[conspire] DMA for SATA? -was: SATA HD transfers slow?
Mark S Bilk
mark at cosmicpenguin.com
Mon Jul 26 02:10:32 PDT 2010
On Sat, Jul 24, 2010 at 07:02:27PM -0400, Luke S Crawford wrote:
>...
Luke, many thanks for your very informative answers to my two
previous posts!
But every SATA transfer is still using up one of my two CPU cores.
See the picture of the KDE4 System Monitor, downloadable from here:
http://cosmicpenguin.com/tmp/SysMon.jpg
The wide peak is the transfer, and the narrow ones about every
second are Firefox.
So I think SATA is using PIO. How can I enable DMA? hdparm fails:
hdparm -d /dev/sdb
/dev/sdb:
HDIO_GET_DMA failed: Inappropriate ioctl for device
Here are copies of the CPU line from "top". The numbers add up
to 100%, so the measurements for both cores are being added
together. The transfer is using 30-45% wa (wait?) and 4 or 8%
system (for disk-disk or disk-/dev/null). Added and multiplied
by 2 these equal about 100% of one of the cores. Each transfer
is a 700+MB file. I used a new one each time so it wouldn't be
in disk cache.
idle
Cpu(s): 11.9%us, 3.2%sy, 0.0%ni, 84.8%id, 0.0%wa, 0.0%hi, 0.2%si, 0.0%st
Cpu(s): 11.3%us, 2.7%sy, 0.0%ni, 85.2%id, 0.8%wa, 0.0%hi, 0.0%si, 0.0%st
idle, during one of Firefox's peaks
Cpu(s): 45.4%us, 2.5%sy, 0.0%ni, 51.5%id, 0.7%wa, 0.0%hi, 0.0%si, 0.0%st
SATA to SATA transfer using mc
Cpu(s): 39.5%us, 15.0%sy, 0.0%ni, 13.9%id, 29.7%wa, 0.3%hi, 1.5%si, 0.0%st
Cpu(s): 31.2%us, 14.1%sy, 0.0%ni, 20.4%id, 32.8%wa, 0.2%hi, 1.3%si, 0.0%st
SATA to SATA transfer using cp
Cpu(s): 28.0%us, 10.7%sy, 0.0%ni, 24.6%id, 36.0%wa, 0.2%hi, 0.5%si, 0.0%st
Cpu(s): 32.4%us, 11.7%sy, 0.0%ni, 19.0%id, 35.3%wa, 0.5%hi, 1.1%si, 0.0%st
SATA to /dev/null transfer using cp
Cpu(s): 16.9%us, 9.1%sy, 0.0%ni, 34.7%id, 37.7%wa, 0.5%hi, 1.1%si, 0.0%st
SATA to /dev/null transfer using cp, probably during a Firefox peak
Cpu(s): 49.2%us, 8.3%sy, 0.0%ni, 4.8%id, 34.5%wa, 1.0%hi, 2.3%si, 0.0%st
SATA to /dev/null transfer using cp, with Firefox stopped
Cpu(s): 3.3%us, 7.4%sy, 0.0%ni, 43.6%id, 44.7%wa, 0.3%hi, 0.6%si, 0.0%st
More information about the conspire
mailing list