On Tue, 2011-02-08 at 20:27 +0100, Stefan Sperling wrote:
> > Also instead of nebulous handwaving about "performance is bad", it
> > would be nice to have actually datasets and actual numbers. We have a
> > VM at the ASF which could be used for hosting a set of benchmarking
> > data; we just need somebody to put together the data and write the
> > scripts which run the benchmarks.
>
> I'll ask Neels about running his set of performance tests on the VM
> (see http://svn.haxx.se/dev/archive-2010-09/0526.shtml).
Wow, looking back at my own results, it seems so far away. Man, it's
slowly getting half a year since I first ran the benchmark.
If you have instructions for me to set up my humble tests on some
machine, I might be tempted to cook up some digest reporting on top of
it. The test script itself might benefit from a review, though.
If we run it on some idling box somewhere, we could simply increase the
extent of the test from 4x4 dir levels to a few more (keeping an eye on
exponential growth so it takes less than a day to run...).
I'm not sure if it's necessary to even care about other runs than
ra_local, since we're testing libsvn_wc performance, right? ra_local
should be where libsvn_wc perf loss is the most visible.
I still have that article about svn 1.7 waiting unreleased and growing a
beard. I don't want to print an article that has to say "svn 1.7 grew
slower than 1.6", so I'd want 1.7 to be same speed as 1.6 for a release.
We can already brag about massive memory usage improvement with deep
WCs. If we can just skip having to say "BUT it's slower in many common
cases", that'd be great.
>
> I know he's not following dev@ closely ATM, but maybe he'll see this one
> anyway cause his name is in it -- but I can more easily catch his attention
> by waving across the yard from my kitchen window :)
(monospace)
[[[
| * |
| . | ______
| | | ~|
| | | ~ | o/
______ | | |____| /| _o/
| ~ | | | / \ /_\ (<-- ascii daughter)
\o | ~ | | |
|\ |____| | |
/ \ | |
|
]]]
:)
And all the neighbors left bewildered.
~Neels
