[coproc_debug] Insufficient CUDA ????
3 Jun 2012 5:50:58 UTC
Topic 196375
(moderation:
I have a warning / Error message occurring that appears to indicate problems. However the GPU WU process and upload as I would expect.
All work units get the message.
Below are messages relating to just one WU extracted from the messages file.
Host Project Date Message
UBUNTU64 Einstein@Home 3/06/2012 9:49:40 AM [coproc_debug] Insufficient CUDA for p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0: need 0.500000
Host Project Date Message
UBUNTU64 Einstein@Home 3/06/2012 10:43:37 AM Starting p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0
Host Project Date Message
UBUNTU64 Einstein@Home 3/06/2012 12:39:22 PM Computation for task p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0 finished
Host Project Date Message
UBUNTU64 Einstein@Home 3/06/2012 12:39:25 PM Started upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_0
UBUNTU64 Einstein@Home 3/06/2012 12:39:25 PM Started upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_1
UBUNTU64 Einstein@Home 3/06/2012 12:39:28 PM Finished upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_0
UBUNTU64 Einstein@Home 3/06/2012 12:39:28 PM Finished upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_1
UBUNTU64 Einstein@Home 3/06/2012 12:39:28 PM Started upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_2
UBUNTU64 Einstein@Home 3/06/2012 12:39:28 PM Started upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_3
UBUNTU64 Einstein@Home 3/06/2012 12:39:30 PM Finished upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_2
UBUNTU64 Einstein@Home 3/06/2012 12:39:30 PM Finished upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_3
UBUNTU64 Einstein@Home 3/06/2012 12:39:30 PM Started upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_4
UBUNTU64 Einstein@Home 3/06/2012 12:39:30 PM Started upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_5
UBUNTU64 Einstein@Home 3/06/2012 12:39:32 PM Finished upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_4
UBUNTU64 Einstein@Home 3/06/2012 12:39:32 PM Finished upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_5
UBUNTU64 Einstein@Home 3/06/2012 12:39:32 PM Started upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_6
UBUNTU64 Einstein@Home 3/06/2012 12:39:32 PM Started upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_7
UBUNTU64 Einstein@Home 3/06/2012 12:39:34 PM Finished upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_6
UBUNTU64 Einstein@Home 3/06/2012 12:39:34 PM Finished upload of p2030.20100912.G44.05+00.67.C.b5s0g0.00000_2656_0_7
Computer ID #4213006
CPU type GenuineIntel
06/17 [Family 6 Model 23 Stepping 12 ***** Q9400
Number of processors 4
Coprocessors NVIDIA GeForce 9800 GTX+ (511MB)
Operating System Linux
2.6.38-15-generic
BOINC client version 6.10.59
Memory 3960.85 MB
Nvidia drivers 295.53 - from UBUNTU X-Swat
Application: Binary Radio Pulsar Search (Arecibo) v1.24 (BRP4cuda32nv270)
WU complete in 6900 - 7000 sec per task for 500 credits
[coproc_debug] Insufficient CUDA ????
)
Hi
I guess you configured your E@H project preferences to crunch 2 GPU tasks in parallel (0.5 GPU per task). However, you might not have enough free Video RAM on the 512 MB card in that host to run two in parallel, so I guess this is BOINC's way of telling you this in the extended logging output (this warning message is only visible when you set special flags in cc_config.xml.
So your 512MB card should execute one GPU task at a time, while your second host should do 2 in parallel. Right?
Cheers
HB
Ahhh !! Edit cc_config and it
)
Ahhh !!
Edit cc_config and it goes away. Thanks
Correct
EDIT:
*** wrong - no cc_config to edit ***
set from web interface just put up with it ...
It works don't fix it ...