[Danometer]

Has anyone successfully downloaded 9.1 yet?

If so, *PLEASE* post any and differences you notice versus 9 or 9.04, whichever you were using.

Those using LibMotoSh (like me), be prepared for LibMoto either making no difference or becoming a hinderance. Other compatibility issues may arise, but from my understanding, the nanokernel for 9.1 may include either LibMoto's processes or ones more efficient. *DISCLAIMER*: I could be way off target. I don't know if there will be a problem, but my best guess is that if there is, it will arise with LibMotoSh.

For those interested, iTunes is SoundJam Plus without an equalizer but with a CD-R burner program. The download is 2.8 MB. The program uses 6144 KB w/VM on (7326 KB w/ VM off), just 1 MB over SoundJam Plus 2.52. The iTunes extension won't load if a SoundJam extension is present, but iTunes can work off of the SoundJam extension. Oddly enough, the Creator Code is "hook".

[Scotttheking]

Hey, what did the equalizer do?

I'm trying to dl 9.1, no luck yet, apple's servers are probably overloaded, I'm gonna try again tomorrow.

In other news, I got 2 more SETI machines running

[Todd Madson]

Graphic equalizer lets you boost the bass and treble parts of the sound.
iTunes sounds curiously flat compared to Soundjam with the equalizer.

Originally posted by Scotttheking:
Hey, what did the equalizer do?

I'm trying to dl 9.1, no luck yet, apple's servers are probably overloaded, I'm gonna try again tomorrow.

In other news, I got 2 more SETI machines running

Try http://apple.doit.wisc.edu/Apple_Sup...tware_Updates/

[pcurtner]

Haven't really tested it yet, but 9.1 and 3.0 seems faster on my G4.

I'll take a closer look over the next day or two.

-p

[Todd Madson]

"seems" too vague - we'd need to do a benchmark as follows:

Ziff-Davis math benchmarks with 9.04 + Cache acceleration
Ziff-Davis math benchmarks with 9.04 + LibMotoSh + Cache acceleration

Ziff-Davis math benchmarks with 9.1 + Cache acceleration
Ziff-Davis math benchmarks with 9.1 + LibMotoSh + Cache acceleration

Then do a comparison with before/after. I think this could be a good
way of benchmarking speed differences in 9.1.

[druber]

of course, you'd also have to use the same wu for all 4 tests.

[Scotttheking]

SETI runs fine with it.
Don't know about speed really, but from watching it for 5 mins, seemed faster.

[Danometer]

I'm getting the same feeling, but I can't be certain. I'll run the current workunit with, then without LibMoto to test its efficacy. I'll test 9.04 vs 9.1 on my wife PowerBook (Pismo/400MHz/320MB) during the MLK holiday. (I got it for her for Christmas, I could have gotten a G4 if I had waited three weeks. )

I have 9.1 running on a iMac SE with 256Mb. If I load LibMotoSh everthing seems OK (Seti runs); except the Calculator will not run. Out of memory message. The Calculator is not a big deal in itself, but there must be other problems with LibMotoSh.

Seti 3.0 had been running about 10 1/2 hr per with 9.04 and LibMotoSh. 9.1 with 3.03 and LibMotoSh were on the way to 20hrs per. (That's why I tried use the Calculator) Will run now without LibMotoSh to see if things speed up at all.

[DocWest]

Does anyone have any proof of libmotosh doing any good?
it makes no difference on my iMac.

I read somewhere that OS 8.1 made it obsolete.

Oh yeah, seti and 9.1: no probs here.

[Danometer]

Test One Complete:

I didn't take this to completion because I felt that doing over half the packet would provide good enough number with less crunching time lost. These are calculated on a PowerBook G3/400 "Pismo" with 320 MB RAM (VM off), no RAM Disk, no LibMoto, only Netscape 4.76 running in the background, blank screen at 1 minute, screen sleep at 5 min, no hard drive sleep, same packet.

9.04: 68.518% completed in 12h59m48.6s CPU time => 5.279%/hr => Packet completion 18h57m
9.1: 68.576% in 12h14m0.5s CPU time => 5.606%/hr => Packet completion in 17h50m

Conclusion: OS 9.1 is 6.3% faster.

Exact benefit will vary, but I feel that all UMA (non-beige) G3 CPU Macs should experience similar performance gains. Some error is possible due to non-equal run time, but the error should not be large enough to counterbalance the calculated gain.

Next test: stick LibMoto in there and see if that helps or hinders.

[This message has been edited by Danometer (edited 01-13-2001).]

[Cosmo]

9.1 seems much faster than 9.04 on my 333 iMac w/ 256.

in 9.04 and seti 2.04 i finished a WU in 12ish hours, 3.0 did about 13h. My first (maybe second) 9.1 unit is about to finish and its time is about 9h 45 min!!! Maybe its just a faster unit, but if its not this is quite the speed increase!!

[Danometer]

These are calculated on a PowerBook G3/400 "Pismo" with 320 MB RAM (VM off), no RAM Disk, no LibMoto, only Netscape 4.76 running in the background, blank screen at 1 minute, screen sleep at 5 min, no hard drive sleep, same packet.

9.1: 68.576% in 12h14m0.5s CPU time => 5.606%/hr => Packet completion in 17h50m
9.1 w/ LibMoto: 68.309% in 12h6m15s => 5.643%/hr => Packet completion in 17h43m

Conclusion: OS 9.1 is 0.67% faster with LibMoto than without.

Definitely upgrade to 9.1 if you're running OS 9. (see above) LibMoto is a personal choice on 9.1, but I would recommend it if you are running 8.1 or earlier. A 7 minute difference on a Pismo 400 becomes almost half an hour on 601s and 603s.

Do not use LibMoto on G4s. The Altivec math functions are faster than LibMoto's. Your performance will be decreased by about 7%.

[This message has been edited by Danometer (edited 01-14-2001).]

[reader50]

I just returned a 3.00 result under 9.1 and got the "upgrade soon" nag screen. Looks like the clock is now ticking on 3.00.

Better run 3.00 for all it is worth. Who knows how much longer we have left.

[Scotttheking]

Well, I'm gonna upgrade to 9.1 as soon as I have time to download the files. My little 56k will take a long time.

And regarding upgrade soon on 3.0, I'm glad, means your times will finally be more in line with all of mine.