Messages in this thread | | | Date | Sun, 21 Mar 1999 22:00:53 +0100 (CET) | From | Jelle Foks <> | Subject | Re: 2.2.3, VIA, IDE, performace -> use bonnie |
| |
On Sun, 21 Mar 1999, Mark Lord wrote:
> Jelle Foks wrote: > > > > On Sat, 20 Mar 1999, Meelis Roos wrote: > ... > > > Now the interesting part: running hdparm -t /dev/hda again (and again > ... > Ooops.. somebody probably used "-T" instead of "-t" that time.. > .. > > Though probably even bonnie can be improved, it's much better than hdparm > > for quick tests. > > But they measure different things. Apples and Oranges. > > Bonnie measures *system* I/O performance. > hdparm measures sequential sustainable disk data transfer rates. >
True. However, in practice, when run on ext2fs, with a reasonably fast cpu (celeron 300a or more), with any UDMA disk that I've seen, bonnie gives you the sequential sustainable disk data transfer rates too when run on a 2GB file.
You can see whether the bottleneck is the CPU or the disk by looking at bonnie's cpu utilization results. The only time I can think of to use hdparm instead of bonnie is when you're doing block device development and are testing if your code has any throughput problems
I think hdparm could be much more useful if it would scan the entire disk and produce the data for a graph of drive speed vs position on the disk.
Cya, Jelle.
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |