S@H version 7.02 released for ATI/AMD GPUs under windows.

Message boards : News : S@H version 7.02 released for ATI/AMD GPUs under windows.
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 · Next

AuthorMessage
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1546
Credit: 26,819,612
RAC: 27
United States
Message 45474 - Posted: 8 Apr 2013, 16:37:26 UTC

Hopefully this will fix the problems most people are having. I'll be looking for min and max driver revisions under which this version functions.

ID: 45474 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1626
Credit: 11,797,955
RAC: 391
Sweden
Message 45475 - Posted: 8 Apr 2013, 17:03:20 UTC - in response to Message 45474.  
Last modified: 8 Apr 2013, 17:07:57 UTC

Hopefully this will fix the problems most people are having. I'll be looking for min and max driver revisions under which this version functions.


Well, I reset the project, but it still sends out 7.01, which as we know fails, and also sends out 7.02 which seems to work. Please stop sending out r1799, we can have a fresh start with 7.02.

Now after a new reset, I get 7.02 but they run as r1799 and fails because no compilation is made of course.

Edit: I give up for now. Still sending out 7.01 fails and fails, not sending out any 7.02 app now. I will do another reset with a removed GPU from the preferences, and only run CPU until this is fixed.
ID: 45475 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1546
Credit: 26,819,612
RAC: 27
United States
Message 45476 - Posted: 8 Apr 2013, 17:09:46 UTC - in response to Message 45475.  

I think 7.01 should stop being sent now.


ID: 45476 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1626
Credit: 11,797,955
RAC: 391
Sweden
Message 45477 - Posted: 8 Apr 2013, 17:19:10 UTC - in response to Message 45476.  
Last modified: 8 Apr 2013, 17:20:11 UTC

I think 7.01 should stop being sent now.



Yup, works better now after last reset I only got 7.02 (ati_opencl_sah) , but now with all my failures, my daily quota for oddly enough the CPU stops me from getting any CPU tasks. I have 10, 7.02 (ati_opencl_sah), and the first one runs just fine now.
ID: 45477 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1626
Credit: 11,797,955
RAC: 391
Sweden
Message 45478 - Posted: 8 Apr 2013, 17:31:51 UTC
Last modified: 8 Apr 2013, 17:41:22 UTC

First 7.02 finished and validated. The stderr is a bit long and messy because I restarted the WU a couple of times to adjust some settings.

http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=13371860

Edit: Also lots of "WARNING: suboptimal workgroup size for PC_find_pulse_kernel" but that is expected with VHAR's, and only VHAR's. If you have set -sbs 256. If setting sbs to 128 they disappear but the app becomes slower.

Edit 2: Just noticed that SETI@home v7 7.01 windows_intelx86 (ati_opencl_sah), was simply replaced by SETI@home v7 7.02 windows_intelx86 (ati_opencl_sah), on the Application details for host 57178 page http://setiweb.ssl.berkeley.edu/beta/host_app_versions.php?hostid=57178, instead of 7.02 being added to the page. I haven't done 32 7.02 WU's this fast :-).
ID: 45478 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1546
Credit: 26,819,612
RAC: 27
United States
Message 45480 - Posted: 8 Apr 2013, 18:53:05 UTC - in response to Message 45479.  

I'm not sure how that would happen. It's possible that's no new server code David did to make new app versions inherit from older ones.
ID: 45480 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1626
Credit: 11,797,955
RAC: 391
Sweden
Message 45481 - Posted: 8 Apr 2013, 19:03:59 UTC - in response to Message 45480.  

I'm not sure how that would happen. It's possible that's no new server code David did to make new app versions inherit from older ones.


Ok I see. On another note, are there only VLAR's in the pipe now? I don't get any new tasks for my GPU. (not getting any for my CPU, I can live with, until another day begins)

Acer-desktop

SETI@home Beta Test 2013-04-08 20:54:41 Requesting new tasks for GPU
SETI@home Beta Test 2013-04-08 20:54:45 Scheduler request completed: got 0 new tasks
ID: 45481 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1546
Credit: 26,819,612
RAC: 27
United States
Message 45482 - Posted: 8 Apr 2013, 19:13:07 UTC - in response to Message 45481.  

