APv7-related issues and errors

Message boards : Number crunching : APv7-related issues and errors
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · Next

AuthorMessage
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1585655 - Posted: 12 Oct 2014, 6:33:37 UTC - in response to Message 1585647.  

7) can't get list of devices for some reason:
[2] AMD ATI Radeon HD 5x00 series (Redwood) (1024MB) driver: 1.4.1523 OpenCL: 1.1
http://setiathome.berkeley.edu/results.php?hostid=6189952
-32 code is "Invalid platform" error.


<core_client_version>7.0.25</core_client_version>
<![CDATA[
<message>
too many boinc_temporary_exit()s
</message>
<stderr_txt>
Running on device number: 0
Priority of worker thread raised successfully
Priority of process adjusted successfully, below normal priority class used
WARNING: boinc_get_opencl_ids failed with code -1
Error: Getting Platforms. (clGetPlatformsIDs)
BOINC assigns device 0
WARNING: BOINC failed to provide OpenCL device, using own enumeration abilities
ERROR: OpenCL kernel/call 'clGetDeviceIDs (second call)' call failed (-32) in file ..\..\..\..\src\GPU_lock.cpp near line 812.
Waiting 30 sec before restart...
Running on device number: 1
Priority of worker thread raised successfully
Priority of process adjusted successfully, below normal priority class used
WARNING: boinc_get_opencl_ids failed with code -1
Error: Getting Platforms. (clGetPlatformsIDs)
BOINC assigns device 1
WARNING: BOINC failed to provide OpenCL device, using own enumeration abilities
ERROR: OpenCL kernel/call 'clGetDeviceIDs (second call)' call failed (-32) in file ..\..\..\..\src\GPU_lock.cpp near line 812.
Waiting 30 sec before restart...
...............

As well as having an 'Alpha' version of Boinc, the CAL driver version equates to Cat 11.8, which is a SDK 2.5 driver, so too old driver too.

Claggy
ID: 1585655 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1585659 - Posted: 12 Oct 2014, 6:45:27 UTC - in response to Message 1585648.  

7A) same as 7) but for iGPU: INTEL Intel(R) HD Graphics (1752MB) OpenCL: 1.2
http://setiathome.berkeley.edu/results.php?hostid=7344909

CPU is a Intel(R) Pentium(R) CPU G2020T @ 2.50GHz, anyone tested if app works on this CPU/GPU combination?

We're only seeing the device fallback mode in the stderr.txt, and not the app startup and device detection.

Claggy
ID: 1585659 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1585699 - Posted: 12 Oct 2014, 8:47:54 UTC - in response to Message 1585659.  

7A) same as 7) but for iGPU: INTEL Intel(R) HD Graphics (1752MB) OpenCL: 1.2
http://setiathome.berkeley.edu/results.php?hostid=7344909

CPU is a Intel(R) Pentium(R) CPU G2020T @ 2.50GHz, anyone tested if app works on this CPU/GPU combination?

We're only seeing the device fallback mode in the stderr.txt, and not the app startup and device detection.

Claggy

That's an interesting, and rather troubling, development. The G2020T - like many other lower-end Intel chips - is listed as having plain 'Intel® HD Graphics'. By contrast, CPUs that we know can be used for crunching on OpenCL projects, like the i7-4770K are listed as 'Intel® HD Graphics 4600' - always with a number. We know that 2000 and 3000 (Sandy Bridge) didn't work, and iGPU crunching started with 2500 and 4000, Ivy Bridge.

If Intel's drivers are now advertising OpenCL 1.2 capability on even the smallest CPUs, as they appear to be doing on the download page, how are we - and more importantly, BOINC - going to be able to distinguish which iGPUs are capable of general-purpose computing?
ID: 1585699 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1585737 - Posted: 12 Oct 2014, 11:53:31 UTC

Here's a problem reported on the Boinc dev forum:

BOINCManager / astropulse crashes system to reboot on Mac OS X

Claggy
ID: 1585737 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 1585808 - Posted: 12 Oct 2014, 15:45:17 UTC - in response to Message 1585655.  


As well as having an 'Alpha' version of Boinc, the CAL driver version equates to Cat 11.8, which is a SDK 2.5 driver, so too old driver too.

Claggy


Thanks for explanation. But I would expect another type of failure if SDK 2.5 would be properly installed and detected... Some additional factor here (maybe BOINC alpha itself that factor? )
ID: 1585808 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 1585811 - Posted: 12 Oct 2014, 15:47:10 UTC - in response to Message 1585699.  

BOINC - going to be able to distinguish which iGPUs are capable of general-purpose computing?

If BOINC reports OpenCL 1.2 for iGPU, not for CPU itself - it should work. Can someone confirm that BOINC detected iGPU as OpenCL-capable, not CPU part?
ID: 1585811 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1585819 - Posted: 12 Oct 2014, 16:10:21 UTC - in response to Message 1585811.  

BOINC - going to be able to distinguish which iGPUs are capable of general-purpose computing?

If BOINC reports OpenCL 1.2 for iGPU, not for CPU itself - it should work. Can someone confirm that BOINC detected iGPU as OpenCL-capable, not CPU part?

We only get that level of detail from the Event Log, not from the website summary. So we need someone here on these boards to experience the problem and post their startup log.

I did try to install the updated driver on my Sandy Bridge, but it's a proprietary laptop build with Optimus as well, so I didn't get very far. I'll try again after a break, now that my day's main testing is over.

And in the end, we'll probably have to consult with Charlie again, once we've gathered the evidence.
ID: 1585819 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1585828 - Posted: 12 Oct 2014, 16:25:50 UTC - in response to Message 1585699.  

7A) same as 7) but for iGPU: INTEL Intel(R) HD Graphics (1752MB) OpenCL: 1.2
http://setiathome.berkeley.edu/results.php?hostid=7344909

CPU is a Intel(R) Pentium(R) CPU G2020T @ 2.50GHz, anyone tested if app works on this CPU/GPU combination?

We're only seeing the device fallback mode in the stderr.txt, and not the app startup and device detection.

Claggy

That's an interesting, and rather troubling, development. The G2020T - like many other lower-end Intel chips - is listed as having plain 'Intel® HD Graphics'. By contrast, CPUs that we know can be used for crunching on OpenCL projects, like the i7-4770K are listed as 'Intel® HD Graphics 4600' - always with a number. We know that 2000 and 3000 (Sandy Bridge) didn't work, and iGPU crunching started with 2500 and 4000, Ivy Bridge.

If Intel's drivers are now advertising OpenCL 1.2 capability on even the smallest CPUs, as they appear to be doing on the download page, how are we - and more importantly, BOINC - going to be able to distinguish which iGPUs are capable of general-purpose computing?

I did use my Bay Trail system at Beta. Like other low end GPUs it also is listed as INTEL Intel(R) HD Graphics (1195MB) OpenCL: 1.2. The only problem I had was the estimated completion time from the server was far to low. Where the GPU takes ~20 hours to complete 1 AP the server was giving me ~1-1.5 hours. So the two tasks that it tried before I stopped ended in 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED.
I had planned to try again with a clean instance of BOINC to see if the estimate issue was just something odd in my client, but I had not gotten to that yet.

The HD Graphics found in the G2020T looks to be the same GPU as the HD Graphics 2500 found in the 3rd generation i3 CPUs, but with a lower maximum clock.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1585828 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14649
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1585852 - Posted: 12 Oct 2014, 17:35:25 UTC - in response to Message 1585828.  

The only problem I had was the estimated completion time from the server was far to low. Where the GPU takes ~20 hours to complete 1 AP the server was giving me ~1-1.5 hours. So the two tasks that it tried before I stopped ended in 197 (0xc5) EXIT_TIME_LIMIT_EXCEEDED.

I had planned to try again with a clean instance of BOINC to see if the estimate issue was just something odd in my client, but I had not gotten to that yet.

No, the estimation error is a pure server-side problem. If you want to avoid it, the only way we know is to inoculate <rsc_fpops_bound> in the individual workunit definition, to some suitably large figure. No need to waste time on the rest of BOINC.
ID: 1585852 · Report as offensive
Profile dnolan
Avatar

Send message
Joined: 30 Aug 01
Posts: 1228
Credit: 47,779,411
RAC: 32
United States
Message 1586549 - Posted: 13 Oct 2014, 23:34:11 UTC

I'm not sure if this is a one-off error or not, but thought I'd post it.
Invalid task
The machine that did it hasn't done a lot on Seti, but has been working away on Einstein for some time and I can't remember ever seeing an error on that project. So far, since the lunatics 0.43 installation, it's only gotten three tasks, one pending, this invalid one, and one that hasn't started yet. It does only GPU work, too.

-Dave
ID: 1586549 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 1586590 - Posted: 14 Oct 2014, 0:35:14 UTC - in response to Message 1586549.  

I'm not sure if this is a one-off error or not, but thought I'd post it.
Invalid task
The machine that did it hasn't done a lot on Seti, but has been working away on Einstein for some time and I can't remember ever seeing an error on that project. So far, since the lunatics 0.43 installation, it's only gotten three tasks, one pending, this invalid one, and one that hasn't started yet. It does only GPU work, too.

-Dave

Hmm, I can't guess much from that:
Info : Building Program (binary, clBuildProgram):main kernels: OK code 0
ERROR: Invalid parameter detected in function (null). File: (null) Line: 0
ERROR: Expression: (null)


Unhandled Exception Detected...

- Unhandled Exception Record -
Reason: Breakpoint Encountered (0x80000003) at address 0x75763226

The Breakpoint probably isn't meaningful, the BOINC API uses a forced break to activate the debugger dump. I hope Raistmer can get something out of it.
                                                                  Joe
ID: 1586590 · Report as offensive
Profile Mike Special Project $75 donor
Volunteer tester
Avatar

Send message
Joined: 17 Feb 01
Posts: 34253
Credit: 79,922,639
RAC: 80
Germany
Message 1586647 - Posted: 14 Oct 2014, 3:44:08 UTC

Looks like oclfft_plan or tune switch were set wrong.


With each crime and every kindness we birth our future.
ID: 1586647 · Report as offensive
Profile dnolan
Avatar

Send message
Joined: 30 Aug 01
Posts: 1228
Credit: 47,779,411
RAC: 32
United States
Message 1586673 - Posted: 14 Oct 2014, 5:50:57 UTC - in response to Message 1586647.  

Looks like oclfft_plan or tune switch were set wrong.


Just an FYI, that machine isn't using the tune switch on the command line,
here's what's in the command line file:
-unroll 12 -ffa_block 12288 -ffa_block_fetch 6144

Not sure what the other item is?

-Dave
ID: 1586673 · Report as offensive
Profile Raistmer
Volunteer developer
Volunteer tester
Avatar

Send message
Joined: 16 Jun 01
Posts: 6325
Credit: 106,370,077
RAC: 121
Russia
Message 1586703 - Posted: 14 Oct 2014, 7:13:17 UTC - in response to Message 1586673.  
Last modified: 14 Oct 2014, 7:15:11 UTC

It's the single case so far for that host: http://setiathome.berkeley.edu/results.php?hostid=7407068 so hardly config error if config was stable between runs... For now just monitor result further and try to catch what condition could be assotiated with error. Not apparent from log.
Could it be out of memory state for that host?
ID: 1586703 · Report as offensive
Profile dnolan
Avatar

Send message
Joined: 30 Aug 01
Posts: 1228
Credit: 47,779,411
RAC: 32
United States
Message 1586766 - Posted: 14 Oct 2014, 12:54:34 UTC

Ok, will watch it and see if any errors come up. The config hasn't changed since lunatics was installed. I don't think it was an out of memory condition, but since I don't actively use the host (it's used mainly as a storage system for my other machines to back up to), I can't be sure. Thanks for looking.

