Folding@Home Weekly Team News - 25th January 2007

OK.

Set up SMP on all suitable machines (actually, I'm eyeing up the M1710 atm) and ppds are as follows:

Monty (X2 3800+) = 866 (P3025)
Bart (Opty 170) = 1056 (P3025)
Maggie (X2 3800+) = 913 (P3025)
Sideshowbob (E6600) = 2X 931 = 1862 (2X P3025)
Homer (X2 4400+) = 966 (P3025)
Krusty (E6300) = 2X 1001 = 2002 :eek: (2X P3027)
Combined SMP PPD = 7665

Combined PPD from non-SMP rigs = 1492

Grand total as things stand at the moment = 9157 :D

BRING IT ON!!!

Wouldn't mind getting some more P3027 WUs, they seem to fold just as quick as the P3025s but give more points :cool:

I had to knock back the OC on the Opty due to stability issues but it's still managing 8 mins a frame - I'm sorely tempted to try a second client on her, she might just do it - might wait a bit though.

As I'm sure you'll understand, I'm really rather pleased at the moment.

And a special mention for Growse - see that speck in the distance getting further away from you? That's me, that is :p

Had a bit of bother setting up Ubuntu on the E6600. For some reason it didn't want to boot from the CD. I tried two different CDs and two different drives, to no avail. Even in VMware, it wasn't interested.
I ended up loading it in VMware from the ISO - which worked a treat (was actually faster than installing from CD on the other rigs).

Stan :)
 
That is awesome Stan. Good work mate! :D

And the difference between native ubuntu and 32 bit xp with vmare is fairly small. Was pulling up to 1430 ppd with native, about 1240 ppd with xp.

I don't know why the difference is. The only thing I can think of is the 32 to 64 bit changeover. But then again, does vmare even have a 64 bit version of the server?
 
Last edited:
Bigstan said:
Monty (X2 3800+) = 866 (P3025)
Bart (Opty 170) = 1056 (P3025)
Maggie (X2 3800+) = 913 (P3025)
Sideshowbob (E6600) = 2X 931 = 1862 (2X P3025)
Homer (X2 4400+) = 966 (P3025)
Krusty (E6300) = 2X 1001 = 2002 (2X P3027)
Combined SMP PPD = 7665

Combined PPD from non-SMP rigs = 1492

Grand total as things stand at the moment = 9157 :D
Click to expand...
:eek: That's some serious output for so few machines! Very nice indeed - should get you a few stomps in time.

shadowscotland said:
Cheesy fun for all, even some of the 'oldies' joined in (do I inclue myself in that statment :confused: ) most amusing.
Click to expand...
Ancient, for sure ;) :D
 
Dumped a few P3025s this afternoon and picked up a few P3027s so PPD will be even higher at the moment :cool:

Stan :)
 
joeyjojo said:
I don't know why the difference is. The only thing I can think of is the 32 to 64 bit changeover. But then again, does vmare even have a 64 bit version of the server?
Click to expand...
The lost performance is going towards the actual emulation environment and all the other things your system is doing with Windows. Think about, it's not just running Windows, it's running Windows and a whole other operating system all of which are vying for the same resources.
 
VMware seems to be very good at taking resources and keeping them.

I have 2 SMP clients running under VMware on my E6600 and it's using so much of the CPU, I couldn't manage to rip a cd with WMP this afternoon - it just seized up.

Rather impressive really.

Stan :)
 
joeyjojo said:
I stop it when doing anything else at all :)
Click to expand...

That rig is mostly for gaming and occasionally playing music. I don't mind stopping it to play games but it may make more sense to move the 5.1 speakers to the rig next to it to play music. That one only has one SMP client on it so should have enough CPU cycles left to play a bit of music :p

Stan :)
 
Bigstan said:
I have 2 SMP clients running under VMware on my E6600 and it's using so much of the CPU, I couldn't manage to rip a cd with WMP this afternoon - it just seized up.
Click to expand...
SMP is designed to use up to four cores per client on a multicore CPU - why do you have two running at the same time?
 
A.N.Other said:
SMP is designed to use up to four cores per client on a multicore CPU - why do you have two running at the same time?
Click to expand...

Because I can run 2 at the same speed, or faster than I can run 1 on most of my AMD rigs.

On the E6600, 1 client was running at 5m 40s a frame, 2 run at 9m 30 a frame. Some of my AMD rigs are running 1 client at 10m 30s a frame.

Stan :)
 
A.N.Other said:
SMP is designed to use up to four cores per client on a multicore CPU - why do you have two running at the same time?
Click to expand...
If you pull up the system monitor in Ubuntu you'll see that the maximum CPU usage for one SMP client is about 80% - when you run two clients that spare 20% is used aswell so you get slower times per WU but you are getting two finishing in less time than if you ran one after the other.

With most AMD machines this isn't really an option unfortunately as running two clients would push the completion past the preferred deadline and would lead to repetition of the same work (which is essentially wasted work).
 
Graph's looking good Stan :)

And smp has died for me. Get's stuck on "Entering MD". Reinstalled with no luck.
 
Last edited:
The graph will keep getting better too :cool: Take another look in the morning after Stanford's day ends ;)

Joey - was it working all right before or is this your first attempt?

Stan :)
 
All sorted now Stan. A common bug, something to do with me changing the machine name from localhost(?)

Any idea how many points an 805d at about 3.7 could get on smp?
 
Anybody ever seen this error before?:
[20:05:43] + Processing work unit
[20:05:43] Core required: FahCore_78.exe
[20:05:43] Core found.
[20:05:43] - Autosending finished units...
[20:05:43] Trying to send all finished work units
[20:05:43] - Error: Length of work/wuresults_01.dat (39227786) exceeds packet limit set (5241856)
[20:05:43] - Error: Could not transmit unit 01 (completed January 30) to work server.
[20:05:43] - 10 failed uploads of this unit.
[20:05:43] - Error: Length of work/wuresults_01.dat (39227786) exceeds packet limit set (5241856)
[20:05:43] Could not transmit unit 01 to Collection server; keeping in queue.
[20:05:43] + Sent 0 of 1 completed units to the server
[20:05:43] - Autosend completed
[20:05:43] Working on Unit 02 [January 31 20:05:43]
Click to expand...

It's given me a new WU to work on but I'd like to get this one home to papa.
 
You must log in or register to reply here.
Back
Top Bottom