Yeah, it's mostly VLARs and most everyone asking for work is not asking for CPU work.
ID: 45482 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1626
Credit: 11,797,955
RAC: 391
Sweden
Message 45483 - Posted: 8 Apr 2013, 19:14:39 UTC - in response to Message 45482.  

Yeah, it's mostly VLARs and most everyone asking for work is not asking for CPU work.


I can wait. I'm pretty good at waiting games :-)
ID: 45483 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 29 May 06
Posts: 1037
Credit: 8,439,427
RAC: 0
United Kingdom
Message 45485 - Posted: 8 Apr 2013, 19:59:38 UTC - in response to Message 45474.  

I'll be looking for min and max driver revisions under which this version functions.

As i keep repeating again and a again, the Max OpenCL device version has to be 'OpenCL 1.2 AMD-APP (938.2)' which is what comes with Cat 12.8,
The OpenCL runtime that comes with Cat 12.10 produces compilations that cause the wrong number of pulses to be found, leading to the rash of AMD/ATI app inconclusives,
Drivers after that produce compilations that use less registers and cause driver restarts on some Wu's, and low GPU usage/High CPU usage on others.
The APP runtime with Cat 13.2 and Cat 13.3 Beta's fail to do the compilations at all.
It is O.K to do the compilations under Cat 12.8/APP runtime 938.2, then upgrade to a later driver, the app will work fine (the compilations are only done when first run)
What you can't do is do the compilations on a later APP runtime, then revert to an earlier driver, compilations are already done wrong, only manual deletion, or a detach/reattach will remove them.

ATI Driver Version Cheat Sheet

Claggy
ID: 45485 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1546
Credit: 26,819,612
RAC: 27
United States
Message 45486 - Posted: 8 Apr 2013, 21:51:27 UTC - in response to Message 45485.  
Last modified: 8 Apr 2013, 22:22:48 UTC

I know you keep repeating it, but that doesn't give me the minimum version required. It also doesn't give me the way to translate 938.2 into a version number of the type that BOINC reports for the driver (1.4.1720, for example).


Didn't see the link to the cheat sheet. Thanks.



Unfortunately there's no way to distinguish the 1.4.1741 that comes with 12.8 from the 1.4.1741 that comes 12.10+.

ID: 45486 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 29 May 06
Posts: 1037
Credit: 8,439,427
RAC: 0
United Kingdom
Message 45487 - Posted: 8 Apr 2013, 22:27:41 UTC - in response to Message 45486.  
Last modified: 8 Apr 2013, 22:47:11 UTC

I know you keep repeating it, but that doesn't give me the minimum version required. It also doesn't give me the way to translate 938.2 into a version number of the type that BOINC reports for the driver (1.4.1720, for example).

1.4.1720 is a CAL driver version, the app is an OpenCL app, the CAL driver versions have all remained the same recently, the APP runtime version is the important version with these apps,
looking at my client_state.xml and sched_request_setiweb.ssl.berkeley.edu_beta.xml files,
it is the <opencl_platform_version>OpenCL 1.2 AMD-APP (938.2)</opencl_platform_version> or <opencl_device_version>OpenCL 1.2 AMD-APP (938.2)</opencl_device_version> sections you should look at,
expect some devices to be only OpenCL 1.1 compliant, depending on Hardware or driver version, i don't expect OpenCL 1.0 devices to work at all, being HD4*** GPUs.

Claggy
ID: 45487 · Report as offensive
Profile Eric J Korpela
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 15 Mar 05
Posts: 1546
Credit: 26,819,612
RAC: 27
United States
Message 45488 - Posted: 8 Apr 2013, 22:37:53 UTC - in response to Message 45487.  
Last modified: 8 Apr 2013, 22:47:41 UTC

I don't think that information is currently available to the plan classes. Do you know which BOINC version it was added in?

