On Wednesday 15 August 2001 08:15 am, Naka Gadjov wrote:
> under UDA33-66-100
> UDA33 --- hdparm -X66 /dev/hda
> UDA66 --- hdparm -X68 /dev/hda
> UDA100 ---hdparm -X69 /dev/hda
This sets udma, but not ata. The two are not necessarily joined.
That said, my ata/100 at udma5 drive, 2.4.7 kernel, on a low inductance
80 wire cable, gets a steady hdparm -t of 35.96mb/sec in run level 3
every d*mn time, but with X runnin (level 5), -t is all over the place
at about 20 to 29mb/sec.
Like I always say tho, hdparm or any other HDD bench for that
matter, has little to do with real world day to day HDD performance.
Specially since most all of us run one window environment or another.
The fact that hdparm -t with X and some other apps/processes runnin
produces varried results, all of which are much lower than runlevel 3
produces illustrates this. Proof is in the puddin in that if I force
udma2 (-X66), hdparm-t scores are the same ... all over the place
between 20 to 29mb/sec with X in level 5.
> PS: use ide0=ata66 in kernel boot parameters to force using 80
> conductor cable.
I'll havt'a look into what that means. I've never passed an
ide?=ata?? parameter to the kernel. I'm skeptical of how it could
change somethin that's strickly a hardware issue, and I get 36mb/sec
without it. I do know that any HDD, whether SCSI or IDE ata/33, 66,
100, or 133 ..... still runs on the 33± mhz PCI bus.
BTW, an ata/100 capable cable has 80 wires .... but only 40
conductors. The other 40 are 'dummy' wires. Puddin again, there's
still only 40 pins on the ends of a 80 wire cable. The other 40 are
spread in between as seperators to lower signal inductance (noise).
A computer is just another kinda radio.
--
Tom Brinkman Galveston Bay
Want to buy your Pack or Services from MandrakeSoft?
Go to http://wwww.mandrakestore.com
