SETI@home v8 beta to begin on Tuesday

Message boards : News : SETI@home v8 beta to begin on Tuesday
Message board moderation

To post messages, you must log in.

Previous · 1 . . . 50 · 51 · 52 · 53 · 54 · 55 · 56 . . . 99 · Next

AuthorMessage
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 58615 - Posted: 15 Jun 2016, 10:41:31 UTC - in response to Message 58613.  
Last modified: 15 Jun 2016, 10:41:46 UTC

On my own iGPU I usually got AP's invalids/inconclusives but OK for MultiBeam.
Seems Intel too frivolous with trigonometry functions precision.
Perhaps to speedup games on their GPU they sacrifice too much precision to produce correctly-enough computations.
I already forced to go separate computation path inside oclFFT to increase precision of FFT transforms (thanks to Einstein's peoples who developed that path). Just too imprecise device :(
News about SETI opt app releases: https://twitter.com/Raistmer
ID: 58615 · Report as offensive
SusieQ
Volunteer tester

Send message
Joined: 12 Nov 10
Posts: 1116
Credit: 29,753,172
RAC: 14,195
United Kingdom
Message 58772 - Posted: 27 Jun 2016, 14:56:58 UTC

Results of my test of older driver.

Still getting invalids and errors, possibly at a slower rate, but now also seeing GPU crash notifications.

Unfortunately, I'm stuck with the two versions I have tried as, when I try to install the newer version downloaded from the Intel website, I get a message to say that it has not been validated for my PC. I guess this is a Windows 10 "feature"

I've reverted back to the newer driver as the older one did seem to be affecting other things and, as my PC is for work, it really ought to be configured optimally for that :-D
ID: 58772 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1644
Credit: 12,123,977
RAC: 15,073
Sweden
Message 58775 - Posted: 27 Jun 2016, 15:38:12 UTC

On my iGPU, the HD4600 in the Intel(R) Core(TM) i7-4790K CPU, I have never had any problems, invalids or errors. Works just fine with both MB and AP. The driver version is 10.18.10.3907.

Sure, it's not the most recent driver, but why upgrade something that works?
WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.
ID: 58775 · Report as offensive
SusieQ
Volunteer tester

Send message
Joined: 12 Nov 10
Posts: 1116
Credit: 29,753,172
RAC: 14,195
United Kingdom
Message 58781 - Posted: 28 Jun 2016, 9:04:54 UTC

I never used to have problems until I moved to Windows 10 so not entirely sure whether it's due to driver alone or the mixture of driver and Windows 10.
ID: 58781 · Report as offensive
Zalster
Volunteer tester

Send message
Joined: 30 Dec 13
Posts: 258
Credit: 12,338,439
RAC: 5,408
United States
Message 58792 - Posted: 30 Jun 2016, 2:23:43 UTC

https://setiweb.ssl.berkeley.edu/beta//forum_thread.php?id=2320

From Grant's post in the enhanced sections.

Looks like v8.13 is failing due to not being able to read or find the .cl files
ID: 58792 · Report as offensive
Zalster
Volunteer tester

Send message
Joined: 30 Dec 13
Posts: 258
Credit: 12,338,439
RAC: 5,408
United States
Message 58794 - Posted: 30 Jun 2016, 2:58:26 UTC

Ok, I took a chance and downloaded Raistmer new app and move the Multibeam_Kernals_r3480.cl to the beta folder and restarted Boinc.

It looks to have worked.

Will have to wait and see if they validate

here's the link to his mail.ru to download, you will need something to expand the zip file

https://cloud.mail.ru/public/4g4V/Gt278t2ku
ID: 58794 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1644
Credit: 12,123,977
RAC: 15,073
Sweden
Message 58797 - Posted: 30 Jun 2016, 3:48:06 UTC - in response to Message 58794.  
Last modified: 30 Jun 2016, 3:48:48 UTC

Ok, I took a chance and downloaded Raistmer new app and move the Multibeam_Kernals_r3480.cl to the beta folder and restarted Boinc.

It looks to have worked.

Will have to wait and see if they validate

here's the link to his mail.ru to download, you will need something to expand the zip file

https://cloud.mail.ru/public/4g4V/Gt278t2ku

Thanks Zalster. I noticed your post just before my 8.12s were about to run out on my 980, so I could apply your fix before any errors occured, when the new 8.13s was about to be started :-)

Very well spotted Sir!!
WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.
ID: 58797 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 15 Jun 16
Posts: 45
Credit: 1,836,741
RAC: 0
Australia
Message 58798 - Posted: 30 Jun 2016, 4:05:19 UTC - in response to Message 58794.  

