From: liljeber@klaava.Helsinki.FI (Mika Pekka Liljeberg) Subject: Re: Benchmarking under Linux (was Re: New 486 Suggestions? Date: 7 Sep 1992 07:39:42 GMT
In article <1992Sep6.212556.14491@wuecl.wustl.edu> ets1@cec2.wustl.edu (Eric Thomas Stuebe) wrote:
> In article <1992Sep4.100706.12473@klaava.Helsinki.FI> torvalds@klaava.Helsinki.FI (Linus Benedict Torvalds) writes:
>
>>But processor speed can be very important under linux: not just for the
>>obvious user-level speedup, but due to better response to disk-drive
>>interrupts and the like. Faster machines may simply read the disk at
>>the full 1:1 interleave - with slower systems, it's possible that the HD
>>driver doesn't keep up, and you get the dreaded 1-block/rotation
>>syndrome, which really hurts when swapping. This problem is probably
>>especially notable on 386SX machines: the 386 interrupt handling is
>>inherently slow, but if the memory badwidth is further reduced by the
>>16-bit bus, interrupt response is probably ever worse.
>
>> Linus
>
> What would you guess that the odds are that this is actually happening to
> anybody? I'm running on the most basic Linux-capable system--16 Mhz 386sx,
> with just 2M of RAM--so I expect that I'm getting hit by it if it's happening
> to anyone. I plan on repartitioning my hard drive soon anyway, I could
> reformat with a 2:1 or even 3:1 interleave at the same time...but I thought
> that the days of 3:1 interleaves were gone forever when I moved into double-
> digit Mhz ratings.
I don't know about odds, but it's happening to me. I have a 25 MHz 386, at least
until my mother board upgrade arrives, and two RLL drives. Using RLL drives is
part of the problem, since 26 sectors/track makes for a lot tighter timing than
15 sectors/track. This wouldn't be a problem, if only my controller was buffered,
but, sadly, it isn't. The controller is speced to perform at 1:1 interleave, but
this is only possible, if the processor doesn't go off carousing, switching
processes and the like. Sigh. The boon and price of a real OS.
On a related note, I have patches for a track buffered hd driver, in case anyone
needs them. The paches bring my data transfer rates up from 56k/s (1 block/rotation)
to about 300k/s (5-6 blocks/rotation), which is a definite imrovement. I have
been using them for a couple of months now, with no problems, so they should be
bug-free. So far I've been unable to interest Linus in the patches (they're not
as clean as he would like, and I haven't had sufficient motivation to clean them
up. Besides, I expect the problem to go away with the mother board upgrade.).
However, if you would like them, drop me a mail.
> --
> Eric Stuebe, CS/EE at Washington University *in St. Louis*
> ets1@cec1.wustl.edu "I didn't do it! Nobody saw me do it--you can't
> estuebe@nyx.cs.du.edu prove nothin'! ... What was the question again?"
Mika