trueg50

What do you set for the core affinity for the i7's with HT?

Is it "1,3,5,7" for VM1 (2,4,6,8 for VM2), or "1,2,3,4" for VM1 (5,6,7,8 for VM2)?

EDIT: seems to be going well with the defaults (permissions error is not letting me change the affinity any ways).

Wow, I'm getting 7,540 PPD with the CPU alone! I never realized they performed that well, I was expecting no more then 3,000 ppd.

Looks like I'll pull in about 17,300 ppd with this plus my GPU!

__________________
Cable manage and liquid cooled Thermal take Armor

Synergy (G5 case mod)

MAD_J

Thx, I just edited my old config I didn't see the options before. Its all working great with 3 cores just as I wanted.

Im just having problems editing the amount of memory allocated to the VM, the option is greyed out.
Edit: Nm figured it out I had to edit the file.

__________________

wire

I might give this a try in the future.

__________________

MadCatMk2

Dude, bump.


Oh wait, not been 24 hours.

Great find mate!

__________________

Automated pausing of Folding@home.

k4m1k4z3

Im using this... had to remove WMware workstation but its worth it... now using 75%-90% CPU for SMP folding. the rest goes to GPU folding and various stuff running.

Great so far. Can anyone confirm that this is bringing them better ppd by only using 1 SMP client?

Mikecdm

Before I edited the file to support 4 cores, I was getting 4kppd with just 2 cores. With 4 cores I'm getting 8k. So it's pretty much the same except that it's in one VM. It also seems to work better for me when I'm folding with the gpu as well. The gpu takes a tiny hit and the cpu takes a small hit due to the gpu using up some resources. In the past, the gpu would take a giant hit and the cpu would take a hit as well.

I've noticed that depending on the wu that the gpu is working on, the smp client gets different ppd.

If the gpu is working on:

1888 = 8K + ppd for the cpu
787 = 7700 - 7800 ppd
353 = 7500 - 7600 ppd.

__________________
How to overclock and understand Nvidia shader straps

k4m1k4z3

Interesting... so the worse the GPU is doing on ppd, the better the CPU does?


The biggest advantage I see from using only 1 client is getting work units completed faster. I'm just too impatient to wait a day and half to get 1 unit completed.

XFreeRollerX

Holy crap this is going to be a RAM saver for sure

__________________
My Awesome HTPC/Server/Gamer Machine Computer related imagesFolding Stats REAL TIME Folding monitor!
Benchmarks Social Clubs and MORE!

DuckieHo

I've dedicated only 3 cores to SMP... this allows GPU Folding and general usage to have access to a free core...

__________________
To answer most of your questions: (1) a fridge cannot cool a PC (2) 64-bit OS for over 3.4GB (3) If a PCIe card fits, it should work (4) Resolution, not screen size (5) If you have a question, it is not news (6) Report, not respond to Spam (7) Single-Rail/Non-Modular PSUs are not always better than Multi-Rail/Modular

Mikecdm

It probably has more to do on the amount of cpu usage that is needed by the gpu. Certain wu's require more cpu usage. When i check the task manager while working on an 1888, cpu usage is 0-2%. Other wu's have 3-6% cpu usage.

Getting them done in 6 hr's is a lot nicer than 12hrs. You must have the cpu at stock clocks if it takes 36hrs to complete a wu.

Using 3 cores was my original plan. I took screen shots using 4 to show that it works. After letting it run for a while using all 4 cores and the gpu, it really didn't make an impact on ppd. I haven't tried using only 3 cores, but I'm positive that I get more total ppd using all 4 cores and the gpu. For basic use, I haven't noticed any type of slowing down.

__________________
How to overclock and understand Nvidia shader straps