well, upgraded hdparm from 5.3 to 5.4... and finally got hard disk performance i have in knoppix (2.4.x kernel) on this slackware 9.1 with 2.6.4 kernel box. hd is fujitsu mpe3064at... hope someone will find this usefull...
and, command added after "Path... " line in /etc/rc.d/rc.S is
hdparm -a 512 -c 3 /dev/hda

before:
/dev/hda:
Timing buffer-cache reads: 128 MB in 2.12 seconds = 60.39 MB/sec
Timing buffered disk reads: 64 MB in 3.53 seconds = 18.13 MB/sec

/dev/hda:
multcount = 16 (on)
IO_support = 3 (32-bit w/sync)
unmaskirq = 1 (on)
using_dma = 1 (on)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 512 (on)
geometry = 13410/15/63, sectors = 12672450, start = 0


after:
/dev/hda:
Timing buffer-cache reads: 240 MB in 2.00 seconds = 119.72 MB/sec
Timing buffered disk reads: 54 MB in 3.04 seconds = 17.75 MB/sec

/dev/hda:
multcount = 16 (on)
IO_support = 3 (32-bit w/sync)
unmaskirq = 1 (on)
using_dma = 1 (on)
keepsettings = 0 (off)
readonly = 0 (off)
readahead = 512 (on)
geometry = 13410/15/63, sectors = 12672450, start = 0

I'm not quite sure why higher cache-reads
please you that much, the physical speed
in this second run was slightly lower?


Cheers,
Tink

not really... average is not lower, if i measure 3-4 times. hdparms benchmark is just orientational, not really precise

I still don't see a reason for excitement or
a recommendation to upgrade?


Cheers,
Tink

dunno.... "works as espected"?!?

What Tinkster is saying, is that an improvement in cache read isn't really much of an improvement at all. Physical read/write to the disk is the bottleneck, rarely the cache.

As you can see, your physical read speed is roughly the same.

--Shade

The size of the cache will make a bigger improvement. 8MB vs 2MB.

Where did you get those two values from?

And they have to do with hdparm how? : )



Cheers,
Tink