• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

A few Questions

S

Syndicated_Death

[H]ard|Gawd
Joined
Feb 8, 2005
Messages
1,603
I started folding with the i7 listed below. I've been doing a fair amount of folding at work already. I Started the i7 just to see where it would stack up against my work machines (e7300 pulling about 1,500 PPD, and a Dual Xeon 5150 pulling about 3K PPD which also has an 8800gt pulling about another 3K PPD).

My GPU (the 6970 below) is pulling about 3500 PPD but my i7 itself is doing really poorly. about 800PPD. so lastnight I decided to kill the GPU client to see how the CPU did by itself. the CPU seemed to do worse when I got home and was just looking at the fah tracker V2. it would slowly go down in PPD.

with just the CPU running I'm getting 10K PPD. so my question is as follows. is the GPU client that reliant on the cpu? am I correct in assuming that I'm just better off running the cpu by itself? or is this an AMD CL issue?
 
Syndicated_Death said:
I started folding with the i7 listed below. I've been doing a fair amount of folding at work already. I Started the i7 just to see where it would stack up against my work machines (e7300 pulling about 1,500 PPD, and a Dual Xeon 5150 pulling about 3K PPD which also has an 8800gt pulling about another 3K PPD).

My GPU (the 6970 below) is pulling about 3500 PPD but my i7 itself is doing really poorly. about 800PPD. so lastnight I decided to kill the GPU client to see how the CPU did by itself. the CPU seemed to do worse when I got home and was just looking at the fah tracker V2. it would slowly go down in PPD.

with just the CPU running I'm getting 10K PPD. so my question is as follows. is the GPU client that reliant on the cpu? am I correct in assuming that I'm just better off running the cpu by itself? or is this an AMD CL issue?
Click to expand...

this is a stanford coding issue... the amd client absolutely eats the cpu... milky way and other dc projects that work on amd/ati gfx cards don't have that issue...

there is a manual fix for the gpu2 client which uses cal not opencl ... amd won't get opencl till the gpu client goes public.
 
yes, the ATI GPU client steals a fair amount of CPU power to run. that is why everyone who buys video cards for folding goes nvidia.

are you doing -bigadv on the i7? should be able to do around 30k ppd @4ghz on an i7 920. there was a SMP client update a few days ago. if you havnt updated, you wont get sent bigadv units. i had to delete the SMP folder (after it finished its WU) before it would download the new files.
 
what's funny is that I only saw the client using like 11-12% of the cpu time and yet the reality of what I was getting was drastic. The folding I'm doing on this computer is extra ppd for the horde not a dedicated boxen. it's my gaming PC otherwise I might have gone nvidia.
 
Your best option is to just drop the GPU client altogether and fold only on your CPU. You should be able to get an easy 25-30k PPD out of it. Even with the CPU usage tweaks, the ATI GPU client will use enough CPU power that your CPU PPD will drop by a larger amount than the increase you'll gain from folding on the GPU.
 
As others have stated, just turn off the GPU client as it isn't worth the electricity, and wear and tear for the time being. Once the ATI GPU3 client is finally released, it might be more worthwhile since the current ATI client is so unoptimized.
 
Where are you getting the 800ppd reading?

800ppd isn't a gpu issue. Its more a mis set variable.

I run -smp 7 and get 16k on my 4.0Ghz 950,.

Have you tried smp 7?
Have you set the gpu to slightly higher process.
 
Patonb said:
Where are you getting the 800ppd reading?

800ppd isn't a gpu issue. Its more a mis set variable.

I run -smp 7 and get 16k on my 4.0Ghz 950,.

Have you tried smp 7?
Have you set the gpu to slightly higher process.
Click to expand...

cant do odd numbers in the new smp client...

it rounds down... if you put smp 7 in you will be running 6 threads..
 
Patriot said:
cant do odd numbers in the new smp client...

it rounds down... if you put smp 7 in you will be running 6 threads..
Click to expand...
Proof? I've been using -smp 7 for a while and I didn't notice any changes once I updated to the new client version. The only thing I read indicated that certain known bad values for the -smp flag were disabled, but there were never any issues for -smp 7, nor any reason why it would have been disabled.
 
Zero82z said:
Proof? I've been using -smp 7 for a while and I didn't notice any changes once I updated to the new client version. The only thing I read indicated that certain known bad values for the -smp flag were disabled, but there were never any issues for -smp 7, nor any reason why it would have been disabled.
Click to expand...

go read the release notes...
this has already been discussed when a5 first came out...
 
Zero82z said:
I would if I could find them.
Click to expand...

yeah they kinda suck about that... I do believe its in the thread that announced the new core...

I will link that stuff when I get home... they may have been mean and not given the general public a heads up

you can always ask how the rounding works at folding forums...
 
Last edited:
Zero82z said:
The best information I found is this post, which mentioned "bad" values (-smp 7 is not a bad value): http://foldingforum.org/viewtopic.php?p=176116#p176116
Click to expand...

well try it and check task manager... ask the client if its a bad value ;)

if its not a "bad value" now.... it um might be in the future... cough... I have seen the rounding 7 to 6 in a4 core and bad value listed in a3>a5 and in another hidden place...
 
Code: 
[01:11:02] Project: 6901 (Run 7, Clone 21, Gen 4)
[01:11:02] 
[01:11:02] Entering M.D.
[01:11:08] Mapping NT from 7 to 7 
[01:11:11] Completed 0 out of 250000 steps  (0%)

Sitting right at 87.5% load, so -smp 7 definately works.

Of course, I just screwed myself out of 18% of a bigadv unit when i restarted this thing....
 
You must log in or register to reply here.
Back
Top