F@H - what client/method for this rig?

I should know better than to ask first, then read the manual!

I was asking how to use a quad core as well as GPU but worked it out.
Should be able to contribute 4 fast cores, 2 atom cores and a 5770 GPU.

Every little helps I guess.
 
What are you running on the atom cores, SMP client?

The amount of points produced by the 5770 probably arent worth the electric its using a ATi are very poor in terms of folding performance
 
The 5770 will get you a decent amount points (they're only slow compared to the Nvidia cards of the same cost). However the Atom is not worth folding on you will find it gets you very few point per day.

/My humble opinion
 
Yes, the atom performance is rubbish and I may take the folding client off.
The 5770 is getting more points than the CPU which surprises me. It's an opteron 1354 @2.2 but maybe that doesn't cut it in the folding world!
 
So this is how it is looking at the moment:
fahmon.jpg


I'm very green to this so am not sure I'm reading it right.
PPD is the overall measure of daily throughput?

I'm disappointed that the Opteron (ML115) is not that much better than the Atom (HAL). Is this just bad config on my part? It's maxing all 4 cores using the -SMP flag.

Would I be better running a Linux VM for the CPU and leave the GPU on Windows?

I know I'm never going to be returning big numbers, but would have hoped for more than I'm seeing at the moment.

Any help would be appreciated.

Cheers
 
Try using HFM.Net for monitoring as this will calculate the bonus you will recieve for the A3 units, that will up the ppd considerably, you should be seeing around 3-5k ppd for that cpu IIRC
 
I've installed that and I'm showing 1375 ppd. Something feels wrong but I don't know what.
 
Will do. Just reinstalled the CPU smp client and am waiting for it to update.
No, the server isn't doing anything other than folding and the 4 cores are maxed out.
 
What was the TPF on the system?

Reason being PPD will be lower further into the unit if theres been a an hour or more where the rig has been off
 
Time per frame

In HFM, highlight your quad core entry. Click view -> Benchmarks viewer.
That should leave a list like below

Could you post that please

Code: 
 Project ID: 6701
 Core: GRO-A3
 Credit: 921
 Frames: 100


 Name: Propus
 Path: C:\smp\
 Number of Frames Observed: 300

 Min. Time / Frame : 00:12:12 - 5,857.8 PPD
 Avg. Time / Frame : 00:14:18 - 4,616.0 PPD
 Cur. Time / Frame : 00:14:56 - 2,790.7 PPD
 R3F. Time / Frame : 00:14:56 - 2,790.7 PPD
 All  Time / Frame : 00:15:14 - 2,735.8 PPD
 Eff. Time / Frame : 01:18:23 - 531.7 PPD
 
Project ID: 6024
Core: GRO-A3
Credit: 474
Frames: 100


Name: ML115 CPU
Path: C:\Users\emailiscrap\FAH\
Number of Frames Observed: 2

Min. Time / Frame : 00:21:03 - 952.0 PPD
Avg. Time / Frame : 00:26:12 - 685.6 PPD
Cur. Time / Frame : 00:31:22 - 572.6 PPD
R3F. Time / Frame : 00:00:00 - 0.0 PPD
All Time / Frame : 00:26:12 - 685.6 PPD
Eff. Time / Frame : 00:30:02 - 598.1 PPD
Click to expand...
 
Those PPD values etc do seem to be off heres my benchmark stats from both CPUs Ive had this rig fold on

Code: 
Project ID: 6024
 Core: GRO-A3
 Credit: 474
 Frames: 100


 Name: Propus - Athlon II X4 @ 3600mhz
 Path: C:\smp\
 Number of Frames Observed: 200

 Min. Time / Frame : 00:05:06 - 7,982.7 PPD
 Avg. Time / Frame : 00:05:28 - 7,193.2 PPD


 Name: Regor - Athlon X2 @ 3600mhz
 Path: C:\smp\
 Number of Frames Observed: 8

 Min. Time / Frame : 00:11:34 - 2,337.2 PPD
 Avg. Time / Frame : 00:18:12 - 1,184.1 PPD

Theres no sensible reason why that should be folding slower than my old dual core, even with a 1400mhz handicap.

Can you try changing your -smp flag to -smp 4 so its forced to use 4 cores, and is the FAHCORE_A3 process at 100% or around about that?
 
I've changed the flag and will monitor.
The gpu exe is taking approx 20% of cpu resource.
 
that explains the poor frame times, try running it with -smp 3 for an hour or two. i would say the usage by the gpu is slowing things down as 3 cores are waiting for one to catch up.
 
You must log in or register to reply here.
Back
Top Bottom