[drbott]

If the CPU monitor is any indication it looks like the new Mac OS X Client will kick some butt. It fully utilizes both CPUs of my dual 450. I'll report back in the morning about how it has done.


BTW, here is the URL for download provided by Ookla earlier.

ftp://serendip.ssl.berkeley.edu/pub/setiathome_macOSXPB_3_0.hqx

[Sjakelien]

Just downloaded SETI@home, and ran it. Considering the buzz-word compliance of OSX, I figured that I would easily run the program continuously 'in the background', because of pre-emptive multitasking.
No way, Jose. My Mac became virtually unusable, speedwise.
So, if I leave this software on my machine, it will be runing only when I'm not working on it.

[genevish]

I can't get it to install. I ope the installer package, enter my (admin) password, and get to the "Installing" screen. It then says "There were errors installing this software. Operation not permitted. Please try installing again. " If I go back one screen, then install again, it says "Preparing SETI@home_OSX" and just stays there. I have to force quit the installer. Trying to install again just repeats the first problem.

Oh well, guess I'l stick to the command line version. Seems to take less processor power anyway.

[mr_sonicblue]

Yea, there's something wrong with the installer app that it won't install anything unless you choose the OS X drive. (And it gets as you said if you go back, and try again.)

Just install to the OS X drive, then move it wherever you want.

[genevish]

Ahh, but I AM installing to the Mac OS X drive. And every other drive. All have the same problem.

[henryhbk]

Of course it says in the release notes "5) When asked to select a destination disk, you must select your OS X System disk. If you select a different disk, the installation will not work properly."


------------------
Henry Feldman
NYU School of
Medicine

[Dr Evil]

Has anyone noticed if the PB 3.0 client is slower than the text based one? On my B&W 350, its telling me that im 42% finished at 10 hours of cpu time. Using the text based one in the termninal, its around 9 hours for the whole work unit and about the same time under 9 running 2.4.

from what i'm seeing so far, i've done 35 percent in 6 hours cpu time on my dual g4/500. now i remember the 2.04 client doing a unit in 7-8 hours on my powerbook g3 (lombard). hows that?

i had both systems running in screensaver, blank screen in 1 minute. nothing else was running.

The version 3.0 changes page says: "The combined effect of FFT optimization, pulse detection, and the extended doppler drift range is that a typical workunit will take about 40% longer to complete with version 3.0 on any given platform."

Originally posted by fizzwinkus:
from what i'm seeing so far, i've done 35 percent in 6 hours cpu time on my dual g4/500. now i remember the 2.04 client doing a unit in 7-8 hours on my powerbook g3 (lombard). hows that?

i had both systems running in screensaver, blank screen in 1 minute. nothing else was running.

ahhh... thanks, agent! i thought i was going crazy

Anyone know if they are going to release a command-line version of 3.0 ???

BTW, having it running doesn't effect performance on my iMac. Try typing 'renice 20 <PID>' in the terminal, where <PID> is the process ID, found with 'ps aux'.

On my machine I could quite happily run SETI, play MP3s, Sherlock index my HD, download from the web, and play about in the terminal, all at the same time with no performance hit.

[Todd Madson]

FYI - On my AGP G4/400 it appears that the block time for my first block was about fifteen hours using the OS X PB graphical client. And this was selecting the "hide Seti @ Home" option so it was not plotting on screen.

The equivalent 3.0 client on the Mac OS 9.04 side would have generated a block in about ten hours. Still pretty slow, but better.

The 2.04 client generated blocks in the 5-6 hour timeframe for me consistently.

[Todd Madson]

Oh, in case you were wondering I did have it set to kick in and blank the screen after one minute. Thems the breaks evidently when it comes to the OS X client.

Anyone with a dual CPU unit have any results to report?

[Todd Madson]

One last post: if you're still having problems installing the client, just log in as root with the password for the account you set-up as admin and install the package at that point. Works like a charm.

[spicyjeff]

On my brand-spankin' new dual G4/500 its taking forever for work units to complete. its only at 40% with about 7 hours elapsed. I was getting about 5 hours per work unit with the CLI 2.4 version and that only used ONE of the processors.

I hope hey do some more optimizations...although it is neat ot see the Seti@home process using 180% of the CPU cycles in ProcessViewer.

[OoklaTheMok]

Got yer mac benchmark unit times here:

On the same machine (G3 500 with 512k cache):

2.04 GUI for Mac OS 9: 9 hr 20 min
2.4 CLI for Mac OS X: 9 hr 15 min
3.0 GUI for Mac OS 9: 7 hr 20 minutes
3.0 GUI for Mac OS X: 12 hr 24 minutes

As you can see, for the benchmark unit, the 3.0 client is much better on a smaller cache with the regular mac client. And there is something seriously wrong with the 3.0 Mac OS X client. Don't run it!

With the 3.0 client, cache size should not matter anymore. A 1 meg cache and 512k cache machine at the same clock speeds should get the same time. Not so with the 2.(0)4 client, where any cache 512k or below will show a performance hit. So 512k cache machines should get better times with the 3.0 client, and because of the extra science, 1 meg cache machines should get slower times.

[Misha]

Originally posted by OoklaTheMok:
and because of the extra science, 1 meg cache machines should get slower times.

This makes no sense. Why should a G3/500/512k work faster at SETI than a G3/500/1mb?

I still say berkeley screwed something up.

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

Someone asked a version 3.0 text only client for OSX: I emailed Seti@home and here's the answer:

Date: Thu, 05 Oct 2000 14:10:54 -0700
From: David Anderson
Subject: Re: Seti@home text-only client for MacOSX/Darwin
X-Accept-Language: en

Thanks for the email. We'll try to have a text-only OSX client soon.
David


>
> Dear Sir:
>
> I have been a Seti@home user for 1.4 years. I enjoyed donating my extra
> computer cycles to your project. I am writing to you to request that you
> update the version 2.4 Seti@home Rhapshody client to 3.0, or include a text
> only feature in the graphical MacOSX client.
>
> Although I am not sure you are the right person to write to, please forward
> this email to the appropriate.
>
> My proposal is this:
>
> 1) Allow an option in the v.3 Seti@home MacOSX Client to run as text app
> in the terminal program
>
> 2) Release an updated v.3.0 Seti@home client for Darwin.
>
> Thank for your time,
>

[gregvr]

Dual processors? Yep!

I ran a test yesterday when I was at work, and had it not connect when it was done with the work unit.

I got home (fairly late) and saw that it took 14.5 hrs of CPU time. Geez-- the new client must REALLY suck down CPU time.

But then I realized-- I left home at 7:30am and got back at 7:30pm-- how could it have possibly taken 14.5 hrs.

Then it hit me-- it really took 7.25 hrs-- but each processor is counted separately!

So I did another test this morning-- I started the OSX client at 6:55am, and checked it again at 7:05am. Sure enough, 20 minutes of CPU time.

Pretty cool, huh?

[OoklaTheMok]

Misha, I don't mean that a 1meg cache machine will be slower than a half-meg cache one. What I meant to say was that on a 1-meg cache machine, 3.0 will be slower than 2.04 or 2.4. Which may not be the case on a 512K cache machine because it would have a disadvantage behind a 1-meg cache machine on 2.04 or 2.4 but in 3.0 the disadvantage disappears.