-Dave
ID: 1586766 · Report as offensive
Doodskop

Send message
Joined: 24 Jul 99
Posts: 1
Credit: 556,372
RAC: 1
United Kingdom
Message 1587358 - Posted: 15 Oct 2014, 20:25:43 UTC

Im getting computation errors for Astropulse on my Mac

Wed 15 Oct 21:12:08 2014 | SETI@home | Starting task ap_07se14aa_B3_P1_00147_20141014_20843.wu_2
Wed 15 Oct 21:12:09 2014 | SETI@home | Computation for task ap_07se14aa_B3_P1_00147_20141014_20843.wu_2 finished
Wed 15 Oct 21:12:09 2014 | SETI@home | Output file ap_07se14aa_B3_P1_00147_20141014_20843.wu_2_0 for task ap_07se14aa_B3_P1_00147_20141014_20843.wu_2 absent



Wed 15 Oct 18:36:08 2014 | | Starting BOINC client version 7.2.42 for x86_64-apple-darwin
Wed 15 Oct 18:36:08 2014 | | log flags: file_xfer, sched_ops, task
Wed 15 Oct 18:36:08 2014 | | Libraries: libcurl/7.26.0 OpenSSL/1.0.1e zlib/1.2.5 c-ares/1.9.1
Wed 15 Oct 18:36:08 2014 | | Data directory: /Library/Application Support/BOINC Data
Wed 15 Oct 18:36:08 2014 | | OpenCL: AMD/ATI GPU 0: Radeon HD 4850 (driver version 1.0, device version OpenCL 1.0, 512MB, 512MB available, 402 GFLOPS peak)
Wed 15 Oct 18:36:08 2014 | | OpenCL CPU: Intel(R) Core(TM) i5 CPU 750 @ 2.67GHz (OpenCL driver vendor: Apple, driver version 1.1, device version OpenCL 1.2)
Wed 15 Oct 18:36:08 2014 | | Host name: Stephens-iMac.local
Wed 15 Oct 18:36:08 2014 | | Processor: 4 GenuineIntel Intel(R) Core(TM) i5 CPU 750 @ 2.67GHz [x86 Family 6 Model 30 Stepping 5]
Wed 15 Oct 18:36:08 2014 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clfsh ds acpi mmx fxsr sse sse2 ss htt tm pbe pni dtes64 mon dscpl vmx smx est tm2 ssse3 cx16 tpr pdcm sse4_1 sse4_2 popcnt
Wed 15 Oct 18:36:08 2014 | | OS: Mac OS X 10.9.5 (Darwin 13.4.0)
Wed 15 Oct 18:36:08 2014 | | Memory: 16.00 GB physical, 733.01 GB virtual
Wed 15 Oct 18:36:08 2014 | | Disk: 930.71 GB total, 732.77 GB free
Wed 15 Oct 18:36:08 2014 | | Local time is UTC +1 hours
ID: 1587358 · Report as offensive
Profile HAL9000
Volunteer tester
Avatar

Send message
Joined: 11 Sep 99
Posts: 6534
Credit: 196,805,888
RAC: 57
United States
Message 1587368 - Posted: 15 Oct 2014, 20:33:36 UTC - in response to Message 1587358.  

Im getting computation errors for Astropulse on my Mac

Wed 15 Oct 21:12:08 2014 | SETI@home | Starting task ap_07se14aa_B3_P1_00147_20141014_20843.wu_2
Wed 15 Oct 21:12:09 2014 | SETI@home | Computation for task ap_07se14aa_B3_P1_00147_20141014_20843.wu_2 finished
Wed 15 Oct 21:12:09 2014 | SETI@home | Output file ap_07se14aa_B3_P1_00147_20141014_20843.wu_2_0 for task ap_07se14aa_B3_P1_00147_20141014_20843.wu_2 absent



