If you want to crunch those WU's faster, try these clients.

http://www.marisan.nl/seti/index.htm

These clients have been compiled to run best with whatever processor instructions your CPU can support. They're amazing! My 2.8GHz Prescott P4/533 Mhz/1Mb L2 went from 3 hours per WU down to 1:20 hours! They work like a champ!

Crunch3r's apps are even faster

they can be found here http://www.guntec.de/Crunch3r/setix86.html

I agree with you about optimised clients from crunch3Rr.

Computer L1 cache L2 cache Floating point Integer memory Time/unitUn-optimised Time/unitOptimised

P4 640 3.2ghz Prescott 16k 2mb 2079 2934 1gb 1 h 45m 40m
P3 1000mhzCoppermine 32k 256k 970 2486 500mb 7h 30m 3h 45m
P3 800mhzCoppermine 32k 256k 691 1221 191mb 8h 12m 3h 45m
P3 600mhzCoppermine 32k 256k 527 920 128mb 10h 15m 4h 30m
P2 350mhzDeschutes 32k 512k 280 489 128mb 16h 30m 9h 30m


Best Regards,
Chris S.

I agree with you about optimised clients from crunch3Rr.

Computer L1 cache L2 cache Floating point Integer memory Time/unitUn-optimised Time/unitOptimised

P4 640 3.2ghz Prescott 16k 2mb 2079 2934 1gb 1 h 45m 40m
P3 1000mhzCoppermine 32k 256k 970 2486 500mb 7h 30m 3h 45m
P3 800mhzCoppermine 32k 256k 691 1221 191mb 8h 12m 3h 45m
P3 600mhzCoppermine 32k 256k 527 920 128mb 10h 15m 4h 30m
P2 350mhzDeschutes 32k 512k 280 489 128mb 16h 30m 9h 30m


Best Regards,
Chris S.

Well, I suppose that depends what version of his app you're using, and what OS it's on.

As far as I know, i'm using the latest (25th december 2005 release) for windows, with no problems except for the occasional app crash, but that's only 1 in every 1000 or so. For windows at least they're defenately quite a bit faster than TMRs

My tests are also based on the latest version for the AMD64 with SSE3 for windows. So i think that is not the case

[quote]
My tests are also based on the latest version for the AMD64 with SSE3 for windows. So i think that is not the case
i've heard there's only a very small gain with using SSE3 over SSE2, so you might want to try the SSE2 version

[quote]

Ah, i'm using SSE2 on a P4 and SSE on P3's
that's probably the reason

[quote]
So i'm still staying with the marsian one )

[quote]

So you found ETI?

(It's marisan, not marsian

[quote]
Thanks for the tip

I can't remeber which one i use but I think it's Marisan. I am running a 1.5 Ghz Celeron M processor with 1 Gb of ram. Without optimized client was around 2 hours 40 min. with is about 50 min.

The real time (stopped manually by clock) will always be a little more than pure cpu time.
Other threads/apps have to get some cpu time to do their work, like anti-virus, the BoincManager or BoincView, firewall, browser, e-mail clients, etc. etc.

Or do you say the reported time in the manager differs from the reported time on the result page at berkeley?
In that case I should ask: Do you use the calibrating core client by truXoft? His version manipulates the reported time and mflops for every wu to get a fair amount of claimed credit. So it is normal to get a different time on the result page than in the manager.


AndyK

Seems weird. Because I'm using Crunch3r's Athlon 64 client with SSE3 instructions (given mine has a Venice core, which added SSE3 instructions to it), and have had no such timing issues. Getting about 44 minutes/WU...

Mine's also the 12/25/05 version...

Can't explain the discrepency in results, unless it's dependent on a difference in core steppings in our respective A64s, or some other difference in the respective systems... Hmm

[quote] Seems weird. Because I'm using Crunch3r's Athlon 64 client with SSE3 instructions (given mine has a Venice core, which added SSE3 instructions to it), and have had no such timing issues. Getting about 44 minutes/WU...

Mine's also the 12/25/05 version...

Can't explain the discrepency in results, unless it's dependent on a difference in core steppings in our respective A64s, or some other difference in the respective systems... Hmm