sbw@naucse.UUCP (Steve Wampler) (08/29/89)
I've run Lenny's step rate exerciser on my system and thought
I'd post the results.
The system: 3B1 3.5MB memory, UNIX 3.5, 40MB internal (miniscribe??)
80MB external (CDC swift, 17.5ms)
WD2010 installed.
Internal Drive (maximum distance: 41,942,528)
Steprate 0 Steprate 14
long seeks: 9 7
random seeks: 6 4
converging seeks: 910 710
sequential seeks: 300 300
(Note: Since I ran these tests on my 'normal' system load, I've
rounded the times. All times represent the consensus of
several runs of 'testit'.)
External Drive (maximum distance: 79,035,904)
Steprate 0 Steprate 14
long seeks: 7 5
random seeks: 2 3!
converging seeks:1510 860!!
sequential seeks: 570 570
On the Swift, I'm assuming that the speed of random seeks is approaching
granularity. I may run that again with more iterations sometime. Note
that for the converging seeks test, the step rate gives a >40% improvement!
As expected, the sequential seek tests show no real difference.
Thanks Lenny, for posting the program! (And for the earlier posting
of ivfix!!)
--
Steve Wampler
{....!arizona!naucse!sbw}lenny@icus.islp.ny.us (Lenny Tropiano) (08/30/89)
In article <1668@naucse.UUCP> sbw@naucse.UUCP (Steve Wampler) writes: |>I've run Lenny's step rate exerciser on my system and thought |>I'd post the results. |> .. Thanks, I'm interested, since you have a nice FAST drive on your UNIX pc ... |>External Drive (maximum distance: 79,035,904) |> |> Steprate 0 Steprate 14 |>long seeks: 7 5 |>random seeks: 2 3! |>converging seeks:1510 860!! |>sequential seeks: 570 570 |> |>On the Swift, I'm assuming that the speed of random seeks is approaching |>granularity. I may run that again with more iterations sometime. Note |>that for the converging seeks test, the step rate gives a >40% improvement! ... We have to realize that the random seek is just that, random. Sometimes you will get results that are better, sometimes you will get results that are worse. It really depends on how step-rate 0 and step-rate 14's random numbers are ... If, for instance, the step-rate 0 is small seeks (using random) which take a minimal amount of time, and step-rate 14 are skewed towards the long seeks, then it would show the results being worse for step-rate 14. Not much can be extrapolated from the random seek, unless the step-rate 14 was the same "RANDOM VALUES". One could of course try this ... srand(same-seed) each time would produce the same values, I believe. |>Thanks Lenny, for posting the program! (And for the earlier posting |>of ivfix!!) You're welcome! Glad to hear some brave soul tried the ivfix! :-) I'm not as fortunate to have a drive with >1024 cylinders... -Lenny -- Lenny Tropiano ICUS Software Systems [w] +1 (516) 589-7930 lenny@icus.islp.ny.us Telex; 154232428 ICUS [h] +1 (516) 968-8576 {ames,pacbell,decuac,hombre,talcott,sbcs}!icus!lenny attmail!icus!lenny ICUS Software Systems -- PO Box 1; Islip Terrace, NY 11752