Running OpenCl on Llano can it be done?

Author	Message
ONH Send message Joined: 11 Dec 11 Posts: 4 Credit: 0 RAC: 0	Message 111543 - Posted: 11 Dec 2011 \| 19:28:21 UTC
Hello I have an AMD Llano system, and it don't want to start the WU. I have Bionc Client 7.0.2, Windows 7 x64, 11.11 Driver. I have 8GB RAM. 11.12.2011 19:59:39 \| \| ATI GPU 0: AMD SUPERSUMO (CAL version 1.4.1607, 512MB, 479MB available, 512 GFLOPS peak) 11.12.2011 19:59:39 \| \| OpenCL: ATI GPU 0: BeaverCreek (driver version CAL 1.4.1607 (VM), device version OpenCL 1.1 AMD-APP-SDK-v2.5 (793.1), 1024MB) 11.12.2011 19:59:39 \| \| ATI GPU is OpenCL-capable The Opencl divice lists that the GPU has 1GB RAM. But the WU won't start.
ID: 111543 \| Reply Quote

Gaurav Khanna Send message Joined: 8 Nov 04 Posts: 12 Credit: 2,281,165 RAC: 469	Message 111556 - Posted: 13 Dec 2011 \| 18:07:59 UTC - in response to Message 111543.
I've been trying on an AMD Fusion E350 box too. I have the latest drivers; updated to the AMD SDK 2.5 and have 1GB of VRAM. BOINC is working fine for me and sending the GPU work. However the OpenCL ATI BRP4 application crashes for me each time with this output: http://albert.phys.uwm.edu/result.php?resultid=61097 [15:42:24][2052][ERROR] Application caught signal 11. ------> Obtained 3 stack frames for this thread. ------> Backtrace: Frame 3: Binary file: ../../projects/albert.phys.uwm.edu/einsteinbinary_BRP4_1.19_i686-pc-linux-gnu__atiOpenCL (0x80b6ae2) Offset info: pthread_mutex_lock+0x64a Source file: erp_boinc_wrapper.cpp (Function: sighandler / Line: 166) Frame 2: Binary file: /opt/AMDAPP/lib/x86/libamdocl32.so (0xf6054199) Offset info: +0xae199 Frame 1: Binary file: /opt/AMDAPP/lib/x86/libamdocl32.so (0xf6054199) Offset info: +0xae199 ------> End of backtrace 15:42:24 (2052): called boinc_finish ____________
ID: 111556 \| Reply Quote

Bikeman (Heinz-Bernd Eggenstein) Volunteer moderator Project administrator Project developer Send message Joined: 28 Aug 06 Posts: 1450 Credit: 1,761,643 RAC: 2,177	Message 111558 - Posted: 13 Dec 2011 \| 19:35:17 UTC - in response to Message 111556. Last modified: 13 Dec 2011 \| 19:38:05 UTC
Hi Gaurav! I've been trying on an AMD Fusion E350 box too. I have the latest drivers; updated to the AMD SDK 2.5 and have 1GB of VRAM. BOINC is working fine for me and sending the GPU work. However the OpenCL ATI BRP4 application crashes for me each time with this output: [...] Binary file: /opt/AMDAPP/lib/x86/libamdocl32.so (0xf6054199) Offset info: +0xae199 Frame 1: [...] I suspect your environment setup is not complete, that lib should have been removed from the resp. directory. See this document: http://developer.amd.com/sdks/amdappsdk/assets/AMD_APP_SDK_Installation_Notes.pdf Section 3: 3 Important Note • For a successful build and correct operation of individually downloaded s when using the Linux Catalyst 11.11 drivers or later and the Linux AMD APP SDK 2.5, delete the old 2.5 runtime files from /opt/AMDAPP/ . To do this for 32-bit Linux OS: a. Go to /opt/AMDAPP/lib/x86. b. Delete the libamdocl32.so and the libOpenCL.so.1 files. c. Create a symlink under /usr/lib using the command # ln –s libOpenCL.so.1 libOpenCL.so To do this for 64-bit Linux OS: a. Go to /opt/AMDAPP/lib/x86. b. Delete the libamdocl32.so and the libOpenCL.so.1 files. c. Go to /opt/AMDAPP/lib/x86_64. d. Delete the libamdocl64.so and the libOpenCL.so.1 files. e. Create a symlink under /usr/lib using the command # ln –s libOpenCL.so.1 libOpenCL.so f. Create a symlink under /usr/lib64 using the command # ln –s libOpenCL.so.1 libOpenCL.so CU HBE P.S.: It escapes me why this cannot be done by the driver installer (at least it could issue a warning.... ____________
ID: 111558 \| Reply Quote

