[Stealth Pyros]

Can someone take a look at my HFM and tell me if this looks pretty decent? I adjusted my GPU client to use a bit less CPU and came to this result in PPD for each client. My goal was to fold with SMP to take advantage of bonus points, while still being able to fold with my standard GPU client. Previously, I couldn't do both. I have an ATI, so when I ran both, my SMP client would get a ridiculous 500PPD while my GPU client earned about 3,000.

 

[Baldy]

Your CPU seems to be getting pretty decent PPD...can't say the same for your GPU though.

Have you checked out the winAFC guide? Should help improve PPD when folding on an ATI GPU as well as a CPU.

 

[aroc91]

Even with the SMP running, your 4870 should be a bit higher...

 

[Finrond]

Quote:

Originally Posted by aroc91 Even with the SMP running, your 4870 should be a bit higher...

He forced the GPU client to use a lower CPU percentage, which would probably explain the lower PPD.

 

[Stealth Pyros]

Quote:

Originally Posted by Baldy Your CPU seems to be getting pretty decent PPD...can't say the same for your GPU though. Have you checked out the winAFC guide? Should help improve PPD when folding on an ATI GPU as well as a CPU.

I did try that guide once before, and the command prompt that WinAFC opens gave me some sort of error. I was lazy and didn't go that route, but I'll probably check it out tonight.

Edit: Actually, using WinAFC would defeat the purpose of SMP; I'd run SMP on core 1 and then GPU on core 2. Using only 1 core for SMP won't let me finish the units on time for bonuses...

Quote:

Originally Posted by Finrond He forced the GPU client to use a lower CPU percentage, which would probably explain the lower PPD.

Joo got it dewd

I balanced it out a little more evenly and came up with this now, squeezing in 200 extra PPD. I think I might be done if everyone agrees that this is pretty good; it's the most PPD I've ever achieved on my sig rig.

 

[zodac]

Quote:

Originally Posted by Stealth Pyros Edit: Actually, using WinAFC would defeat the purpose of SMP; I'd run SMP on core 1 and then GPU on core 2. Using only 1 core for SMP won't let me finish the units on time for bonuses... Joo got it dewd I balanced it out a little more evenly and came up with this now, squeezing in 200 extra PPD. I think I might be done if everyone agrees that this is pretty good; it's the most PPD I've ever achieved on my sig rig.

Indeed it would.

Anyway, that PPD looks about the best you could hope for honestly. What does the 4870 get on its own?

 

[mmx+]

Quote:

Originally Posted by zodac Indeed it would. Anyway, that PPD looks about the best you could hope for honestly. What does the 4870 get on its own?

According to the first post, about 3000

 

[zodac]

I don't read first posts.

 

[Stealth Pyros]

Lol, yea it gets roughly 3,200 at max, so I think mixing SMP with GPU as I've done is the best. It seems like there's something going on with the servers though? I'm not able to work on neither client right now. GPU isn't getting its new unit, and SMP isn't continuing the unit I had. About to reboot and try again. SMP is stuck at "Using Gromacs checkpoints," I let it sit there all day while I was at work and still nothing when I got home.

 

[ablearcher]

Quote:

Originally Posted by zodac I don't read first posts.

And I don't read past the first page. I guess we are both tanked

Anyhow, great job, Stealth Pyros! For a HD4870, that is rather decent PPD. Which GPU client are you using? GPU2 or GPU3 (beta)?

 

[zodac]

Quote:

Originally Posted by ablearcher And I don't read past the first page. I guess we are both tanked Anyhow, great job, Stealth Pyros! For a HD4870, that is rather decent PPD. Which GPU client are you using? GPU2 or GPU3 (beta)?

Not quite; I don't need to read the first post. My input is pretty much consistent whether I read the title only, or the whole post.

You however, never seem to be quite on topic. It's embarrassing really. How can one person be so wrong, so often?

 

[ablearcher]

Quote:

Originally Posted by zodac Not quite; I don't need to read the first post. My input is pretty much consistent whether I read the title only, or the whole post. You however, never seem to be quite on topic. It's embarrassing really. How can one person be so wrong, so often?

Simple, I ignore what I don't like

 

[Stealth Pyros]

Quote:

Originally Posted by ablearcher And I don't read past the first page. I guess we are both tanked Anyhow, great job, Stealth Pyros! For a HD4870, that is rather decent PPD. Which GPU client are you using? GPU2 or GPU3 (beta)?]

Ummmmm... uhhh... hmmmm...

Huh?

I don't know, I assume GPU2.

 

[ablearcher]

Quote:

Originally Posted by Stealth Pyros Ummmmm... uhhh... hmmmm... Huh? I don't know, I assume GPU2.

GPU2, then. The GPU3 client is beta, right now. I guess the only benefit for ATi users was not needing the forcegpu flag (sometimes).

 

[Stealth Pyros]

Quote:

Originally Posted by ablearcher Simple, I ignore what I don't like

As do I, and then I go troll Apple threads... (lol, hit at all those who think I'm a troll for some reason)

GPU3: Doubt it has anything worth going through the trouble installing. I'd hope that the final release does offer some improvement on the CPU hogging issues they have.

 

[ablearcher]

Quote:

Originally Posted by Stealth Pyros As do I, and then I go troll Apple threads... (lol, hit at all those who think I'm a troll for some reason)

Ach, I only bother if I am really...

eh...

I'd rather not say.

EDIT: CPU3 is a nVidia beta, right now. But the beta client itself also recognizes newer ATi GPU, abating the need for the forcegpu flag.

 

[zodac]

Quote:

Originally Posted by ablearcher Simple, I ignore what I don't like

Quote:

Originally Posted by ablearcher GPU2, then. The GPU3 client is beta, right now. I guess the only benefit for ATi users was not needing the forcegpu flag (sometimes).

Doesn't apply to 4 series.

An example of above?

 

[ablearcher]

Quote:

Originally Posted by zodac Doesn't apply to 4 series. An example of above?

My HD4350 always needed the flag in GPU2, otherwise, no GPU would be "present"

Unless if you don't count the HD4350 as a HD4000 series card, mr. NVidia owner.

 

[zodac]

Quote:

Originally Posted by ablearcher My HD4350 always needed the flag in GPU2, otherwise, no GPU would be "present" Unless if you don't count the HD4350 as a HD4000 series card, mr. NVidia owner.

I just this Stanford heard about you (I had nothing to do with that

), so added that feature to the client; any 4350 in California needs to use the -forcegpu flag.

 

[ablearcher]

Quote:

Originally Posted by zodac I just this Stanford heard about you (I had nothing to do with that ), so added that feature to the client; any 4350 in California needs to use the -forcegpu flag.

Say all you want, the client will not even initiate without the flags