[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Re: [dm-devel] [PATCH 1/2] dm-userspace: use ring buffer instead of system call



FT> It's against on the top of the following patches:

Would you mind sending a patch against a vanilla tree?  I'm having a
hard time getting the patches (old and new) to apply cleanly and in
the right order.

FT> In my experiments, the ring buffer interface provides 10% better
FT> write performance with disktest benchmark, though dbench gives
FT> comparable performances (about 340 MB/s).

This is good to hear.  I am interested in testing it in some of my
real-life situations where performance is slower due to the high
latency.  This mostly affects the boot time of Xen guests.

I have added kernel caching back into my local version (which only
adds 200 lines).  It improves Xen guest boot performance
significantly, but does not affect dbench performance.  I will be
interested to see what kind of performance I can get from your
ringbuffer version, as well as possibly a combination of the two.

Thanks!

-- 
Dan Smith
IBM Linux Technology Center
Open Hypervisor Team
email: danms us ibm com

Attachment: pgpWKUjt2lk0I.pgp
Description: PGP signature


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]