Message boards :
Number crunching :
Something changed with linux crunching
Message board moderation
Author | Message |
---|---|
Send message Joined: 17 Feb 09 Posts: 24 Credit: 3,446,458 RAC: 631 |
Started a dual intel machine up running fedora core 10 linux. This machine has done MW w/u's before fine. Not this time. Downloaded the first w/u and it errored straight away with "SIGILL: illegal instruction", This machine is running seti and Einstein fine. Its fc10 (2.6.27.5-117.fc10.i686) Same thing with a fedora core 8 linux with dual AMD cpus. (2.6.23.1-42.fc8) This has run seti for ages fine. Both machines used to run MW fine. Any thing changed?. Nairb |
Send message Joined: 29 Sep 10 Posts: 1 Credit: 6,548 RAC: 0 |
Seeing similar issue this morning on my linux box. Found boinc client down... on attempted restart got the following: ... 11-Oct-2010 11:10:48 [Milkyway@home] Restarting task de_11_2s_5_609775_1286475414_2 using milkyway version 4 SIGSEGV: segmentation violation Stack trace (10 frames): /usr/lib64/libboinc.so.6(boinc_catch_signal+0x4d)[0x7f23f6d8bd2d] /lib64/libc.so.6[0x3015032a20] /usr/lib64/libboinc.so.6(_ZN14APP_CLIENT_SHM10reset_msgsEv+0x26)[0x7f23f6d87356] boinc[0x411f0f] boinc[0x412633] boinc[0x42908c] boinc[0x418304] boinc[0x4490ec] /lib64/libc.so.6(__libc_start_main+0xfd)[0x301501ec5d] boinc[0x4076f9] Exiting... Running: Linux 2.6.34.7-56.fc13.x86_64 Bob |
Send message Joined: 8 May 10 Posts: 576 Credit: 15,979,383 RAC: 0 |
Started a dual intel machine up running fedora core 10 linux. This machine has done MW w/u's before fine. Not this time. Downloaded the first w/u and it errored straight away with "SIGILL: illegal instruction", This is probably my fault. It looks like I accidentally built the 32 bit Linux one with SSE2 and the old processor doesn't have it. |
Send message Joined: 8 May 10 Posts: 576 Credit: 15,979,383 RAC: 0 |
Seeing similar issue this morning on my linux box. Found boinc client down... on attempted restart got the following: Not sure about this one. |
Send message Joined: 17 Feb 09 Posts: 24 Credit: 3,446,458 RAC: 631 |
Haaaaa, I thought something must have changed. 2 out of 2 machines with a fault looked a bit iffy. Are we going to have a new app?. I only have 1 windoz pc. The others are all linux of one sort or another Ta Nairb |
Send message Joined: 17 Feb 09 Posts: 24 Credit: 3,446,458 RAC: 631 |
Ummmm is this a linux problem or an old cpu problem. I dont have a "modern" cpu running linux to check. And I dont have an old cpu running windows xp to check. If its a SSE2 issue its not an linux problem is it??. Just checking Ta Nairb |
Send message Joined: 9 May 10 Posts: 3 Credit: 1,589,065 RAC: 0 |
I seem to be seeing a problem on my CentOS 5.5 x64 machine running a nVidia GTX 275. System appears to have received a recent kernel update (over weekend). As a result, I also upgraded the GPU device drivers from nVidia (dtd 11/11/10). Since then, I seem to be seeing frequent compute errors (machine ID 194981).... most message indicate within task info following error: Error executing gpu__integral_kernel3 error message: unknown error Any suggestions for repair? |
Send message Joined: 9 May 10 Posts: 3 Credit: 1,589,065 RAC: 0 |
I have since moved my NVIDIA driver back to the prior release (oct 2010) and driver / GPU seem to be performing well on this project and others. |
©2024 Astroinformatics Group