I am getting computation error

tullio
tullio
Joined: 22 Jan 05
Posts: 53
Credit: 137342
RAC: 0
Topic 90865

I am getting computation error both on a Linux host with no graphic board and on this Windows 8.1 PC with an ATI/AMD graphic booard which is recognized by Einstein@home. Sorry if I chose this thread but I am new to Drupal.

Tullio

Oliver Behnke
Oliver Behnke
Moderator
Administrator
Joined: 4 Sep 07
Posts: 320
Credit: 8545955
RAC: 0

No problem, I moved your post

No problem, I moved your post :-)

Oliver Behnke
Oliver Behnke
Moderator
Administrator
Joined: 4 Sep 07
Posts: 320
Credit: 8545955
RAC: 0

Which of your computers are

Which of your computers are we talking about? At least one (10258) lacks the necessary OpenCL installation and the other one (12544has an issue with the BOINC client not correctly identifying/propagating the GPU type. Just have a look at any failing task's error output, it's pretty informative.

Oliver

tullio
tullio
Joined: 22 Jan 05
Posts: 53
Credit: 137342
RAC: 0

10258 is the Linux host with

10258 is the Linux host with SuSE 13.1 and no graphic board driver installed although it has a Geforce GTX 750 in the PCIe slot. 12544 is a Windows 8.1 PC with BOINC 7.4.27 which runs Einstein@home Binary Radio Pulsar Search 1.39 (BRP5-opencl-ati). I used to run Albert@home on the Linux host but I see that the new tasks use a GPU so I put NNT on that host.

SETI@home also  sees the GPU.

tullio
tullio
Joined: 22 Jan 05
Posts: 53
Credit: 137342
RAC: 0

It seems to fail on 1.40

It seems to fail on 1.40 Binary Radio Pulsar Search while doing fine on 1.39 in Einstein@home.

tullio
tullio
Joined: 22 Jan 05
Posts: 53
Credit: 137342
RAC: 0

Albert@home thinks I have an

Albert@home thinks I have an Intel GPU but I have an ATI/AMD GPU. Einstein@home and SETI@home identify it correctly.

Richie
Richie
Joined: 26 Jul 14
Posts: 41
Credit: 154864070
RAC: 0

I had that same problem a

I had that same problem a short moment (my computer has only Nvidia card but received some Intel / ATI work... and those two options were unchecked all the time in the Project Preferences). I did some pseudo change (changed something back and forth) into the preferences and saved the changes. Then I forced update command from the Boinc manager for Albert a few times. I don't know which had an effect on what, but wrong sort of work stopped coming anyway.

tullio
tullio
Joined: 22 Jan 05
Posts: 53
Credit: 137342
RAC: 0

SETI@home uses the following

SETI@home uses the following definition: opencl_ati_cat132. Einstein@home uses opencl-ati. Since I am not an expert on graphic boards, I don't know the difference.

Claggy
Claggy
Joined: 29 Dec 06
Posts: 122
Credit: 4040969
RAC: 0

According to a stderr at Seti

According to a stderr at Seti your host has two GPUs, through Boinc is reporting the most capable to the server:

Quote:
SETI7 update by Raistmer

OpenCL version by Raistmer, r2489

Number of OpenCL platforms: 1

OpenCL Platform Name: AMD Accelerated Parallel Processing
Number of devices: 2
  Max compute units: 2
  Max work group size: 256
  Max clock frequency: 775Mhz
  Max memory allocation: 536870912
  Cache type: None
  Cache line size: 0
  Cache size: 0
  Global memory size: 1073741824
  Constant buffer size: 65536
  Max number of constant args: 8
  Local memory type: Scratchpad
  Local memory size: 32768
  Queue properties:
    Out-of-Order: No
  Name: Caicos
  Vendor: Advanced Micro Devices, Inc.
  Driver version: 1268.1 (VM)
  Version: OpenCL 1.2 AMD-APP (1268.1)
  Extensions: cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_atomic_counters_32 cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_d3d10_sharing cl_khr_dx9_media_sharing cl_amd_image2d_from_buffer_read_only
  Max compute units: 6
  Max work group size: 256
  Max clock frequency: 844Mhz
  Max memory allocation: 401866752
  Cache type: None
  Cache line size: 0
  Cache size: 0
  Global memory size: 1607467008
  Constant buffer size: 65536
  Max number of constant args: 8
  Local memory type: Scratchpad
  Local memory size: 32768
  Queue properties:
    Out-of-Order: No
  Name: Devastator
  Vendor: Advanced Micro Devices, Inc.
  Driver version: 1268.1 (VM)
  Version: OpenCL 1.2 AMD-APP (1268.1)
  Extensions: cl_amd_fp64 cl_khr_global_int32_base_atomics cl_khr_global_int32_extended_atomics cl_khr_local_int32_base_atomics cl_khr_local_int32_extended_atomics cl_khr_3d_image_writes cl_khr_byte_addressable_store cl_khr_gl_sharing cl_ext_atomic_counters_32 cl_amd_device_attribute_query cl_amd_vec3 cl_amd_printf cl_amd_media_ops cl_amd_media_ops2 cl_amd_popcnt cl_khr_d3d10_sharing cl_khr_dx9_media_sharing cl_amd_image2d_from_buffer_read_only

The error your GPU  tasks report here is:

Quote:
<core_client_version>7.4.27</core_client_version>
<![CDATA[
<message>
(unknown error) - exit code -161 (0xffffff5f)
</message>
<stderr_txt>
Activated exception handling...
[23:07:50][5536][INFO ] Starting data processing...
GPU type not found in init_data.xml
[23:07:50][5536][ERROR] Failed to get OpenCL platform/device info from BOINC (error: -161)!
[23:07:50][5536][ERROR] Demodulation failed (error: -161)!
23:07:50 (5536): called boinc_finish

I wonder is there is an api Bug present in the BRP apps where GPUs and OpenCL platforms aren't aligned correctly,

this api Bug fix might be related (Albert GPU apps predate it):

http://boinc.berkeley.edu/gitweb/?p=boinc-v2.git;a=commit;h=328d87be2625ce68a04b0d1caf29e3826eab25f3

http://albert.phys.uwm.edu/apps.php

in the meantime tullio can you update to Boinc 7.4.36 please.

Claggy

tullio
tullio
Joined: 22 Jan 05
Posts: 53
Credit: 137342
RAC: 0

I have both an onboard

I have both an onboard graphic, called Devastator, and a graphic board , called Caicos, on this Windows PC. SETI recognizes both, Einstein and Albert only the first. SETI GPU tasks are completed and validated, as are BRP4G in Einstein. BRP5 seem to fail both in Einstein and Albert. Thanks for your suggestions. I am really new to the use of graphic boards. Maybe one day I shall install the driver for my Geforge GTX 750 in my SUN Linux host with SuSE 13.1.

Tullio

tullio
tullio
Joined: 22 Jan 05
Posts: 53
Credit: 137342
RAC: 0

I installed BOINC 7.4.36 as

I installed BOINC 7.4.36 as suggested by Claggy but things do not change and all Albert@home tasks fail with computation error after 2 seconds.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.