Problems running ATI/OpenCL 7.00 apps - report here

Message boards : SETI@home Enhanced : Problems running ATI/OpenCL 7.00 apps - report here
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 · Next

AuthorMessage
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45224 - Posted: 15 Mar 2013, 20:28:28 UTC

First task is error (7)
It didn't run for many seconds.

http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=13175882



The other 3 tasks runs at 17% and ticking
ID: 45224 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45225 - Posted: 15 Mar 2013, 21:07:39 UTC

Got me a couple of more erros...just a couple.
iteration=12 and using all 3 gpu's doesn'y seem wise.
Rig is not stable...

running 2 now with iteration=6 and that seems to work.

ID: 45225 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45226 - Posted: 15 Mar 2013, 21:14:09 UTC
Last modified: 15 Mar 2013, 21:18:14 UTC

Rig isn't stable.
ID: 45226 · Report as offensive
Grumpy Swede
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1700
Credit: 13,216,373
RAC: 0
Sweden
Message 45227 - Posted: 15 Mar 2013, 21:15:08 UTC - in response to Message 45225.  
Last modified: 15 Mar 2013, 21:20:06 UTC

Got me a couple of more erros...just a couple.
iteration=12 and using all 3 gpu's doesn'y seem wise.
Rig is not stable...

running 2 now with iteration=6 and that seems to work.



The errors you got there hasn't anything to do with your GPU's, or the new Beta release Ver 7.00. They would come even on the CPU's.

It's just resends of the "empty" (no telescope data, only headers data) +20K files, giving "SETI@home error -7 Garbled encoded workunit", discussed in the last 8 messages in this thread:

SETI@home 6.99 (aka x41zb) for Windows NVIDIA released
ID: 45227 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45228 - Posted: 15 Mar 2013, 21:19:14 UTC - in response to Message 45227.  

Got me a couple of more erros...just a couple.
iteration=12 and using all 3 gpu's doesn'y seem wise.
Rig is not stable...

running 2 now with iteration=6 and that seems to work.



The errors you got there hasn't anything to do with your GPU's, or the new Beta release Ver 7.00. They would come even on the CPU's.

It's just resends of the "empty" (no telescope data) +20K files, giving "SETI@home error -7 Garbled encoded workunit", discussed in the last 8 messages in this thread:

SETI@home 6.99 (aka x41zb) for Windows NVIDIA released


Well, I got me a nice blue screen for that....
ID: 45228 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45230 - Posted: 15 Mar 2013, 21:37:45 UTC
Last modified: 15 Mar 2013, 21:40:51 UTC

Then it is this serie of tasks that makes my comp very unstable...

12mr10ac.20754.67.9.14.25_3

I just aborted 25 of thoose tasks...

NNT now and run cache dry.
I'll try to run 3 tasks at once again.
Wish me luck.
ID: 45230 · Report as offensive
Grumpy Swede
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1700
Credit: 13,216,373
RAC: 0
Sweden
Message 45231 - Posted: 15 Mar 2013, 21:44:25 UTC - in response to Message 45230.  
Last modified: 15 Mar 2013, 21:45:08 UTC

Then it is this serie of tasks that makes my comp very unstable...

12mr10ac.20754.67.9.14.25_3



Find the +20K files in the Beta directory ( there are a few legit 20K ones too, the 20K ones with the same name, but different extensions as the currently running WU's, should not be touched)

After you found them on the disk, identify them in your Boinc manager, and abort them. Deleting them from the disk directly will only result in them being resent to you again.

Find them, Flog them, Forget them. :-)

Edit: Good, you killed them :-)
ID: 45231 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45232 - Posted: 15 Mar 2013, 22:13:05 UTC - in response to Message 45231.  

Then it is this serie of tasks that makes my comp very unstable...

12mr10ac.20754.67.9.14.25_3



Find the +20K files in the Beta directory ( there are a few legit 20K ones too, the 20K ones with the same name, but different extensions as the currently running WU's, should not be touched)

After you found them on the disk, identify them in your Boinc manager, and abort them. Deleting them from the disk directly will only result in them being resent to you again.

Find them, Flog them, Forget them. :-)

Edit: Good, you killed them :-)



Yep.

The errors where just errors.
The blue screen I got from too low iteration when running on all 3 gpu's
I am back to iteration=20 again and 3 tasks.

Thanks for the information regarding the error 7's

ID: 45232 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45234 - Posted: 15 Mar 2013, 22:33:00 UTC

I dunno...

3 tasks iteration=20
And another bluescreen.
running 2 now with iteration=8
See if that works.

ID: 45234 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45236 - Posted: 16 Mar 2013, 7:10:00 UTC - in response to Message 45234.  

I dunno...

3 tasks iteration=20
And another bluescreen.
running 2 now with iteration=8
See if that works.



Did not, now reinstalling driver....
ID: 45236 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45237 - Posted: 16 Mar 2013, 8:05:33 UTC

Now running Cat 13.1 3 tasks on 3 gpu's, iteration=20 and all seems stable.

ID: 45237 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45238 - Posted: 16 Mar 2013, 8:14:25 UTC

I am noticeing a frequent use of Cpu 100% from time to time.

