are
using insanely wide window size, that is). The objects in such a
case will be sorted solely by size, larger to smaller, and we need
to find a good delta base within that window. It may not be as
horrible as fast-import (which only tries to delta against the
previous single object), but with ~70k paths in a single revision
with history that is ~1m deep, I was pessimistic to see a size-only
sort to drop even a pair of blobs from the same path within the
default window size of 10. But it seems that such a pessimism is
unwarranted, which is a good news ;-).
next prev parent reply other threads:[~2021-09-14 5:49 UTC|newest]
Thread overview: 55+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-09-07 21:17 [PATCH 0/4] pack-bitmap: permute existing namehash values Taylor Blau
2021-09-07 21:17 ` [PATCH 1/4] t/helper/test-bitmap.c: add 'dump-hashes' mode Taylor Blau
2021-09-08 1:37 ` Ævar Arnfjörð Bjarmason
2021-09-08 2:24 ` Taylor Blau
2021-09-07 21:17 ` [PATCH 2/4] pack-bitmap.c: propagate namehash values from existing bitmaps Taylor Blau
2021-09-07 21:18 ` [PATCH 3/4] midx.c: respect 'pack.writeBitmapHashcache' when writing bitmaps Taylor Blau
2021-09-08 1:40 ` Ævar Arnfjörð Bjarmason
2021-09-08 2:28 ` Taylor Blau
2021-09-09 8:18 ` Ævar Arnfjörð Bjarmason
2021-09-09 9:34 ` Ævar Arnfjörð Bjarmason
2021-09-09 14:55 ` Taylor Blau
2021-09-09 15:50 ` Ævar Arnfjörð Bjarmason
2021-09-09 16:23 ` Taylor Blau
2021-09-09 14:47 ` Taylor Blau
2021-09-13 0:38 ` Junio C Hamano
2021-09-14 1:15 ` Taylor Blau
2021-09-07 21:18 ` [PATCH 4/4] t5326: test propagating hashcache values Taylor Blau
2021-09-08 1:46 ` Ævar Arnfjörð Bjarmason
2021-09-08 2:30 ` Taylor Blau
2021-09-17 8:56 ` Ævar Arnfjörð Bjarmason
2021-09-17 17:32 ` Taylor Blau
2021-09-17 19:22 ` Ævar Arnfjörð Bjarmason
2021-09-13 0:46 ` Junio C Hamano
2021-09-14 1:12 ` Taylor Blau
2021-09-14 2:05 ` Junio C Hamano
2021-09-14 5:11 ` Taylor Blau
2021-09-14 5:17 ` Taylor Blau
2021-09-14 5:27 ` Jeff King
2021-09-14 5:31 ` Taylor Blau
2021-09-14 5:23 ` Jeff King
2021-09-14 5:49 ` Junio C Hamano [this message]
2021-09-14 22:05 ` [PATCH v2 0/7] pack-bitmap: permute existing namehash values Taylor Blau
2021-09-14 22:06 ` [PATCH v2 1/7] t/helper/test-bitmap.c: add 'dump-hashes' mode Taylor Blau
2021-09-14 22:06 ` [PATCH v2 2/7] pack-bitmap.c: propagate namehash values from existing bitmaps Taylor Blau
2021-09-14 22:06 ` [PATCH v2 3/7] midx.c: respect 'pack.writeBitmapHashcache' when writing bitmaps Taylor Blau
2021-09-14 22:06 ` [PATCH v2 4/7] p5326: create missing 'perf-tag' tag Taylor Blau
2021-09-16 22:36 ` Jeff King
2021-09-17 4:14 ` Taylor Blau
2021-09-14 22:06 ` [PATCH v2 5/7] p5326: don't set core.multiPackIndex unnecessarily Taylor Blau
2021-09-16 22:38 ` Jeff King
2021-09-14 22:06 ` [PATCH v2 6/7] p5326: generate pack bitmaps before writing the MIDX bitmap Taylor Blau
2021-09-16 22:45 ` Jeff King
2021-09-17 4:20 ` Taylor Blau
2021-09-14 22:06 ` [PATCH v2 7/7] t5326: test propagating hashcache values Taylor Blau
2021-09-16 22:49 ` Jeff King
2021-09-16 22:52 ` [PATCH v2 0/7] pack-bitmap: permute existing namehash values Jeff King
2021-09-17 21:21 ` [PATCH v3 " Taylor Blau
2021-09-17 21:21 ` [PATCH v3 1/7] t/helper/test-bitmap.c: add 'dump-hashes' mode Taylor Blau
2021-09-17 21:21 ` [PATCH v3 2/7] pack-bitmap.c: propagate namehash values from existing bitmaps Taylor Blau
2021-09-17 21:21 ` [PATCH v3 3/7] midx.c: respect 'pack.writeBitmapHashcache' when writing bitmaps Taylor Blau
2021-09-17 21:21 ` [PATCH v3 4/7] p5326: create missing 'perf-tag' tag Taylor Blau
2021-09-17 21:21 ` [PATCH v3 5/7] p5326: don't set core.multiPackIndex unnecessarily Taylor Blau
2021-09-17 21:21 ` [PATCH v3 6/7] p5326: generate pack bitmaps before writing the MIDX bitmap Taylor Blau
2021-09-17 21:21 ` [PATCH v3 7/7] t5326: test propagating hashcache values Taylor Blau
2021-09-17 22:12 ` [PATCH v3 0/7] pack-bitmap: permute existing namehash values Jeff King
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=xmqqilz34sgt.fsf@gitster.g \
--to=gitster@pobox.com \
--cc=git@vger.kernel.org \
--cc=me@ttaylorr.com \
--cc=peff@peff.net \
/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).