Ok, I took a chance and downloaded Raistmer new app and move the Multibeam_Kernals_r3480.cl to the beta folder and restarted Boinc.

It looks to have worked.

Will have to wait and see if they validate

Thanks.
I've done likewise.
Will have to keep an eye on things as 2 WUs finished after less than 2 min crunching.
Grant
Darwin NT.
ID: 58798 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1644
Credit: 12,123,977
RAC: 15,073
Sweden
Message 58800 - Posted: 30 Jun 2016, 5:25:18 UTC
Last modified: 30 Jun 2016, 5:27:56 UTC

Well, they do validate of course with this "manual" adding of the right .cl file. (I've had a couple of them validated) However, most will go to "Validation pending", unless the wingman is a CPU app, or CUDA apps. I doubt that there are many of us BETA testers, who have realized that the .cl file is missing, and gone down Zalster's road to a fix.
WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.
ID: 58800 · Report as offensive
Winterknight
Volunteer tester

Send message
Joined: 15 Jun 05
Posts: 707
Credit: 5,439,085
RAC: 2,729
United Kingdom
Message 58803 - Posted: 30 Jun 2016, 7:58:51 UTC - in response to Message 58798.  

Ok, I took a chance and downloaded Raistmer new app and move the Multibeam_Kernals_r3480.cl to the beta folder and restarted Boinc.

It looks to have worked.

Will have to wait and see if they validate

Thanks.
I've done likewise.
Will have to keep an eye on things as 2 WUs finished after less than 2 min crunching.

Just woken up to this "cannot open..." but now done the same.
Can I request Eric is relocated to Sydney, so all his changes are done early in the earths day, not midnight UK time.
ID: 58803 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 58804 - Posted: 30 Jun 2016, 8:14:27 UTC - in response to Message 58800.  
Last modified: 30 Jun 2016, 8:16:40 UTC

Well, they do validate of course with this "manual" adding of the right .cl file. (I've had a couple of them validated) However, most will go to "Validation pending", unless the wingman is a CPU app, or CUDA apps. I doubt that there are many of us BETA testers, who have realized that the .cl file is missing, and gone down Zalster's road to a fix.

Of course new deployment will be required as soon as Eric will read mail/forums....

Until then would be good to establish what plan classes affected by CL file omission.
News about SETI opt app releases: https://twitter.com/Raistmer
ID: 58804 · Report as offensive
MarkJ
Volunteer tester

Send message
Joined: 18 Oct 09
Posts: 48
Credit: 73,283
RAC: 0
Australia
Message 58807 - Posted: 30 Jun 2016, 12:07:29 UTC - in response to Message 58772.  

I've reverted back to the newer driver as the older one did seem to be affecting other things and, as my PC is for work, it really ought to be configured optimally for that :-D

I've had a bug raised with Intel since Feb 2016 regarding their OpenCL not producing valid results for the HD Graphics 530. I've tried quite a few of their drivers (under Win7).
ID: 58807 · Report as offensive
SusieQ
Volunteer tester

Send message
Joined: 12 Nov 10
Posts: 1116
Credit: 29,753,172
RAC: 14,195
United Kingdom
Message 58808 - Posted: 30 Jun 2016, 13:09:58 UTC - in response to Message 58807.  

Hi Mark

Good to know.

Not sure if you'd checked the details of the client in question but it has an Intel HD Graphics 530 too.

At least this shows it's not just a Windows 10 issue and also looks as though it wouldn't have made any difference even if I had been able to install their most up to date driver :-)
ID: 58808 · Report as offensive
Zalster
Volunteer tester

Send message
Joined: 30 Dec 13
Posts: 258
Credit: 12,338,439
RAC: 5,408
United States
Message 58809 - Posted: 30 Jun 2016, 14:22:03 UTC - in response to Message 58804.  

Well, they do validate of course with this "manual" adding of the right .cl file. (I've had a couple of them validated) However, most will go to "Validation pending", unless the wingman is a CPU app, or CUDA apps. I doubt that there are many of us BETA testers, who have realized that the .cl file is missing, and gone down Zalster's road to a fix.

Of course new deployment will be required as soon as Eric will read mail/forums....

Until then would be good to establish what plan classes affected by CL file omission.



