On Monday 09 April 2001 19:01, you wrote:
> On Mon, 9 Apr 2001, Penguin wrote:
> |If it is broken then how John is getting those throughputs. (only
> |difference is that he does not have i810 board ). That was
> |precisely the reason I wanted the version of hdparm on John's machine.
> |John .........
>
> hdparm-3.9-6
>
> hdparm only sets the stuff on the hard disk... it doesn't interact much
> with the kernel.. so there's no question of it being broken.
> if it's broken it wouldnt' be able to set the DMA and stuff.
Yes, but kernel 2.4.x, iam informed, has changed in a lot of ways in
the manner in which hardware results are reported. Yes, there may not be much
of kernel interaction, but its the kernel which is feeding those results to
hdparm AFAIK.
I am using the 4.x version of hdparm - no change.
> Would you by anychance have any hardware that generates a LOT of
> interrupts ?? some of the USB drivers are supposed to be badly implemented
> for webcams and stuff.
(only
> |difference is that he does not have i810 board ).
I think thats the over-riding difference and Kingsly's suggn fits with
another theory i got. i810 hogs RAM for its DRI, and maybe thats causing the
slowdown. I tried dropping the i810 module and ran a console only kernel -
did not help.
Rajesh, grab bonnie and run the test - compare it with someone with like
hardware higher throughput. Hopefully tomorrow i get my 80 cable.
Ashwin
(Glad that this thread is alive - i gave up on this 2 months back when
someone said hdparm is broken and demonstrated that with bonnie the results
were similar)
----------------------------------------------
Find out more about this and other Linux India
mailing lists at http://lists.linux-india.org/
