[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: faster client pre-1.0: neon prefetching, multithreading

From: Branko Čibej <brane_at_xbc.nu>
Date: 2003-02-11 19:37:03 CET

Brandon Ehle wrote:

> solo turn wrote:
>
>> do you have maybe an explanation wyh the ssh access method has the
>> same scalability behaviour than local access, and webdav seems to be
>> factor 3? is it just the asynchrounous, or is it the continouus
>> opening of a connection and get a little information from the server?
>>
>>
> BTW, in case you haven't noticed, I've had the call tree for checkout
> and import posted here for a while for reference.
>
> http://www.pipedreaminteractive.com/svn/checkout.png
>
> and
>
> http://www.pipedreaminteractive.com/svn/import.png
>
> Its very visibile that nothing takes up much time except the delta
> code right now. Until that disappears, I don't think its very useful
> to track performance of the systems (at least for import & export) as
> a lot could change once the delta algorithm is optional or made faster.

Uh, are you counting CPU time or runtime? If ra_dav is 3x slower than
ra_local, hten you're probably not counting in the time when the client
is blocked.

Besides, sending deltas for checkout and import is total nonsense. The
fix is not speeding up vdelta (bet: you won't get a 2x speedup anyway),
but to remove it from the checkout and import code path entirely.

-- 
Brane Čibej   <brane_at_xbc.nu>   http://www.xbc.nu/brane/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Feb 11 19:37:48 2003

This is an archived mail posted to the Subversion Dev mailing list.