comment
and update changelog with benchmark results
This commit is contained in:
parent
0518b62d2b
commit
dfa1c21b8a
2 changed files with 19 additions and 1 deletions
|
@ -24,7 +24,7 @@ git-annex (8.20200618) UNRELEASED; urgency=medium
|
|||
* Fix reversion that broke passing annex.* and remote.*.annex-*
|
||||
git configs with -c. (Since version 8.20200330.)
|
||||
* Bring back git-annex branch read cache. This speeds up some operations;
|
||||
git-annex sync --content gets up to 3x faster.
|
||||
git-annex sync --content --all gets 20% faster.
|
||||
|
||||
-- Joey Hess <id@joeyh.name> Thu, 18 Jun 2020 12:21:14 -0400
|
||||
|
||||
|
|
|
@ -0,0 +1,18 @@
|
|||
[[!comment format=mdwn
|
||||
username="joey"
|
||||
subject="""comment 9"""
|
||||
date="2020-07-06T16:23:07Z"
|
||||
content="""
|
||||
Cache brought back in [[!commit e72ec8b9b23346c4b741a071e9edae4460b233c9]].
|
||||
|
||||
Benchmarking git-annex sync --content (w/o --all) with 10k files, there
|
||||
was very little change, it's only 0.4% faster. With --all, it's 20% faster.
|
||||
|
||||
That's kind of weird, because w/o --all it was doing 3 redundant queries,
|
||||
and --all only adds one more. The first result makes me think that a) git
|
||||
cat-file is doing its own caching or takes good advantage of disk cache and
|
||||
b) roundtrip time communicating with it does not seem significant so
|
||||
--buffer would not be much of an improvement. But the second result muddies
|
||||
the waters, and I guess it would still be worth trying the --buffer
|
||||
optimisation.
|
||||
"""]]
|
Loading…
Add table
Reference in a new issue