On 17-Dec-99 Nimrod Mesika wrote:
> Evgeny Stambulchik wrote:
>
> > As far as performance is considered, it's not RPM that usually matters
> > (inspite
> > of a big hype), but the amount of cache on disk. Until recently (1-2
> > years), the
> > maximum cache on IDE disks was 128KB, while about each SCSI disk had 0.5MB.
>
> How come? The OS cache is much bigger (at least several MB).
What's "OS cache"? The amount of memory reported by "top" as "cached"? Well,
it's adjusted dynamically by the OS. Not quite related to the question.
> I thought
> that a level-2 cache has to be bigger to have any effect?
Bigger than what? Level-2 cache is typically around 1MB (depends on hardware).
> If the block
> was not found in the OS cache, how can you expect to find it in the
> drive's cache?
If a drive has large cache of its own, it can communicate with controller (and
thus the OS) in bursts at the hypotetical 33/66 MB/s rate, accummulating data
(at the media I/O rate, ~10MB/s) between them; during these periods, CPU cycles
can be spent on other tasks.
> mmm.. maybe it's the read ahead and write back buffers that make the
> difference? A drive with a large buffer can get much more work done by
> itself without involving the OS.. maybe.
Exactly.
Regards,
Evgeny
____________________________________________________________
/ Evgeny Stambulchik <[EMAIL PROTECTED]> \
/ Plasma Laboratory, Weizmann Institute of Science, Israel \ \
| Phone : (972)8-934-3610 == | == FAX : (972)8-934-3491 | |
| URL : http://plasma-gate.weizmann.ac.il/~fnevgeny/ | |
| Finger for PGP key >=====================================+ |
|______________________________________________________________|
=================================================================
To unsubscribe, send mail to [EMAIL PROTECTED] with
the word "unsubscribe" in the message body, e.g., run the command
echo unsubscribe | mail [EMAIL PROTECTED]
