Thanks for your tests.
Can you send the real configuration you used for those tests and the real
results. i.e:
- Kernel version and configuration files
- Hardware descriptions(at leasts what is involved in the tests)
- Results (time or speed).
Just have to be aware that using tagged commands with bad hard disks
decreases performances.
On the other hand, your tests seems to be only sequential IOs.
A hard disk is normally optimized for random IOs as well as sequential IOs.
Please, be as accurate as possible.
Gerard.
On Sat, 2 Nov 1996 psychos@xeo.net wrote:
> Has anybody performed any real speed tests between the two?
>
> I did a quick test by timing how long it took to sync, write 100mb from
> /dev/zero to a file, sync, write 200mb from /dev/zero to another file to
> clear the first file from the cache (running with 192mb ram), and read
> back the first file, then rm the two and sync again to clean things up. I
> did this about 5 times for each one (performing it right after booting),
> and noticed that the 53c7,8xx driver was about 15% fast than the 53c8xx
> driver. I'm running with a 53c810 btw...
>
> It also seems very noticeably faster overall just using the system with
> the 53c7,8xx driver (which is what I normally use).