QMD Cores

dicecca112 · Apr 19, 2005

The last time my personal rig, and not some hold over backup rig was folding, QMD cores were not out yet. I'm curious as to if my TPF are right. Right now i am getting about 15:09 per frame on a p1912 ALA-dipeptide umb2. Is that normal should it be faster?

ChasR · Apr 19, 2005

Sig rig #1 is popping out p1910s @ a rate of 10:31/frame, 616ppd, 154ppd/GHz. Your's is putting out 142ppd/GHz which is about right considering memory latencies and FSB. Tracking these QWUs is a bit tricky because the frame time is quite variable. My number is a snapshot not an average so it may be a bit high, yours could be a bit low. One thing for sure is you can't trust what EMIII says the frame time is on a QWU, you have to look in the log.

dicecca112 · May 2, 2005

after I recieved my new mobo, I started hitting 10-11min per frame.

veryhumid · May 2, 2005

From the rig in my sig, I've seen them down to 13:something and up to 15:something. For some reason I remember them lower, but maybe that was a different QMD and i got tired of looking at the logs

ChasR · May 2, 2005

P4cs should fold one instance of p191x at a rate of around 143-150ppd/GHz. P4es should be in the 155-160ppd/GHz. If you get a p1900 or p1901 frame times will be much lower but points are lower too.

the garynator · May 2, 2005

ack chasr, have a bit of money sittin' in your rigs?!?!

psyshack · May 21, 2005

ChasR said:
P4cs should fold one instance of p191x at a rate of around 143-150ppd/GHz. P4es should be in the 155-160ppd/GHz. If you get a p1900 or p1901 frame times will be much lower but points are lower too.

Will a P4 2.0a @ 2.8 get qmd's and will it run them worth a fud. Cause all mine gettins is tinkers and gromacs.

ChasR · May 21, 2005

psyshack said:
Will a P4 2.0a @ 2.8 get qmd's and will it run them worth a fud. Cause all mine gettins is tinkers and gromacs.

I've got a P4a 1.5 @ stock turning out 200ppd on QMDs. Turn on bigpackets and -advmethods. You must have 512MB of ram. If you have integrated graphics, you'll have to share as little system ram as possible with the video subsystem. You ought to get 375ppd folding QMDs on your rig.

psyshack · May 21, 2005

ChasR said:
I've got a P4a 1.5 @ stock turning out 200ppd on QMDs. Turn on bigpackets and -advmethods. You must have 512MB of ram. If you have integrated graphics, you'll have to share as little system ram as possible with the video subsystem. You ought to get 375ppd folding QMDs on your rig.

Ok only 256 in that machine. More ram... lolololol thanks

muddocktor · May 22, 2005

Yeah, you can buy ram for cheap right now.

When I get my RMA'd VX back I think I'll throw it into my IC7-G machine so I can comfortably run QMD's on it.

psyshack · May 22, 2005

Ok add'ed a 256 stick for 512. Got a QMD 1295,, 450 pointer I thinks. Took 328 meg of ram and went to work. Its been running about a hour now and EMIII hasnt reported a thing. Look in log and its getting work done and progressing. Does EMIII not like QMDs?

dicecca112 · May 22, 2005

EMIII is weird with the qmds have the time its telling me its taking 2hrs per fram, when in reality its taken 22min per frame (running two clients, with one I get about 13-15min)

psyshack · May 22, 2005

dicecca112 said:
EMIII is weird with the qmds have the time its telling me its taking 2hrs per fram, when in reality its taken 22min per frame (running two clients, with one I get about 13-15min)

No kidding. It just posted at 4 of 100 with like a 4min per frame time. Then hit 100% and got confused. Now status unknown. Dont really care about the times and such. Just a tool to tell if machines are still folding. Now time per frame problem. lmao. Maybe it will sort itself out.

EMIII is soooo confused LMAO!!!

ChasR · May 22, 2005

preconvergence times are virtually untrackable. Usually convergence occurs by frame 7. After that time, frame times are pretty uniform on a decicated machine. The QMDs are p19xx.

ChasR · May 23, 2005

muddocktor said:
Yeah, you can buy ram for cheap right now.

When I get my RMA'd VX back I think I'll throw it into my IC7-G machine so I can comfortably run QMD's on it.

Muddoc,
I bought some VX for my IC7-Max3 and found the motherboard won't supply adequate stable vdimm to get latency below 2.5-3-3-6. It's a known issue with the IC7 (but one I forgot when I ordered the VX).

Audioaficionado · May 23, 2005

TCCD ram may not go as high and tight as VX but at least it's close and vdimm is very reasonable to hit its max bandwidth.

QMD Cores

dicecca112

Member

ChasR

Senior Member

dicecca112

Member

veryhumid

Member

ChasR

Senior Member

the garynator

Chief folding_monkey

psyshack

Member

ChasR

Senior Member

psyshack

Member

muddocktor

Retired

psyshack

Member

dicecca112

Member

psyshack

Member

ChasR

Senior Member

ChasR

Senior Member

Audioaficionado

Sparkomatic Moderator

Similar threads