Am 04.01.2012 16:03, schrieb Andreas Kloeckner:
On Wed, 04 Jan 2012 15:40:12 +0100, Omar<[email protected]> wrote:
Hi everyone,
1. is there a method to view the current memory usage of the gpu?
2. What could this error mean?
"pyopencl.RuntimeError: clWaitForEvents failed: out of resources"
Mind you, all i do is call a Kernel-object and .wait() for the resulting
event like this:
kernel(...).wait()
That's one of the ways Nvidia GPUs say 'you had a segfault in your
kernel'. (Do note, however, that the error message also arises in a few
other contexts--but 'segfault' is the most common one.)
Thanks, but now i ran the examples provided by pyopencl and transpose.py
gives me a similar error (out of memory). Here's the log:
('benchmarking', 'Silly', 1568, 0.8430220258184548, 'GB/s')
('benchmarking', 'Silly', 1696, 0.8314101965016066, 'GB/s')
('benchmarking', 'Silly', 1856, 0.8336174518261653, 'GB/s')
('benchmarking', 'Silly', 2048, 0.6867860790222692, 'GB/s')
('benchmarking', 'Silly', 2208, 0.9463517188892961, 'GB/s')
('benchmarking', 'Silly', 2432, 0.6080511709729958, 'GB/s')
('benchmarking', 'Silly', 2624, 0.6751110587742045, 'GB/s')
('benchmarking', 'Silly', 2880, 0.7625399316670668, 'GB/s')
('benchmarking', 'Silly', 3136, 0.7611269735011305, 'GB/s')
('benchmarking', 'Silly', 3424, 0.6619695642140532, 'GB/s')
('benchmarking', 'Silly', 3744, 0.6421031214735665, 'GB/s')
('benchmarking', 'Silly', 4096, 0.4020922138938147, 'GB/s')
Traceback (most recent call last):
File "transpose.py", line 212, in <module>
benchmark_transpose()
File "transpose.py", line 180, in benchmark_transpose
events[-1].wait()
pyopencl.MemoryError: clWaitForEvents failed: mem object allocation failure
pyopencl version 2011.2
Andreas
_______________________________________________
PyOpenCL mailing list
[email protected]
http://lists.tiker.net/listinfo/pyopencl
