I have a ST310212A (Capacity: 9.54 GB) drive in the RAQ 3...

Doing a
[root@www /admin]# /sbin/hdparm -Tt /dev/hda

/dev/hda:
Timing buffer-cache reads: 64 MB in 2.11 seconds =30.33 MB/sec
Timing buffered disk reads: 32 MB in 3.24 seconds = 9.88 MB/sec


http://www.oreillynet.com/pub/a/linux/2000/06/29/hdparm.html?page=1 gives some tips on speeding this up quite a bit -- however you have to be in single-user mode and as far as I know you cannot do this remotely....

[root@www /admin]# /sbin/hdparm /dev/hda

/dev/hda:
multcount = 32 (on)
I/O support = 0 (default 16-bit)
unmaskirq = 0 (off)
using_dma = 1 (on)
keepsettings = 0 (off)
nowerr = 0 (off)
readonly = 0 (off)
readahead = 8 (on)


Any reason why I/O is set to 16-bit by default and shouldn't it also be 32? As well shouldn't xfermode be set to 66 (UltraDMA mode 2) ?

Well since nobody replied I went ahead and did it:

1) enable 32 bit access:
/sbin/hdparm -c1 /dev/hda

2) enable dma:
/sbin//hdparm -d1 /dev/hda

3) enable xfermode to UltraDMA/66 & turn on unmaskirq
/sbin/hdparm -X66 -u1 /dev/hda

BEFORE:

Timing buffer-cache reads: 64 MB in 2.11 seconds =30.33 MB/sec
Timing buffered disk reads: 32 MB in 3.24 seconds = 9.88 MB/sec

AFTER:

Timing buffer-cache reads: 64 MB in 1.26 seconds =50.79 MB/sec
Timing buffered disk reads: 32 MB in 1.72 seconds =18.60 MB/sec

....I did this on my raq3.. I don't necessarily recommend you do it since the hard drive on yours might be different and take different parameters..but, it has made a big difference on mine - especially when disk swapping happens.

Heck that's some good improvements - I'm a novice on the unic side and am still getting to grips with using telnet etc. on the raq - but getting there (bt anyone know of a good resource). As leat when my server is crawling I've found the top and kill command (usefull when you right code that is in a permanant loop!) - I've run the hdparm and got similar results :

/dev/hda:
multcount = 32 (on)
I/O support = 0 (default 16-bit)
unmaskirq = 0 (off)
using_dma = 1 (on)
keepsettings = 0 (off)
nowerr = 0 (off)
readonly = 0 (off)
readahead = 8 (on)
geometry = 39693/16/63, sectors = 40011300, start = 0

Timing buffer-cache reads: 128 MB in 2.63 seconds = 48.67 MB/sec
Timing buffered disk reads: 64 MB in 3.69 seconds = 17.34 MB/sec

But this is on a RAQ4 iwth 128mb ram / Top also shows a constant (approx) 100mb of ram being used, and when I sort by memory usage I always have at least 15 CaspEng using 11% memory each there? Is that ASP stuff? why so many?

Anyone tried the HD performance upgrade on the RAQ4 - Any other tip to increase performance? (phew!)

[Edited by Matthew.A on 03-08-2001 at 05:09 AM]