Ugh, this is going to get ugly. OpenCL version detection is very limited and OpenCL device driver version detection is non-existent. The original code was written assuming a single version number would identify the driver and all its components.
ID: 45488 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1626
Credit: 11,797,955
RAC: 391
Sweden
Message 45489 - Posted: 9 Apr 2013, 8:07:41 UTC

We need some non VLAR's from the splitters, or we will never be able to test 7.02. I haven't been able to get anything other than CPU tasks since the initial 9 or so 7.02s I crunched just after the release of 7.02.
ID: 45489 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 45504 - Posted: 10 Apr 2013, 11:56:51 UTC - in response to Message 45482.  

Yeah, it's mostly VLARs and most everyone asking for work is not asking for CPU work.

Eric, ATi app should be capable to do VLARs too. LEts' test this at least at beta!
ID: 45504 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 45505 - Posted: 10 Apr 2013, 12:00:35 UTC - in response to Message 45486.  

I know you keep repeating it, but that doesn't give me the minimum version required. It also doesn't give me the way to translate 938.2 into a version number of the type that BOINC reports for the driver (1.4.1720, for example).

<edit>
Didn't see the link to the cheat sheet. Thanks.
</edit>

<edit2>
Unfortunately there's no way to distinguish the 1.4.1741 that comes with 12.8 from the 1.4.1741 that comes 12.10+.
</edit2>


Eric, we need to convince David to report OpenCL runtime version for OpenCL devices. "Driver" versions have nothing to do with computations BOINC performs.
ID: 45505 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45507 - Posted: 10 Apr 2013, 14:53:29 UTC

Did some tasks during this night and the day on 5850 and 5970.
I ran 2 tasks, 1 on each card. Worked like charm :)

I am now running 1 task on 5850 and 2 tasks on 5970.
It looks like it is working.

Link to computor:

http://setiweb.ssl.berkeley.edu/beta/hosts_user.php?userid=14842
ID: 45507 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45508 - Posted: 10 Apr 2013, 15:00:29 UTC
Last modified: 10 Apr 2013, 15:21:38 UTC

Something strange happend now.

All tasks in BM stoped counting in progres meter including 1 docking task and 1 wu-prop task.
1 SETI beta was at 92% when the "clock" stoped.
And the task was completed when the "clock" started to count again.
And a new task was at 9%.

Windows Gui worked and SIV showed that all 3 GPU's was running.
A bit strange.....Might be BM 7.0.60, I dunno.

ideas anyone??


Hapend again after 20mins. All BM counters stoped for about 1min and then resumed. Win GUI works and SIV reports activity on all cpu and gpu's. Must be BM 7.0.60 I'll look at BM project page for clues.
ID: 45508 · Report as offensive
Richard Haselgrove
Volunteer tester

Send message
Joined: 3 Jan 07
Posts: 1444
Credit: 3,265,669
RAC: 0
United Kingdom
Message 45509 - Posted: 10 Apr 2013, 15:30:53 UTC - in response to Message 45508.  

Something strange happend now.

All tasks in BM stoped counting in progres meter including 1 docking task and 1 wu-prop task.
1 SETI beta was at 92% when the "clock" stoped.
And the task was completed when the "clock" started to count again.
And a new task was at 9%.

Windows Gui worked and SIV showed that all 3 GPU's was running.
A bit strange.....Might be BM 7.0.60, I dunno.

ideas anyone??


Hapend again after 20mins. All BM counters stoped for about 1min and then resumed. Win GUI works and SIV reports activity on all cpu and gpu's. Must be BM 7.0.60 I'll look at BM project page for clues.

It's been happening to me ever since I got a Windows 7 laptop, over two years ago - nothing new there.

But there is a v7.0.62 to test, with "major overhaul of work-fetch logic" (just what we need, in the last few days leading up to a major release). Best concentrate on testing that for now.
ID: 45509 · Report as offensive
TRuEQ & TuVaLu
Volunteer tester
Avatar

Send message
Joined: 28 Jan 11
Posts: 619
Credit: 2,580,051
RAC: 0
Sweden
Message 45511 - Posted: 10 Apr 2013, 16:18:10 UTC

running 7.0.62 now
ID: 45511 · Report as offensive
1 · 2 · 3 · 4 · Next

Message boards : News : S@H version 7.02 released for ATI/AMD GPUs under windows.


 
©2019 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.