Gaurav Khanna Send message Joined: 8 Nov 04 Posts: 12 Credit: 2,281,165 RAC: 469	Message 111564 - Posted: 14 Dec 2011 \| 12:34:24 UTC - in response to Message 111558.
Thanks, HBE. Unfortunately, this does not appear to be the problem, it seems. I am planning to experiment with slightly older drivers (11.5 -- 11.10) to see if anything by chance just works out .. Will report back soon. ____________
ID: 111564 \| Reply Quote

Gaurav Khanna Send message Joined: 8 Nov 04 Posts: 12 Credit: 2,281,165 RAC: 469	Message 111611 - Posted: 21 Dec 2011 \| 12:38:15 UTC - in response to Message 111564.
An update: With the the just-released 11.12 catalyst, I actually have things working on the APU! I seem to have a minor problem left to address. After the first data-set processing in a workunit is complete, it is unable to start the next one. For example see below: http://albert.phys.uwm.edu/result.php?resultid=64181 http://albert.phys.uwm.edu/result.php?resultid=64159 http://albert.phys.uwm.edu/result.php?resultid=64137 They all crash with: Error during OpenCL kernel setup: PS (error: -5) during the processing of the second data-file AFTER completing the processing of the first data file successfully. Comments? ____________
ID: 111611 \| Reply Quote

Bernd Machenschalk Volunteer moderator Project administrator Project developer Send message Joined: 15 Oct 04 Posts: 1926 Credit: 1,921,900 RAC: 763	Message 111616 - Posted: 21 Dec 2011 \| 18:45:27 UTC - in response to Message 111611.
This is a problem Oliver needs to look into, which means it won't be fixed this year. BM
ID: 111616 \| Reply Quote

Author

Message

ONH
Send message
Joined: 11 Dec 11
Posts: 4
Credit: 0
RAC: 0

Hello

I have an AMD Llano system, and it don't want to start the WU.
I have Bionc Client 7.0.2, Windows 7 x64, 11.11 Driver. I have 8GB RAM.

11.12.2011 19:59:39 | | ATI GPU 0: AMD SUPERSUMO (CAL version 1.4.1607, 512MB, 479MB available, 512 GFLOPS peak)
11.12.2011 19:59:39 | | OpenCL: ATI GPU 0: BeaverCreek (driver version CAL 1.4.1607 (VM), device version OpenCL 1.1 AMD-APP-SDK-v2.5 (793.1), 1024MB)
11.12.2011 19:59:39 | | ATI GPU is OpenCL-capable

The Opencl divice lists that the GPU has 1GB RAM. But the WU won't start.

ID: 111543 | Reply Quote

Gaurav Khanna
Send message
Joined: 8 Nov 04
Posts: 12
Credit: 2,281,165
RAC: 469

I've been trying on an AMD Fusion E350 box too. I have the latest drivers; updated to the AMD SDK 2.5 and have 1GB of VRAM. BOINC is working fine for me and sending the GPU work.

However the OpenCL ATI BRP4 application crashes for me each time with this output:

http://albert.phys.uwm.edu/result.php?resultid=61097

[15:42:24][2052][ERROR] Application caught signal 11.