Wed 15 Oct 18:36:08 2014 | | Starting BOINC client version 7.2.42 for x86_64-apple-darwin
Wed 15 Oct 18:36:08 2014 | | log flags: file_xfer, sched_ops, task
Wed 15 Oct 18:36:08 2014 | | Libraries: libcurl/7.26.0 OpenSSL/1.0.1e zlib/1.2.5 c-ares/1.9.1
Wed 15 Oct 18:36:08 2014 | | Data directory: /Library/Application Support/BOINC Data
Wed 15 Oct 18:36:08 2014 | | OpenCL: AMD/ATI GPU 0: Radeon HD 4850 (driver version 1.0, device version OpenCL 1.0, 512MB, 512MB available, 402 GFLOPS peak)
Wed 15 Oct 18:36:08 2014 | | OpenCL CPU: Intel(R) Core(TM) i5 CPU 750 @ 2.67GHz (OpenCL driver vendor: Apple, driver version 1.1, device version OpenCL 1.2)
Wed 15 Oct 18:36:08 2014 | | Host name: Stephens-iMac.local
Wed 15 Oct 18:36:08 2014 | | Processor: 4 GenuineIntel Intel(R) Core(TM) i5 CPU 750 @ 2.67GHz [x86 Family 6 Model 30 Stepping 5]
Wed 15 Oct 18:36:08 2014 | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clfsh ds acpi mmx fxsr sse sse2 ss htt tm pbe pni dtes64 mon dscpl vmx smx est tm2 ssse3 cx16 tpr pdcm sse4_1 sse4_2 popcnt
Wed 15 Oct 18:36:08 2014 | | OS: Mac OS X 10.9.5 (Darwin 13.4.0)
Wed 15 Oct 18:36:08 2014 | | Memory: 16.00 GB physical, 733.01 GB virtual
Wed 15 Oct 18:36:08 2014 | | Disk: 930.71 GB total, 732.77 GB free
Wed 15 Oct 18:36:08 2014 | | Local time is UTC +1 hours

That is the 4th error documented above. ATI HD 4xxx cards do not seem to be very happy on MAC.
SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[
ID: 1587368 · Report as offensive
Profile Zalster Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 27 May 99
Posts: 5517
Credit: 528,817,460
RAC: 242
United States
Message 1588487 - Posted: 17 Oct 2014, 21:09:24 UTC - in response to Message 1587358.  
Last modified: 17 Oct 2014, 21:09:47 UTC

I'm linking this guys thread from over the question and answer for Macs here for someone to try and help him. I don't know anything about ATI Graphic cards.


AstroPulse v7 v7.04 (opencl_ati_mac) - KABOOM
http://setiathome.berkeley.edu/forum_thread.php?id=75912&postid=1588481
ID: 1588487 · Report as offensive
Urs Echternacht
Volunteer tester
Avatar

Send message
Joined: 15 May 99
Posts: 692
Credit: 135,197,781
RAC: 211
Germany
Message 1588500 - Posted: 17 Oct 2014, 21:42:06 UTC - in response to Message 1588487.  

I'm linking this guys thread from over the question and answer for Macs here for someone to try and help him. I don't know anything about ATI Graphic cards.


AstroPulse v7 v7.04 (opencl_ati_mac) - KABOOM
http://setiathome.berkeley.edu/forum_thread.php?id=75912&postid=1588481

The fix (new app version) for this issue is on the way. No timeframe when it will appear here has been set for now. Sorry, i don't have more info yet.
_\|/_
U r s
ID: 1588500 · Report as offensive
Profile [AF>Le_Pommier] Jerome_C2005
Volunteer tester

Send message
Joined: 1 Oct 03
Posts: 4
Credit: 428,809
RAC: 0
Mexico
Message 1588504 - Posted: 17 Oct 2014, 21:50:50 UTC

Thanks Zalster for reporting my post.
ID: 1588504 · Report as offensive
Previous · 1 · 2 · 3 · 4 · Next

Message boards : Number crunching : APv7-related issues and errors


 
©2024 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.