News on new OpenCL driver...

Moderators: mhouston, Site Moderators, PandeGroup

Re: News on new OpenCL driver...

Postby k1wi » Fri Apr 01, 2011 2:26 am

I guess it's possible that the core is being sent out to 4xxx cards now that it's on adv?
Image
k1wi
 
Posts: 1152
Joined: Tue Sep 22, 2009 10:48 pm

Re: News on new OpenCL driver...

Postby gwildperson » Fri Apr 01, 2011 4:42 am

k1wi wrote:I guess it's possible that the core is being sent out to 4xxx cards now that it's on adv?


Possible, but not likely. They said performance was really bad, which I interpret to mean that OpenCL 1.0 is inferior to core_11.
gwildperson
 
Posts: 726
Joined: Tue Dec 04, 2007 8:36 pm

Re: News on new OpenCL driver...

Postby k1wi » Fri Apr 01, 2011 5:03 am

gwildperson wrote:
k1wi wrote:I guess it's possible that the core is being sent out to 4xxx cards now that it's on adv?


Possible, but not likely. They said performance was really bad, which I interpret to mean that OpenCL 1.0 is inferior to core_11.


Given his log and I think another user experiencing the same thing in the Core16/ATI P11293 thread, it is possible that the assign is wrong and Stanford is unintentionally assigning them to 4xxx cards when they have adv methods or beta enabled...

That is, rather than them "manipulating" the files.

I'm 10 hours from my bigadv finishing, but I can try firing up my 4770 to see if I catch one of the work units.
k1wi
 
Posts: 1152
Joined: Tue Sep 22, 2009 10:48 pm

Re: News on new OpenCL driver...

Postby Scandalman » Fri Apr 01, 2011 6:24 am

My 6 series cards are still being assigned core 11 WU's even though I am running 11.2 and 7.1.21. I know that not everyone is receiving core 16 WU's from the prof's remarks so this is an observation not a complaint. I have another machine with dual 5750's installed which has been updated to catalyst 11.3, will update it's client to 7.1.21 today and see if I am lucky. I am not going to bother installing my new hardware until I see that it will be used (effectively).
User avatar
Scandalman
 
Posts: 45
Joined: Fri Jan 14, 2011 11:38 am
Location: England

Re: News on new OpenCL driver...

Postby Scandalman » Fri Apr 01, 2011 7:18 am

Oh well, my 6950 (with disabled shaders enabled) and two 5750's are now running Catalyst 11.3/FAH Client 7.1.21 and... (FX: Tadaaaaa!) core 11.
User avatar
Scandalman
 
Posts: 45
Joined: Fri Jan 14, 2011 11:38 am
Location: England

Re: News on new OpenCL driver...

Postby 7im » Fri Apr 01, 2011 7:30 am

Remember the limited number of core 16s going out?

Just cause you got the hardware doesn't mean you'll get the 16s.
How to provide enough information to get helpful support
Tell me and I forget. Teach me and I remember. Involve me and I learn.
User avatar
7im
 
Posts: 14648
Joined: Thu Nov 29, 2007 4:30 pm
Location: Arizona

Re: News on new OpenCL driver...

Postby p0wer » Fri Apr 01, 2011 8:28 am

gwildperson wrote:Are Core 16 WUs being assigned to a 4850? Apparently you manipulating the files and you don't trust Vijay to be telling the truth. :evil:

Please read the whole topic again, then this sentence:
p0wer wrote:Did that because core 16 WU was indeed a bit slow on my Radeon HD4850 and I want to go back to the old core.

YES, I have HD4850 and I've been assigned a core 16 wu. Errors do happen, ya know, especially when there's a pressure from people. Besides, how and why am I supposed to trick F@H servers to assign core 16 wu to my card?
User avatar
p0wer
 
Posts: 70
Joined: Fri Mar 18, 2011 7:46 am
Location: Poland

Re: News on new OpenCL driver...

Postby k1wi » Fri Apr 01, 2011 10:41 am

@ p0wer: I highlighted the possibility that this might be the case in my previous post, that HD 4xxx's may unintentionally being assigned core16's by Stanford if they have beta/adv methods enabled (not the user's fault). I had a bit of a chat and have since been advised that the right people have now been contacted regarding this (this may have slipped through their filtering/threads their monitoring), but obviously it's the middle of the night in California so any resolution will need to wait until they're back on board.

k1wi
k1wi
 
Posts: 1152
Joined: Tue Sep 22, 2009 10:48 pm

Re: News on new OpenCL driver...

Postby gwildperson » Fri Apr 01, 2011 5:21 pm

Sorry if you took my question to mean you were somehow cheating. That was not my intention. I was just trying to sort out if this was an assignment error on the part of Stanford or something else. I see that it has now been sorted out and perhaps fixed by now. In any case, it's easy to remove "advanced" and not ask the server for late-beta assignments.
gwildperson
 
Posts: 726
Joined: Tue Dec 04, 2007 8:36 pm

Re: News on new OpenCL driver...

Postby p0wer » Fri Apr 01, 2011 8:46 pm

gwildperson wrote:In any case, it's easy to remove "advanced" and not ask the server for late-beta assignments.

Isn't it?

But once you get a WU you don't want for any reason, it's not that straightforward to remove it.
User avatar
p0wer
 
Posts: 70
Joined: Fri Mar 18, 2011 7:46 am
Location: Poland

Re: News on new OpenCL driver...

Postby 7im » Fri Apr 01, 2011 9:42 pm

reference: dump
User avatar
7im
 
Posts: 14648
Joined: Thu Nov 29, 2007 4:30 pm
Location: Arizona

Re: News on new OpenCL driver...

Postby p0wer » Sat Apr 02, 2011 7:54 am

I just thought that if there are plans for cutting GPU2 client and there's fahcore_16 being already tested, maybe we can get some quick return bonus for GPU2 units? 8-)
User avatar
p0wer
 
Posts: 70
Joined: Fri Mar 18, 2011 7:46 am
Location: Poland

Re: News on new OpenCL driver...

Postby 7im » Sat Apr 02, 2011 5:54 pm

QRB started in SMP, and was recently added to CPU client. It's not unrealistic to expect that for all clients.
User avatar
7im
 
Posts: 14648
Joined: Thu Nov 29, 2007 4:30 pm
Location: Arizona

Previous

Return to ATI specific issues

Who is online

Users browsing this forum: No registered users and 1 guest

cron