Re: disk benchmarks & linux

Martin von Loewis (martin@mira.isdn.cs.tu-berlin.de)
Sun, 14 Sep 1997 10:55:27 +0200

Messages sorted by: [ date ][ thread ][ subject ][ author ]
Next message: Peter L Jones: "Fw: Another fix for 2.0.31p9 smbfs"
Previous message: Ricky Beam: "fmount system call, second attempt"
Maybe in reply to: Larry McVoy: "disk benchmarks & linux"

> So I'm polishing up yet another whizzy benchmark and I'm getting screwed,
> somewhat, by the fact that Linux doesn't have raw devices so the buffer
> cache can throw me off when I'm doing I/O on /dev/sda or something similar.
>
> Is there any way to force the data out of the cache other than to try and
> use up all of memory?

When you open a block device with O_SYNC, data is guaranteed to be on disk
before write(2) returns. AFAIK, there is no way to prevent read
caching on Linux.

Regards,
Martin

Next message: Peter L Jones: "Fw: Another fix for 2.0.31p9 smbfs"
Previous message: Ricky Beam: "fmount system call, second attempt"
Maybe in reply to: Larry McVoy: "disk benchmarks & linux"