* some -o discard performance numbers
@ 2009-11-16 15:48 Christoph Hellwig
0 siblings, 0 replies; only message in thread
From: Christoph Hellwig @ 2009-11-16 15:48 UTC (permalink / raw)
To: linux-btrfs
I played around with the btrfs discard support now that I've finished
trying up the ATA TRIM support to the block layer. These numbers are
with an OCZ-Vertex SSD with the 1.4 firmware, on a 2.6.32-rc7 kernel
and current git btrfs-progs with my patch do discard the whole device
at mkfs time.
I used this little script (with the discard option comment out for the
non-discard run):
---- snip ----
#!/bin/sh
OPTS=
OPTS="-o discard"
./mkfs.btrfs /dev/sdb
mount -t btrfs $OPTS /dev/sdb /mnt/
cp -a ../linux-2.6 /mnt/
echo 1 > /proc/sys/vm/drop_caches
time rm -rf /mnt/linux-2.6
time sync
umount /mnt/
---- snip ----
And here are the numbers:
---- snip ----
rm -rf sync
With TRIM:
real 0m19.375s real 0m8.603s
user 0m0.065s user 0m0.000s
sys 0m11.122s sys 0m0.070s
real 0m16.267s real 0m8.277s
user 0m0.049s user 0m0.001s
sys 0m11.175s sys 0m0.117s
real 0m16.039s real 0m9.883s
user 0m0.058s user 0m0.000s
sys 0m10.993s sys 0m0.157s
real 0m16.277s real 0m8.419s
user 0m0.057s user 0m0.000s
sys 0m11.196s sys 0m0.161s
real 0m16.264s real 0m9.017s
user 0m0.063s user 0m0.000s
sys 0m11.066s sys 0m0.152s
Without:
real 0m15.796s real 0m0.181s
user 0m0.043s user 0m0.000s
sys 0m10.950s sys 0m0.133s
real 0m15.591s real 0m0.190s
user 0m0.067s user 0m0.000s
sys 0m11.157s sys 0m0.114s
real 0m15.877s real 0m0.230s
user 0m0.064s user 0m0.000s
sys 0m11.162s sys 0m0.115s
real 0m15.171s real 0m0.187s
user 0m0.051s user 0m0.000s
sys 0m11.156s sys 0m0.120s
real 0m15.965s real 0m0.194s
user 0m0.049s user 0m0.000s
sys 0m11.327s sys 0m0.134s
^ permalink raw reply [flat|nested] only message in thread
only message in thread, other threads:[~2009-11-16 15:48 UTC | newest]
Thread overview: (only message) (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2009-11-16 15:48 some -o discard performance numbers Christoph Hellwig
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox