[Danometer]

On my older computers (PowerMac 7100/80), I have SETI 3 running on OS 8. On my iMac (Rev B), SETI 3 runs on OS 9.04. [I'm not brave enough to be a guinea pig for OS X.]

OS 9 runs SETI 3 most efficiently with the disk cache set at the maximum possible value. As a test to see how big a difference disk cache size makes under OS 8, I set one 7100 to maximum disk cache (6144K) and the other to minimum (96K). Contrary to what I expected, the smaller disk cache processed more in the same amount of time. Thinking this was a fluke, I repeated it two more times. Same results. The two machines are configured identically.

Is anyone else experiencing this? Granted, there is still a possibility that coincidence is the explanation (1 in 8 if Pascal's triangle is to be believed), but I wonder. Also, is there any possible explanation for this?

[Todd Madson]

Are you saying:
OS9 with smallest possible disk cache = slower.
OS9 with largest possible disk cache = faster.

OS8 with smallest possible disk cache = faster.
OS8 with largest possible disk cache = slower.

Is this correct? I don't have an OS 8 machine anymore, but am
running OS 9. I think I'll do some experimenting.

How much RAM are you running in the machine(s) in question?
What machines? What processors? What disk drives (speed/size?)

Thanks.

[Danometer]

That's what it appear to be. I'll experiment further tonight.

The specs on the two are identical:

PowerMac 7100/80 (PPC 601)
80 MB RAM (2 chips w/8 MB @ 80ns, 2 w/32 MB @ 60 ns)
OS 8.0 bare minimum install plus Peek-A-Boo and LibMotoSh
700 MB SCSI HD (aw, ain't it cute)
576 KB RAM Disk on each
SETI 3 on each

One is running at 96 K Disk Cache, the other at 7680K Disk Cache. They are not working on identical packets, but by Pascal's triangle there is a 7 in 8 chance that the differences are real, 1 in 8 that they are luck. The 1 in 8 chance is from the 50-50 chance of getting a more intensive packet three times in a row.

[reader50]

If RAM is tight on the 7100s, giving an extra 6 MB to the disk cache may have made things tighter for the rest of the OS. That would increase the VM load and slow everything down. The answer would be more real RAM.

[Edit] OK, you have plenty of RAM for just running minimum OS & SETI. Turn VM off, and go back to 2.04. I am milking 2.04 for as much as I can until 3.0 becomes mandatory. 3.0 seems to be generally flakey in a number of ways. Once the "upgrade" is mandatory, I will fool around with 3.0 some more under OS 9. But I will probably go back to the 2.4 CLI client under OS X. Avoid the 3.0 GUI client under OS X at any cost - it is a dog. At best, it gives 1/3 the speed of 2.04 under OS 9 and 1/2 the speed of 3.0 under OS 9. Usually a bit worse.

[This message has been edited by reader50 (edited 12-07-2000).]

[Danometer]

Mystery solved.

The one with the smaller disk cache setting was set to blank screen after 1 minute. The one with the larger disk cache setting was set to blank screen after 4 hours. Duh!

I'll check back tomorrow to see if it made a difference.