Computation error at 6 seconds

Message boards : Bug tracker : Computation error at 6 seconds

To post messages, you must log in.

AuthorMessage
Peppernrino
Avatar

Send message
Joined: 30 Jan 17
Posts: 17
Credit: 486,804,901
RAC: 0
   
Message 507 - Posted: 27 Jun 2017, 19:07:08 UTC

Tasks are erroring out lately on 750ti on Linux BILD. Any ideas?

Peppernrino
ID: 507 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sergei Chernykh
Project administrator
Project developer

Send message
Joined: 5 Jan 17
Posts: 506
Credit: 72,451,573
RAC: 0
   
Message 508 - Posted: 27 Jun 2017, 20:20:25 UTC

Initializing prime tables...terminate called after throwing an instance of 'std::bad_alloc'
Host #2209: Memory	1971.45 MB, Swap space: 0 MB

That host has only 2 GB of memory, it just doesn't have enough free memory. Try to create 4 GB swap file.
ID: 508 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Peppernrino
Avatar

Send message
Joined: 30 Jan 17
Posts: 17
Credit: 486,804,901
RAC: 0
   
Message 510 - Posted: 27 Jun 2017, 21:10:26 UTC - in response to Message 508.  

Added 2GB of RAM, and it is working fine now.

Thank you very much! :D
ID: 510 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Matt

Send message
Joined: 26 Nov 17
Posts: 2
Credit: 281,268,350
RAC: 156,444
    
Message 692 - Posted: 24 Dec 2017, 8:03:18 UTC
Last modified: 24 Dec 2017, 8:10:07 UTC

Is it ok if i post to this thread since my problem is VERY similar, I get a computation error after about 7-8 seconds ever since I've added in a second GPU for boincing. The main GPU (the one that was in there first anbd the strongest) is able to compute fine but the second card will reliably fail with the computations. Both cards are fine and working, not SLI, I did have to enter the <use all gpus> line in cc_config to have them both be used, when I go to the tasks page the below is what I see. The top half doesn't seem to add much.

Stderr output
<core_client_version>7.8.3</core_client_version>
<![CDATA[
<message>
(unknown error) - exit code -1 (0xffffffff)</message>
<stderr_txt>
Initializing prime tables...done
OpenCL.cpp, line 294: Preferences:
<project_preferences>


<max_jobs>0</max_jobs>
<max_cpus>3</max_cpus>
<kernel_size_amd>21</kernel_size_amd>
<kernel_size_nvidia>17</kernel_size_nvidia>
</project_preferences>


OpenCL.cpp, line 307: Kernel size for NVIDIA GPU has been set to 17
OpenCL.cpp, line 440: clEnqueueWriteBuffer returned error -4
18:24:31 (6364): called boinc_finish(-1)

</stderr_txt>
]]>
ID: 692 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sergei Chernykh
Project administrator
Project developer

Send message
Joined: 5 Jan 17
Posts: 506
Credit: 72,451,573
RAC: 0
   
Message 693 - Posted: 24 Dec 2017, 10:16:40 UTC - in response to Message 692.  

OpenCL.cpp, line 440: clEnqueueWriteBuffer returned error -4

Error -4 is CL_MEM_OBJECT_ALLOCATION_FAILURE. What are your second GPU's specs? Does it have at least 2 GB of memory?
ID: 693 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Matt

Send message
Joined: 26 Nov 17
Posts: 2
Credit: 281,268,350
RAC: 156,444
    
Message 694 - Posted: 26 Dec 2017, 12:16:45 UTC - in response to Message 693.  

It's a 560 Ti so no it has 1024 of memory.

Also I added another project and allocated this card towards it to see if same error. Did not fail, so successfully crunching with this card on another project.

If that 2GB memory minimum is a project specific thing then I guess all is working as intended, must have missed the line that mentioned it.
ID: 694 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sergei Chernykh
Project administrator
Project developer

Send message
Joined: 5 Jan 17
Posts: 506
Credit: 72,451,573
RAC: 0
   
Message 695 - Posted: 26 Dec 2017, 13:16:16 UTC - in response to Message 694.  

It's a 560 Ti so no it has 1024 of memory.

Also I added another project and allocated this card towards it to see if same error. Did not fail, so successfully crunching with this card on another project.

If that 2GB memory minimum is a project specific thing then I guess all is working as intended, must have missed the line that mentioned it.

You need to have 2 GB of GPU memory to run GPU version of Amicable Numbers, so it's normal that 560 Ti errors out.
ID: 695 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Lamplighter55

Send message
Joined: 23 Jan 18
Posts: 2
Credit: 6,118,393
RAC: 0
   
Message 749 - Posted: 3 Mar 2018, 10:12:40 UTC

My iMac (i7 late 2015 AMD R9 GPU M395X 4GB) errors out on this build 'Amicable Numbers up to 10^20 v2.08 (opencl_amd) x86_64-apple-darwin'. Prior to this release it was running fine with v2.06.
ID: 749 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sergei Chernykh
Project administrator
Project developer

Send message
Joined: 5 Jan 17
Posts: 506
Credit: 72,451,573
RAC: 0
   
Message 750 - Posted: 3 Mar 2018, 10:21:52 UTC - in response to Message 749.  
Last modified: 3 Mar 2018, 10:22:08 UTC

MacOS didn't like new OpenCL code:
:1516:29: error: comparison of integers of different signs: 'ulong' (aka 'unsigned long') and 'int'
if (amicableCandidateIndex >= candidatesDataHighBitOffsets.x) value_ulong |= 0x100000000UL;

I'll fix it ASAP.
ID: 750 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sergei Chernykh
Project administrator
Project developer

Send message
Joined: 5 Jan 17
Posts: 506
Credit: 72,451,573
RAC: 0
   
Message 751 - Posted: 3 Mar 2018, 10:52:25 UTC - in response to Message 749.  

Lamplighter55
Can you try again with version 2.09?
ID: 751 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Lamplighter55

Send message
Joined: 23 Jan 18
Posts: 2
Credit: 6,118,393
RAC: 0
   
Message 752 - Posted: 3 Mar 2018, 12:25:39 UTC - in response to Message 751.  

Thanks Sergei, your quick fix looks like it's working fine. I'll report if there are any further issues on my (MacOS) client side.
ID: 752 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Sergei Chernykh
Project administrator
Project developer

Send message
Joined: 5 Jan 17
Posts: 506
Credit: 72,451,573
RAC: 0
   
Message 759 - Posted: 9 Mar 2018, 9:21:20 UTC

MacOS GPU version broke again when the project switched to the new type of work units, sorry for the mess. I'll fix it later today when I have time.
ID: 759 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Bug tracker : Computation error at 6 seconds


©2024 Sergei Chernykh