Is part of calculation being done on cpu as with ap tasks??
Or is this something with 13.1 driver?
ID: 45238 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45239 - Posted: 16 Mar 2013, 9:59:17 UTC
Last modified: 16 Mar 2013, 10:34:54 UTC

13.1 seems to work fine....

But tasks are running partially 100% on cpu then low% on gpu.
And then they are running 60% on gpu and very low % on cpu.
All the time on all 3 cores.

Is that supposed to be so?
ID: 45239 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45240 - Posted: 16 Mar 2013, 11:12:32 UTC
Last modified: 16 Mar 2013, 11:16:32 UTC

Another observation is that 2 GPU tasks on the ATI 5970(2GPU-card) is using only 1 cpu core when it should be using 0.977 x 2 cpu cores.

It's when I free cpu cores I can see this. Running x cpu tasks which decrease and increase when changing number of cores.


Conclusion HD app is taking almost twice the runtime to complete with 13.1 driver.
But again, I don't get driver restarts or bluescreens, yet...
ID: 45240 · Report as offensive
Profile Mike
Volunteer tester
Avatar

Send message
Joined: 16 Jun 05
Posts: 2530
Credit: 1,074,556
RAC: 0
Germany
Message 45243 - Posted: 16 Mar 2013, 11:29:06 UTC - in response to Message 45239.  

13.1 seems to work fine....

But tasks are running partially 100% on cpu then low% on gpu.
And then they are running 60% on gpu and very low % on cpu.
All the time on all 3 cores.

Is that supposed to be so?


Thats the drivers.

With each crime and every kindness we birth our future.
ID: 45243 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 45245 - Posted: 16 Mar 2013, 11:44:20 UTC - in response to Message 45239.  
Last modified: 16 Mar 2013, 11:46:34 UTC

13.1 seems to work fine....

But tasks are running partially 100% on cpu then low% on gpu.
And then they are running 60% on gpu and very low % on cpu.
All the time on all 3 cores.

Is that supposed to be so?


Our inner tests showed that Cat 13.1 hardly can be preferred vs Cat 12.8 or Cat 12.1 for example.

Number of period iterations for PulseFind set to:20
increase in this param will make app running longer though stable, in case that PartialPulseFind kernel takes too long (it can be another kernel that causes driver restart, then changes in this param will give nothing but slowdown).

Maximum single buffer size set to:256MB
Change in this param can affect both performance and stability.
Try 64MB, 128MB.

Also, as I understand, you trying to run "stock" app with non-stock params.
App tweaking comes in "anonymous platform" usage and not the primary topic in this thread. Here we find conditions where stock config works and where it doesn't work (to exclude such hosts).
ID: 45245 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45246 - Posted: 16 Mar 2013, 12:58:36 UTC - in response to Message 45243.  

13.1 seems to work fine....

But tasks are running partially 100% on cpu then low% on gpu.
And then they are running 60% on gpu and very low % on cpu.
All the time on all 3 cores.

Is that supposed to be so?


Thats the drivers.


So this app isn't ready for 13.1?
ID: 45246 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45247 - Posted: 16 Mar 2013, 13:06:52 UTC - in response to Message 45245.  

13.1 seems to work fine....

But tasks are running partially 100% on cpu then low% on gpu.
And then they are running 60% on gpu and very low % on cpu.
All the time on all 3 cores.

Is that supposed to be so?


Our inner tests showed that Cat 13.1 hardly can be preferred vs Cat 12.8 or Cat 12.1 for example.

I use to prefere 12.3 as I found it stable.
Tried 12.4 when upgraded box.
Went back to 12.3.
Upgraded again, now to 12.6 which had another feature fixed for switching sound source that I've been waiting for.

Now tried 13.1 and it works, but slow. No driver restarts and blue screens.


Also, as I understand, you trying to run "stock" app with non-stock params.
App tweaking comes in "anonymous platform" usage and not the primary topic in this thread. Here we find conditions where stock config works and where it doesn't work (to exclude such hosts).


This files comes with the package: mb_cmdline-7.00-opencl_ati_sah.txt
Is it not to be used?

I will now downgrade to 12.8 since you say that is what this app preferes.
I have no clue how to have my team members use only 12.8 since they most likely will use what AMD recomends is "the latest driver". But when it is realeased I will inform them.

ID: 45247 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45250 - Posted: 16 Mar 2013, 13:58:58 UTC

Now using 12.8

When using Firefox and scroll a page with the mouse's scrollwheel it decreases the 5970 GPU usage alot and comp almost freezes.

The 5850 still has the same gpu usage when I scroll and it's this card that shows the picture.

I will ask the moderator to remove alot of my posts in this thread.
ID: 45250 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 29 May 06
Posts: 1037
Credit: 8,440,339
RAC: 0
United Kingdom
Message 45251 - Posted: 16 Mar 2013, 14:02:09 UTC - in response to Message 45250.  

Now using 12.8

It only really matters what driver/APP runtime you initially do the compilations on, 938.2 works, 1016.4 doesn't.

Claggy
ID: 45251 · Report as offensive
1 · 2 · 3 · 4 · Next

Message boards : SETI@home Enhanced : Problems running ATI/OpenCL 7.00 apps - report here


 
©2021 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.