------> Obtained 3 stack frames for this thread.
------> Backtrace:
Frame 3:
Binary file: ../../projects/albert.phys.uwm.edu/einsteinbinary_BRP4_1.19_i686-pc-linux-gnu__atiOpenCL (0x80b6ae2)
Offset info: pthread_mutex_lock+0x64a
Source file: erp_boinc_wrapper.cpp (Function: sighandler / Line: 166)
Frame 2:
Binary file: /opt/AMDAPP/lib/x86/libamdocl32.so (0xf6054199)
Offset info: +0xae199
Frame 1:
Binary file: /opt/AMDAPP/lib/x86/libamdocl32.so (0xf6054199)
Offset info: +0xae199
------> End of backtrace

15:42:24 (2052): called boinc_finish
____________

ID: 111556 | Reply Quote

Bikeman (Heinz-Bernd Eggenstein)
Volunteer moderator
Project administrator
Project developer
Avatar

Send message
Joined: 28 Aug 06
Posts: 1450
Credit: 1,761,643
RAC: 2,177

Hi Gaurav!

I've been trying on an AMD Fusion E350 box too. I have the latest drivers; updated to the AMD SDK 2.5 and have 1GB of VRAM. BOINC is working fine for me and sending the GPU work.

However the OpenCL ATI BRP4 application crashes for me each time with this output:

[...]
Binary file: /opt/AMDAPP/lib/x86/libamdocl32.so (0xf6054199)
Offset info: +0xae199
Frame 1:
[...]

I suspect your environment setup is not complete, that lib should have been removed from the resp. directory. See this document:

http://developer.amd.com/sdks/amdappsdk/assets/AMD_APP_SDK_Installation_Notes.pdf

Section 3:

3 Important Note • For a successful build and correct operation of individually downloaded s when using the Linux Catalyst 11.11 drivers or later and the Linux AMD APP SDK 2.5, delete the old 2.5 runtime files from /opt/AMDAPP/ . To do this for 32-bit Linux OS: a. Go to /opt/AMDAPP/lib/x86. b. Delete the libamdocl32.so and the libOpenCL.so.1 files. c. Create a symlink under /usr/lib using the command # ln –s libOpenCL.so.1 libOpenCL.so To do this for 64-bit Linux OS: a. Go to /opt/AMDAPP/lib/x86. b. Delete the libamdocl32.so and the libOpenCL.so.1 files. c. Go to /opt/AMDAPP/lib/x86_64. d. Delete the libamdocl64.so and the libOpenCL.so.1 files. e. Create a symlink under /usr/lib using the command # ln –s libOpenCL.so.1 libOpenCL.so f. Create a symlink under /usr/lib64 using the command # ln –s libOpenCL.so.1 libOpenCL.so

CU
HBE

P.S.: It escapes me why this cannot be done by the driver installer (at least it could issue a warning....
____________

ID: 111558 | Reply Quote

Gaurav Khanna
Send message
Joined: 8 Nov 04
Posts: 12
Credit: 2,281,165
RAC: 469

Thanks, HBE. Unfortunately, this does not appear to be the problem, it seems. I am planning to experiment with slightly older drivers (11.5 -- 11.10) to see if anything by chance just works out ..

Will report back soon.
____________

ID: 111564 | Reply Quote

Gaurav Khanna
Send message
Joined: 8 Nov 04
Posts: 12
Credit: 2,281,165
RAC: 469

An update: With the the just-released 11.12 catalyst, I actually have things working on the APU! I seem to have a minor problem left to address. After the first data-set processing in a workunit is complete, it is unable to start the next one. For example see below:

http://albert.phys.uwm.edu/result.php?resultid=64181
http://albert.phys.uwm.edu/result.php?resultid=64159
http://albert.phys.uwm.edu/result.php?resultid=64137

They all crash with: Error during OpenCL kernel setup: PS (error: -5) during the processing of the second data-file AFTER completing the processing of the first data file successfully.

Comments?
____________

ID: 111611 | Reply Quote

Bernd Machenschalk
Volunteer moderator
Project administrator
Project developer
Avatar

Send message
Joined: 15 Oct 04
Posts: 1926
Credit: 1,921,900
RAC: 763

This is a problem Oliver needs to look into, which means it won't be fixed this year.

BM

ID: 111616 | Reply Quote