Re: Subversion trunk (r1078338) HTTP(/WC?) performance problems?

John Beranek <john_at_redux.org.uk> writes:

> svn export
> ==========
>
> 1.6.16 from localhost trunk server (10 iterations)
> ra_neon: 1.43
> ra_serf: 2.91
>
> trunk from localhost trunk server (10 iterations)
> ra_neon: 1.48
> ra_serf: 2.95
>
> So, neon and serf retain their speeds relative to each other...

Are you measuring CPU, elapsed time, or something else? I see 1.7/serf
slightly faster that 1.7/neon.

Server 1.6

client 1.6/neon
real 0m8.319s
user 0m2.910s
sys 0m0.850s

client 1.7/neon
real 0m8.278s
user 0m2.410s
sys 0m0.990s

client 1.7/serf
real 0m7.514s
user 0m2.280s
sys 0m2.520s

Server 1.7

client 1.6/neon
real 0m8.571s
user 0m2.970s
sys 0m0.890s

client 1.7/neon
real 0m8.562s
user 0m2.380s
sys 0m1.000s

client 1.7/serf
real 0m7.732s
user 0m1.910s
sys 0m1.500s

The one small anomaly is 1.7/serf against a 1.6 server, it is fast but
it uses a bit more CPU. (I'm using r1080251).

-- 
Philip