10:25 p.m.
On Donnerstag 22 April 2010, Michael Rule wrote:
NV Tesla T10,
I assume its something I am doing wrong...
If you're using NV hardware, it helps to know the CUDA literature. You
must supply a local_size to partition your work into NV's thread
blocks. If you don't you're only submitting one thread block, which has
hardware size limits.
Andreas
PS: Please keep replies on the list. Thanks.
10:37 p.m.
New subject: why do calls for large number of threads fail ?
thank you, that solved all my problems ! ( I mean.... that one
problem, which was pervasive in my program )
On Thu, Apr 22, 2010 at 4:25 PM, Andreas Klöckner
<lists(a)informa.tiker.net> wrote:
On Donnerstag 22 April 2010, Michael Rule wrote:
> NV Tesla T10,
> I assume its something I am doing wrong...
If you're using NV hardware, it helps to know the CUDA literature. You
must supply a local_size to partition your work into NV's thread
blocks. If you don't you're only submitting one thread block, which has
hardware size limits.
Andreas
PS: Please keep replies on the list. Thanks.
midnight
New subject: why do calls for large number of threads fail ?
The hardware-independent way to solve this is to
call clGetKernelWorkGroupInfo() with an argument
of CL_KERNEL_WORK_GROUP_SIZE to get the maximum number of work-items that
you can enqueue at once.
See also CL_KERNEL_LOCAL_MEM_SIZE to find out how much local memory will be
available for that kernel.
Andreas, I assume that this is also exposed through PyOpenCL somehow :)
Cheers,
David
On Thu, Apr 22, 2010 at 4:37 PM, Michael Rule <mrule7404(a)gmail.com> wrote:
thank you, that solved all my problems ! ( I mean.... that one
problem, which was pervasive in my program )
On Thu, Apr 22, 2010 at 4:25 PM, Andreas Klöckner
<lists(a)informa.tiker.net> wrote:
> On Donnerstag 22 April 2010, Michael Rule wrote:
>> NV Tesla T10,
>> I assume its something I am doing wrong...
>
> If you're using NV hardware, it helps to know the CUDA literature. You
> must supply a local_size to partition your work into NV's thread
> blocks. If you don't you're only submitting one thread block, which has
> hardware size limits.
>
> Andreas
>
> PS: Please keep replies on the list. Thanks.
>
_______________________________________________
PyOpenCL mailing list
PyOpenCL(a)host304.hostmonster.com
http://host304.hostmonster.com/mailman/listinfo/pyopencl_tiker.net
12:13 a.m.
New subject: why do calls for large number of threads fail ?
On Donnerstag 22 April 2010, David Garcia wrote:
The hardware-independent way to solve this is to
call clGetKernelWorkGroupInfo() with an argument
of CL_KERNEL_WORK_GROUP_SIZE to get the maximum number of work-items that
you can enqueue at once.
See also CL_KERNEL_LOCAL_MEM_SIZE to find out how much local memory will be
available for that kernel.
Andreas, I assume that this is also exposed through PyOpenCL somehow :)
http://documen.tician.de/pyopencl/reference.html#pyopencl.Kernel.get_work...
Andreas
12:25 a.m.
New subject: why do calls for large number of threads fail ?
Thanks! Is it missing CL_KERNEL_LOCAL_MEM_SIZE?
David
On Thu, Apr 22, 2010 at 6:13 PM, Andreas Klöckner
<lists(a)informa.tiker.net>wrote:
On Donnerstag 22 April 2010, David Garcia wrote:
> The hardware-independent way to solve this is to
> call clGetKernelWorkGroupInfo() with an argument
> of CL_KERNEL_WORK_GROUP_SIZE to get the maximum number of work-items that
> you can enqueue at once.
>
> See also CL_KERNEL_LOCAL_MEM_SIZE to find out how much local memory will
be
> available for that kernel.
>
> Andreas, I assume that this is also exposed through PyOpenCL somehow :)
http://documen.tician.de/pyopencl/reference.html#pyopencl.Kernel.get_work...
Andreas
_______________________________________________
PyOpenCL mailing list
PyOpenCL(a)host304.hostmonster.com
http://host304.hostmonster.com/mailman/listinfo/pyopencl_tiker.net
3344
days inactive
3344
days old
5 comments
3 participants
participants (3)
-
Andreas Klöckner -
David Garcia -
Michael Rule