Well we know it was the 8.13 OpenCl_sah and OpenCL_SoG as they both immediately went to "can't read cl" then postponed. Don't have the event log anymore as it cleared upon restarting BOINC.
ID: 58809 · Report as offensive
SusieQ
Volunteer tester

Send message
Joined: 12 Nov 10
Posts: 1116
Credit: 29,753,172
RAC: 14,195
United Kingdom
Message 58810 - Posted: 30 Jun 2016, 14:44:57 UTC
Last modified: 30 Jun 2016, 15:01:34 UTC

It's missing from the 8.13 opencl_intel_gpu_sah

OK with the cl file copied over as per earlier posts
ID: 58810 · Report as offensive
Richard Haselgrove
Volunteer tester

Send message
Joined: 3 Jan 07
Posts: 1451
Credit: 3,266,428
RAC: 27
United Kingdom
Message 58811 - Posted: 30 Jun 2016, 14:51:51 UTC - in response to Message 58809.  

Don't have the event log anymore as it cleared upon restarting BOINC.

Yes, you do. It's in stdoutdae.txt
ID: 58811 · Report as offensive
Profile Raistmer
Volunteer tester
Avatar

Send message
Joined: 18 Aug 05
Posts: 2423
Credit: 15,878,738
RAC: 0
Russia
Message 58816 - Posted: 30 Jun 2016, 21:48:03 UTC - in response to Message 58810.  

It's missing from the 8.13 opencl_intel_gpu_sah

OK with the cl file copied over as per earlier posts

Looking to application page 8.13 seems complete failure, for AMD too :/
Hope it will be fixed before weekend...
News about SETI opt app releases: https://twitter.com/Raistmer
ID: 58816 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1644
Credit: 12,123,977
RAC: 15,073
Sweden
Message 58817 - Posted: 30 Jun 2016, 22:01:42 UTC
Last modified: 30 Jun 2016, 22:32:08 UTC

This one for the iGPU, fails, even with default settings (default start at the end) Aborted it, because at max it got to 26.54 percent. Edit: I have the right .cl file (MultiBeam_Kernels_r3480.cl)

Fails with these errors. Seems to be different Edit: Line numbers at each restart.

ERROR: OpenCL kernel/call 'Enqueueing kernel:PC_find_pulse_kernel_cl,pass 5' call failed (-6) in file ..\analyzePoT.cpp near line XXXX

http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=24202194

Arecibo tasks seems to work though. I'm now running another Guppi with default settings, let's see how that goes.

Edit: No, Guppi can't be run on my iGPU with 8.13. Worked perfectly with 8.12.

Next one aborted with the same type of failure:

http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=24203917

I'll abort all Guppi's on the iGPU, and continue to run the Arecibo's I have on it.
WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.
ID: 58817 · Report as offensive
Zalster
Volunteer tester

Send message
Joined: 30 Dec 13
Posts: 258
Credit: 12,338,439
RAC: 5,408
United States
Message 58818 - Posted: 30 Jun 2016, 22:47:11 UTC - in response to Message 58817.  

There is a new work class 8.14 that just rolled out.

Lets see how it does
ID: 58818 · Report as offensive
Linux? You're kidding me!!
Volunteer tester
Avatar

Send message
Joined: 10 Mar 12
Posts: 1644
Credit: 12,123,977
RAC: 15,073
Sweden
Message 58819 - Posted: 30 Jun 2016, 23:50:02 UTC
Last modified: 1 Jul 2016, 0:11:55 UTC

No, Guppi fails totally on my iGPU with 8.14 SETI@home v8 (opencl_intel_gpu_sah) also.

Aborted, since it will only repeat itself.

In this case:

ERROR: OpenCL kernel/call 'Enqueueing kernel:PC_find_pulse_partial_kernel_cl,pass 3' call failed (-6) in file ..\analyzePoT.cpp near line 2340.
Waiting 30 sec before restart...

WU:
http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=24211237

I'll abort all Guppi's on my iGPU here on Beta, and let it run the Arecibo tasks it got, and then put it at use on Main, with 8.12 (opencl_intel_gpu_sah) MB8_win_x86_SSSE3_OpenCL_Intel_r3430.exe, where it works flawlessly, even with Guppi tasks.

(It used to work here on Beta too, with r3430) see two examples: http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=24198175, and http://setiweb.ssl.berkeley.edu/beta/result.php?resultid=24198719

In other words: "r3480, is fail fail fail with Guppi's on my iGPU HD4600, while r3430 was win win win". :-)
WARNING!! "THIS IS A SIGNATURE", of the "IT MAY CHANGE AT ANY MOMENT" type. It may, or may not be considered insulting, all depending upon HOW SENSITIVE THE VIEWER IS, to certain inputs to/from the nervous system.
ID: 58819 · Report as offensive
Previous · 1 . . . 50 · 51 · 52 · 53 · 54 · 55 · 56 . . . 99 · Next

Message boards : News : SETI@home v8 beta to begin on Tuesday


 
©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.