Hardware Canucks

Hardware Canucks (http://www.hardwarecanucks.com/forum/)
-   HardwareCanucks F@H Team (http://www.hardwarecanucks.com/forum/hardwarecanucks-f-h-team/)
-   -   FahCore17 for AMD is Here!! (http://www.hardwarecanucks.com/forum/hardwarecanucks-f-h-team/60017-fahcore17-amd-here.html)

SwingOP3 March 4, 2013 08:11 PM

FahCore17 for AMD is Here!!
 
Announced today is the beta release of FahCore17. It uses OpenCL so the AMD cards should show some muscle. I saw a post saying that a 7970 was doing ~47,000 ppd. They do use one cpu core so there will be some tuning to do but that goes for the NVidia cards as well.

More info here as well as the beta section of the Folding Forum.

G777 March 4, 2013 08:22 PM

1 Attachment(s)
I've never seen a gpu folding client use these many cpu cycles. This is on a GTX 670

great_big_abyss March 4, 2013 09:05 PM

Awesome! I set it up on a 5870 and a 6950. Let's see what they can do.

IRQ Conflict March 5, 2013 12:27 AM

So, F@H finally got serious about folding on ATI hardware again? After the Xseries debacle back in, what? 2007? We shall see.

Quote:

Originally Posted by SwingOP3 (Post 694370)
I saw a post saying that a 7970 was doing ~47,000 ppd.

That's nothing. I saw a post where the Titan was getting 307,392 PPD.

samhell44 March 5, 2013 04:14 AM

yea but a titan is not a normal card

sswilson March 5, 2013 04:22 AM

Quote:

Originally Posted by G777 (Post 694371)
I've never seen a gpu folding client use these many cpu cycles. This is on a GTX 670

This has been a discussion in the pre-beta comment thread (you had to have access to the file rather than just using the -beta tag) and from what I've gathered a big part of the cpu usage is related to debugging / troubleshooting code that will not be part of the public release.

So are these actually giving out points? The pre-beta testing wasn't.

I'm going to keep my fingers crossed though, not only does this core apparently bring AMD back into the fold, it also will make use of kepler GPUs. My 680 has been sitting idle while I wait for them to either come up with something that will use more than 80% of the cards power on more than just a few WUs and/or the public release of QRB for GPUs.

IRQ Conflict March 5, 2013 04:27 AM

Quote:

Originally Posted by samhell44 (Post 694416)
yea but a titan is not a normal card

6x the performance of a 7970? That is whats not normal. It's the software.

Looks to me that membership/active folders have decreased significantly. Must be why they are trying (once again) to bring ATI cards back into the fold. (sorry bout that, I had too).

G777 March 5, 2013 05:58 AM

Quote:

Originally Posted by sswilson (Post 694417)
This has been a discussion in the pre-beta comment thread (you had to have access to the file rather than just using the -beta tag) and from what I've gathered a big part of the cpu usage is related to debugging / troubleshooting code that will not be part of the public release.

So are these actually giving out points? The pre-beta testing wasn't.

I'm going to keep my fingers crossed though, not only does this core apparently bring AMD back into the fold, it also will make use of kepler GPUs. My 680 has been sitting idle while I wait for them to either come up with something that will use more than 80% of the cards power on more than just a few WUs and/or the public release of QRB for GPUs.

Interesting... the V7 client definitely shows points and a QRB bonus:

Project:7661 (5,0,2)
Base Credit:1600
Estimated Credit:6341.48
Estimated PPD:24030.88
Estimated TPF:3min 48secs

Previously though I was getting about 40,000 PPD on 762X WU's. With only ~2% cpu usage per client:sad:

Hopefully the cpu usage goes down on the new units; if as you say a large part of that is debugging

great_big_abyss March 5, 2013 06:14 AM

So, it looks like the new WUs are active on 6xxx series cards, as my backup 6950 currently in my main rig is doing about 12K ppd (more than double its usual output). However, 5xxx cards don't seem to be affected. The 5870 in my son's rig hasn't changed its PPD.

Also, the hex core in my son's rig is still pulling about 15K ppd, which is what it usually does. There obviously aren't any more cpu cycles being devoted to the video card in that rig. However, my main rig's 3570K has seen production fall from 25K to 15K. So, definitely a difference. I'll continue running the new work units on my main rig to see what they do...

sswilson March 5, 2013 06:46 AM

Quote:

Originally Posted by G777 (Post 694434)
Interesting... the V7 client definitely shows points and a QRB bonus:

Project:7661 (5,0,2)
Base Credit:1600
Estimated Credit:6341.48
Estimated PPD:24030.88
Estimated TPF:3min 48secs

Previously though I was getting about 40,000 PPD on 762X WU's. With only ~2% cpu usage per client:sad:

Hopefully the cpu usage goes down on the new units; if as you say a large part of that is debugging

Is that on your 670? Are you running any form of monitoring software (AB or Precision)? I'm wondering if the new core fully utilizes kepler cards (like the 762X WUs did) or if the power is only running around 80% (like it does for WUs writting for fermi like the 8XXX ones we've been seeing recently).


All times are GMT -7. The time now is 10:47 AM.