On Thursday, 17 February 2000 at 10:35:17 -0800, Parag Patel wrote:
>
> Hello. I have a friend's quad PPro box temporarily sitting in my garage
> that I've been using to play with 4.0-CURRENT and vinum. Since the last
> series of bug-fixes a few weeks ago, everything works as advertised.
>
> Just for my own curiosity, I've been running some simple (probably
> questionable) tests involving find|cpio of the source tree and then
> buildworld -j16 on various vinum volumes.
>
> So I thought I'd forward the results to current, for what it's worth.
>
> The biggest (non)surprise is how big a difference softupdates makes.
> Nice!
>
> buildworld -j16:
> src & obj on IDE disk (softupdates)
> 5676.29 real 7701.09 user 6133.60 sys
> src & obj on /noraid volume (1 drive, softupdates)
> 6053.86 real 7969.94 user 7601.81 sys
I'm not overly happy with these results. I suspect that a throttling
mechanism I put in a few months back, which limited the number of
simultaneous transactions per drive, might have something to do with
it. Could you change the values of VINUM_MAXACTIVE and
DRIVE_MAXACTIVE (in /sys/dev/vinum/vinumvar.h) to 30000 each
(effectively disabling the throttling), recompile the kld and try
again please? Previously, I had limited the number of transactions to
10, but I've just completed a make world here, and the drives had up
to 400 outstanding transactions each, so there's quite a possibility
the limitation is causing performance problems.
Greg
--
Finger [EMAIL PROTECTED] for PGP public key
See complete headers for address and phone numbers
To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message
