Top Genome-crunchers 11-18 march 2001.

[Liquid Snake]

Uh oh. I think they discovered our secret
http://www.dslreports.com/forum/rema...ding;mode=flat

[Rattledagger]

I think it's only folding they're actively trying to run multiple clients of in this thread.
If not, I see no reason to keep running genome if 5 copies on a P3-500 does all in 12-16 hours.

BTW, can some of you with nearly double production post your result compared to seti@home with one/two instances of genome running?

[This message has been edited by Rattledagger (edited 22 March 2001).]

[Mark F]

I tried to run a second client on my machine with a PIII@733, Win2K, 256MB. The 1st client slowed down and the second was very slow to filter. Several hours later it had only counted to 58 out of 70something. As soon as I stoped the 1st client the second one finished counting quickly. So no double production for me.

Bye-Bye Guyver. Just as I was getting close. I guess no 3Way dance. Even looks like rocketmanx, is starting to catch up again. So my grand plan of joining the dance and turning one machine back to S@H won't be happening.



------------------
Mark F. (A+ & Network+)
--------------------------------------------------
OH NO, my retractable cup holder swallowed a DVD...
and burped out a movie

[-=Narcissus=-]

Guyver,

Have you added more machines?? If not, there's your proof about the 2 clients vs 1... You're really flying now!!

I'm running 2 clients at the same time on my personal machine (C466), but it isn't without it's problems . I should be the only one running 2 clients.

Way to go guyv, I'll be trying to catch you this summer...

Fred

[-=Narcissus=-]

Haha!

Just noticed that OcUK's production dropped over the last few days... 6th place, here we come!

Fred

[Guyver]

As listed above, except that I decided to run 2 sessions on the 1Ghz system as well.

That's it....

I figured I'd slowly start to pull away, not that 35 degree angle offshoot according to the graphic...

Bring on the extra machines dude, it's all for a good cause...

(Note: Now I have to actually pay for the extra goodies I put into place this last week... )

For however short it lasts, it kind of feels good to be first at something... (Tried for close to 2 years with SETI)

MarkF - Have you tried tweaking your cache settings for the processor? I run my 866 and 1Ghz with a cache timing of 4. The dual 933 and 850 have no extra tweak settings available...

Narcissus - How many systems do you and your friends have running now?

Guyver

[-=Narcissus=-]

Computers.... Let's see...

(I'm not sure about everything, some are more than 1000km away! )
C466
C266@400
C300
CII-550
PIII-667
PIII-700@933
PIII-800
PII-400 (double CPU)

I might be missing one or two... Keep in mind, some of these computers aren't running g@h all the time, either being closed or on another OS.

Now if we could get things running behind a firewall, processor power would certainly double.

Fred

EDIT: oupsies

[This message has been edited by -=Narcissus=- (edited 23 March 2001).]

[Mark F]

<font face="Verdana, Arial, Helvetica" size="2">Originally posted by Guyver:
Have you tried tweaking your cache settings for the processor? I run my 866 and 1Ghz with a cache timing of 4. </font>

If you mean the L2 cache lataency, it's at 0.

The prossessor involved is a PIII-550@733 on a BX MoBo, memory is running 2-2-2. Both clients had there priority set to normal. Anything else you might need to know?

If this system could double it's out put, I will try it on others to help our cause!!

Mark F.

[Jorden]

Mark, try setting the priority to HIGH in Win2k. If you don't use your PC at that time for much else, like playing a heavy 3D game, your time should go up.
Also, take care you still have a good amount of swapfile around. Genome takes a pretty bit of swap file.

In Win9x/ME you have to set the CPU idle time to LOW, for at least One of the sessions you're running, if you want to do something with your PC

Jord.

[Guyver]

Mark,

You may be hurting yourself, by setting the L2 cache latency to 0 (I seem to have lost performance, when I set mine below 4) - ie - missed reads, re-reads.

You may want to play with the numbers, and see what shows optimal output...

Also, I wouldn't go higher than normal for Win2K priority. By going higher, you may cause system level processes (network/disk reads/writes) to suffer, again causing lower performance.

With my 3 sessions running on my 2k box, I leave them on the lowest setting, and they still get 33% each (Avg).

You may want to actually take them back to the lowest setting, instead of normal, and see what shakes loose. As long as nothing else is running, they'll get 99-100% of the processor anyway.

I also went through my services list, and shutdown any services not necessary for my installation.

( Not trying to disagree with you Jorden, I'm just running from my own experiences... If I'm wrong, my bad... )

Guyver


[This message has been edited by Guyver (edited 23 March 2001).]

[Guru]

Liquid Snake
What times do you get on that old pentium?

What times could I expect from a cel 400@450 and a Katmai 500535?



------------------
Join the MURC SETI team! | SETI @ MURC

According to the latest official figures, 43% of all statistics are totally worthless..

[Liquid Snake]

I dunno, I don't check on it. I just let it run. But I think the P133 takes about a week for 99 aa-length data block. The P166 probably takes 6 days for the same.

[Jorden]

Guru, my P3-450 with 256Mb RAM and WinMe takes about 24 hours to crunch 2 GU's

Jord.

[Guyver]

WooHoo - 4000 UNITS!!!!

1000+ Units in 4 days.... Yeah, Baby!!!

O.K. Maybe I'm a little psycho (With my girth, I'm not a little anything... ), but anyway, I just started GH#4 on my dual 933, and GH#3 on my 1Ghz...

Guyver

[This message has been edited by Guyver (edited 23 March 2001).]

[Rattledagger]

Jorden, if you set genome to HIGH on NT, you can't run any games the normal way.
I've seen some recommend changing the priority of games to HIGH, but if they stops responding it's nearly impossible to kill them. The only worse thing you can do is to set a process to REALTIME.

Guru, is it very cold in Finland, or is you using some special cooling on the Katmai 500535

As for run-time, it's just to re-format my equation to this:
[seti@home cpu-time * 0,3 * aa-length] / times-faster = run-time g@h
and you can calculate yourself.
Since no-one has posted any numbers, you can use 13,38 as times-faster.

[This message has been edited by Rattledagger (edited 24 March 2001).]