WINSMP Frame Times?

I get around 12-13 min per frame depends what i do usally just surfing.
This is on 2 smp clients.

(awaits new parts for folding farm)
 
On the two rigs at 3.5GHz, running two WinSMP clients, I get about 11.5 min per frame.
On the rig at 3.0GHz, running two WinSMP clients, I get 13.5 min per frame.
On the rig at 2.7GHz, running two WinSMP clients, I get 14.5 min per frame.
On the rig at 2.4GHz, running two WinSMP clients, I get 18.5 min per frame.

All are Q6600s with 2 GHz RAM.
 
Q6600 with 2 win smp's and affinity changer @ 3.15ghz

Project : 2653
Core : SMP Gromacs
Frames : 100
Credit : 1760


-- C:\ --

Min. Time / Frame : 12mn 31s - 2024.82 ppd
Avg. Time / Frame : 15mn 02s - 1685.85 ppd
Cur. Time / Frame : 12mn 39s - 2003.48 ppd
R3F. Time / Frame : 12mn 39s - 2003.48 ppd
Eff. Time / Frame : 12mn 58s - 1954.55 ppd


-- F:\ --

Min. Time / Frame : 12mn 26s - 2038.39 ppd
Avg. Time / Frame : 14mn 07s - 1795.32 ppd
Cur. Time / Frame : 12mn 43s - 1992.98 ppd
R3F. Time / Frame : 12mn 43s - 1992.98 ppd
Eff. Time / Frame : 14mn 30s - 1747.86 ppd
 
ok thats good to know some times :) just wondering where I would be if I decreased my clock.. finally all nice and stable @ 3.825.. having to run more vcore than for primestable and is getting @ tad warm with my fans at only 40% or so
 
Pneumonic said:
ok thats good to know some times :) just wondering where I would be if I decreased my clock.. finally all nice and stable @ 3.825.. having to run more vcore than for primestable and is getting @ tad warm with my fans at only 40% or so
Click to expand...

i know my system is stable @ 3.33ghz but requires 1.5v's instead of 1.425v for 3.15ghz and @ 1.5v's it starts to get a bit toasty as well.

@ 3.15ghz it's only hitting low 50's with 2 win smp's.

@ 3.33ghz cause the extra 0.75v's it was hitting high 60's!
 
the B3 quad I had for a week hit +70c at 2.5 on (1.3v) at 75% load - passive cooling a B3 quad is not a good idea. I hit 100% load for a second before it the system shut down due to my coolant boiling :eek:

As the thermal design was lower that the hotest P4's I though it would work - I was wrong.
 
My office work station - [email protected]
Project : 2653 1x WinSMP + affinity

Min. Time / Frame : 20mn 45s - 1221.40 ppd
Avg. Time / Frame : 23mn 44s - 1067.87 ppd
Cur. Time / Frame : 20mn 48s - 1218.46 ppd
R3F. Time / Frame : 20mn 48s - 1218.46 ppd
Eff. Time / Frame : 21mn 25s - 1183.38 ppd

Home rig [email protected]

VMware/Ubuntu server - 10.5 mins a step - 2400ppd
 
Last edited:
Ok G/F has a 2180 rig @ 3ghz works out around 1610 ppd / 15 min 44 second frames

Just got home to check the Quad and has been stable all day :) so well happy about that getting around 5000ppd off the 2 clients ave frame time of 9mins 30 seconds

Edit: Thats both rigs using that core affinity thing
 
*Pulls Hair Out!!!!*

well... rig has processed 2 lots of 2 units and then throws an utter spaz at 60% through the next lot of 2.. decreased the clock :ekk: tbh this is damm annoying I have no idea why this is so unstable yet my other have no fuss what so ever... not like its even running hot compaired to a couple of the other rigs..

am getting very closed to sticking it under water.. as I cant run my fans much faster 24/7 or the misses is going to go mental

Edit:

For the record I am running at an average of 54/54/51/52 across my cores thats by no means hot?! (anyone else confirm their temps would be nice) and I have read that the instruction sets that the SMP client uses are different to Prime95 sets and such hense why a Primestable setup is not always F@H stable.. and that the F@H SSE instruction sets can cause high peaks in CPU coretemps that arent always reported by coretemp and the like!? But I do not see these being +15/20c spikes above full load temps? that would cause the core to fail or overheat? any ideas anyone?

Also anyone else notice that the afinity loader thing for SMP loads cores 1 and 2 more than 3 and 4? normally my coretemps are full load are even (within 1c) on average over a period of time.. So I can only assume that the difference I am seeing above is because the afinity loaderything is causing that as when I ran my first set of units through without it all my cores were within the 1c difference I expect whereas the last set and the set that got to 60% when I was using the afinity thing caused a bigger variance
 
Last edited:
does this look like an issue with the unit and not my clock??

hmm well looking back over my logs because I got another failure today when I got back in from work.. first unit last night looks like it did properly fail... second one appears to have been after a restart (fahmon was saying client hung but says that a lot! should have checked that really).. So only 1 of the units was a proper fail and it restarted that unit using standard loops.. this is the unit that has failed again.. last night it failed @ 60% and this afternoon @ 61% so maybe it is an issue with the unit and not the clock.. The other client has run through the whole day no issues on another unit will return that in 10 mins or so..
 
You must log in or register to reply here.
Back
Top Bottom