[Danometer]

SETI 3 is behaving a bit erratically. First packet using 3 took 20.3 hours. Second packet should be complete after 14 hours under similar conditions. Third packet, ¿quien sabe?

I am currently using a RAM Disk set at 1280 K, and that increased performance by 10-12% under SETI 2.04. I have "blank screen" set at 1 minute. What I would like are other modifications that I can easily make (ie, freeware/shareware) that would increase work unit crunching efficiency. A processor or computer upgrade is not possible at this time (not enough money).

I have an iMac Rev B (G3-233) 160 MB (VM off) running OS 9.04.

I have heard of LibMotoSh. I would like to know from where it can be downloaded.

I have heard of "cache overclocking". What is it, what does it do, and how is it done?

Any suggestions are welcomed. Thank you for your time.


[This message has been edited by Danometer (edited 10-05-2000).]

[wlonh]

LibMotoSh is here:
http://www.mot.com/SPS/PowerPC/suppo...g_macuser.html


powerlogix cache profiler is free, http://www.powerlogix.com/support/software.html

XLR8's Mach Speed Control is better, imo: http://xlr8.com/support/cpuupgrades/DL-MAChSpeed.html not sure if it is free anymore...

and, as ever, a search of this forum would have yielded this info



whatever, i know for a fact that i have posted the link to LibMotoSh more than once

[This message has been edited by wlonh (edited 10-05-2000).]

[Danometer]

Thank you for the information.

Searching for LibMotoSh on the MacNN forums produced a signal-to-noise ratio of almost zero. I read lots of complaints about LibMotoSh causing bridges to collapse and starting a nuclear war and lots of bragging about it being the best thing since sliced bread, but no hyperlinks for downloading. That's why I resorted to this.

I'll install LibMotoSh before beginning the third SETI 3 packet. We'll see what difference it makes.

[Misha]

I also noticed that the packets seem to speed up under 3.0... the first packet on the iMac DV took about 12 hours. The most recent about 8.5. More anomolies... but now the iMac is crunching units 2-3 hours faster than it used to with 2.04.

If only the other Macs with 1 MB of cache could do that too...

[Danometer]

Testing is finished.

The times for the standardized workunit available through Ars Technica on my iMac Rev B 160 MB RAM and SETI 3:

No modifications: 23 hours 15 minutes
With RAM Disk at 1280 KB: 20 hours 56 minutes
With RAM Disk and XLR8 MACh Speed Control setting cache speed at 155 MHz (3:2): 18 hours 9 minutes
With RAM Disk, XLR8, and LibMotoSh: 14 hours 17 minutes

The PowerLogix G3/G4 extension was too unstable to use, even at 117 MHz (2:1), which is the normal cache speed. It crashed on alternating application openings (open-crash-open-crash-you get the picture), even without LibMotoSh.

As yet, nuclear war hasn't started. (Search for "LibMotoSh" in the MacNN forums and you'll see what I'm babbling about.)

Keep on crunching…

[wlonh]

yeah, people make the strangest claims re: LibMotoSh...

and it is a 'cheap hack', etc etc, but it will not cause harm to your Mac regardless of what anyone has said/posted

i remember one other thing to possibly speed up SETI: quit all background processes and the Finder while running SETI, you can use freeware Process Guard to do this... available here: http://www.illenberger.net/fearonni/...cessguard.html

[OldGuy]

I have a slot loading iMac, and the tips suggested have not produced the results claimed by some.
Recently I had my iMac in for service due to spiderweb cracks in the case (BTW, Apple says they are not there. The service tech and I can both see them but Apple can't seem to acknowlege them.).I was using all the tips everyone suggested (LibMotosh, XLR8, turning off AppleTalk, removing RamDoubler, etc)
The iMac has been running for a little over a week with just the OS (less all the bells and whistles like Speech Manager, QuickDraw etc, etc.), SUM and Seti@Home. The work units were running at about the same speed.
Now over the past four days, I have been reinstalling all the stuff to my HD, the units haven't slowed down.
I would think if LibMotoSh, for example, actually provided an actual increase in speed, wouldn't Apple have used it in the OS by now? Its been around long enough that they should have noticed it by now.
I'm wondering if there is any real reason to disable or add anything to get more speed?
I do agree with blanking the screen after 1 minute and using the RAM Disk though. That cut my work unit times by about 45 minutes.
Just my two cents worth.

[Todd Madson]

Apple cannot use LibMotoSh as part of the operating system as it is a software product of Motorola, or so I understand.

They should use it as part of the operating system, but I don't believe they do.

[OoklaTheMok]

Hey Danometer could you submit those benchmarks to the teamlambchop page? For the 3.0 client my 2 times are the only ones on the list so far...getting lonely...