[Cbe-oss-dev] ps3vram performance

Siarhei Siamashka siarhei.siamashka at gmail.com
Sat Sep 5 11:58:51 EST 2009


On Tuesday 01 September 2009, Siarhei Siamashka wrote:
> On Tuesday 01 September 2009, Siarhei Siamashka wrote:
> > How fast is ps3vram supposed to be? For me swap on ps3vram works
> > noticeably slower than swap on hdd (if benchmarked by a simple program
> > which allocates a large buffer and then reads/writes bytes at
> > pseudo-random locations).
> >
> > Also 'dd if=/dev/ps3vram of=/dev/null bs=1M' shows throughput something
> > like 29MB/s and this does not look very impressive. Does anybody have
> > better performance? I probably have some non-optimal kernel
> > configuration (using 2.6.29 kernel), but first would like to
> > know what level of performance is normally expected.
>
> OK, nevermind. Found the expected numbers here:
> http://lists.ozlabs.org/pipermail/cbe-oss-dev/2008-April/004588.html

Now using the kernel from
http://git.kernel.org/?p=linux/kernel/git/geoff/ps3-linux.git;a=summary
and default ps3 config, still get the same pitiful numbers:

# hdparm -t /dev/ps3vram

/dev/ps3vram:
 Timing buffered disk reads:   94 MB in  3.00 seconds =  31.32 MB/sec


# time dd if=/dev/ps3vram of=/dev/null bs=1M count=229 iflag=direct
229+0 records in
229+0 records out
240123904 bytes (240 MB) copied, 7.32892 s, 32.8 MB/s

real    0m7.338s
user    0m0.003s
sys     0m0.569s

I'm using European 80GB model, firmware 2.80

What could be wrong? Is the performance reported in the initial ps3vram
submission post still reachable? Or was it crippled in one of the firmware
updates, making ps3vram practically useless?

-- 
Best regards,
Siarhei Siamashka


More information about the cbe-oss-dev mailing list