PPD # correct ?

M

mickeykool

Weaksauce
Joined
Oct 4, 2004
Messages
92
Back to folding to heat up my office in the basement. :D

I'm not sure if these numbers are correct thou.

i7 @3.4 - 2080 PPD per day
along w/
NV 260 - 7846 PPD per day

Shouldn't I be getting more PPD on the CPU than my 260?

Thanks
 
Are you running SMP on the i7? That looks more like regular than SMP. Post us a log of the CPU client (or post what WU it's working on)

 
That depends.....

on number of cores used....

with a i7 combined with a GTX260, (should be 8 cores with hyperthreading) you should start the client with

-smp X

Where X is number of cores - 1 (or 2)

so assuming 8 cores

-smp 7

now, stanford does not say that they support smp 7 (and other prime numbers) as it might become unstable, but in general it works fine. The client may detect the smp 7 being a problem and run smp 6

The limitation point wise for the setup is the fact that a GTX260 folds for the most PPD on GPU2 folding. And if you want to use all 8 cores, you need to be running "GPU3 -advmethods"

additionally to get the full points you need to set up a "passkey" see the sticky threads how to set that up, this will allow you to receive bonus points. For reference, my i7 2600K at 4.6 GHz gets right around 30,000 PPD using -smp 8, and not using the GTX260 for any folding at all.


 
--- Opening Log file [December 10 03:10:32 UTC]


# Windows SMP Console Edition #################################################
######################################################################
[03:10:32]
[03:10:32] Loaded queue successfully.
[03:10:32]
[03:10:32] + Processing work unit
[03:10:32] Core required: FahCore_a3.exe
[03:10:32] Core found.
[03:10:32] Working on queue slot 05 [December 10 03:10:32 UTC]
[03:10:32] + Working ...
[03:10:32]
[03:10:32] *------------------------------*
[03:10:32] Folding@Home Gromacs SMP Core
[03:10:32] Version 2.27 (Dec. 15, 2010)
[03:10:32]
[03:10:32] Preparing to commence simulation
[03:10:32] - Ensuring status. Please wait.
[03:10:42] - Looking at optimizations...
[03:10:42] - Working with standard loops on this execution.
[03:10:42] - Previous termination of core was improper.
[03:10:42] - Going to use standard loops.
[03:10:42] - Files status OK
[03:10:42] - Expanded 1771983 -> 1966776 (decompressed 110.9 percent)
[03:10:42] Called DecompressByteArray: compressed_data_size=1771983 data_size=1966776, decompressed_data_size=1966776 diff=0
[03:10:42] - Digital signature verified
[03:10:42]
[03:10:42] Project: 7142 (Run 0, Clone 42, Gen 387)
[03:10:42]
[03:10:42] Entering M.D.
[03:10:48] Using Gromacs checkpoints
[03:10:48] Mapping NT from 8 to 8
[03:10:48] Resuming from checkpoint
[03:10:48] Verified work/wudata_05.log
[03:10:48] Verified work/wudata_05.trr
[03:10:48] Verified work/wudata_05.edr
[03:10:49] Completed 107614 out of 500000 steps (21%)
[03:12:33] Completed 110000 out of 500000 steps (22%)

I changed the -smp to -smp 8, dunno if will make any difference.

I followed the guide for SMP setting and did get the bonus code as well.
 
Keep an eye on the time between percentage points. Since that wasn't a full TPF, let's assume 2:30. Punch that in here with WU 7142 selected and you're looking at almost 30k PPD. That's bang on for a 2600k at that clock.
 
Nice. Welcome back! Great time to fold again.
 
A 2:40 on those WUs would probably need an overclock of around 4.5GHz. (that is what I got on a 7151 and a 6994).

~doh, you have a 920, not a 2600. Too many processors out these days.
 
Last edited:
Good morning,

I switched to HFM and now I'm getting 16500 PPD on CPU and 7800 on GPU.

Thanks for your replies.
 
That sounds about right for an older I7 quad with a GPU leeching off of it. It looks like you have it set up correctly.
 
At 4.6GHz with two 460's leeching off it, I see 35k plus 11k plus 11k, respectively.
 
Yea just started and so far completed 4 or 5. I'm thinking of building another rig for this purpose (aka space heater) Should I pick up a 8120 bulldozer or what?
 
mickeykool said:
Yea just started and so far completed 4 or 5. I'm thinking of building another rig for this purpose (aka space heater) Should I pick up a 8120 bulldozer or what?
Click to expand...

No, the architecture underperforms... Look at Intel... 2500K or 2600K.
 
I wouldn't suggest a 2500K, mine at 3.7ghz gets 16k-20k ppd doing windows SMP.
 
Zink said:
I wouldn't suggest a 2500K, mine at 3.7ghz gets 16k-20k ppd doing windows SMP.
Click to expand...
My 2600k at 4.6ghz gets between 22 and 30k on SMP in windows, with regular bigadv going extinct the two will perform very similarly.

Mickeykool, I would keep an eye out in the dc fs/ft section for some of the 2p rigs that crop up cheap every few weeks. You will be able to get a 16+ thread machine for about the same as a 2500k rig.
 
You must log in or register to reply here.
Back
Top