<#part sign=pgpmime>
On Tue, 3 Apr 2012 10:14:09 +0300, Alex Toldayev <[email protected]> 
wrote:
> Hi everyone!
> 
> I'd like to ask if there is any chance to build an opencl program from
> several .cl files?
> 
> I've found this useful thread
> 
> http://stackoverflow.com/questions/7622293/how-to-structure-large-opencl-kernels
> 
> They say there is the feature in OpenCL 1.2 and with Nvidia compiler you
> can #include "anotherpart.cl" file. But including doesn't work for me (i
> have an ATI card):
> 
> /tmp/OCLG3u4M4.cl(8): catastrophic error: cannot open source file "add.cl"
>   #include "add.cl"
>                    ^
> 
> 1 catastrophic error detected in the compilation of "/tmp/OCLG3u4M4.cl".
> Compilation terminated.
> 
> Internal error: clc compiler invocation failed.

Add options=["-I", "."] to build().

> And I'd like to ask whether anybody uses linking object files in openCL 1.2
> (I can't find useful reference on the web)? Or, maybe, there is a simpler
> way to split a file of kernels? (It seems the issue isn't very hard. So if
> the problem remains I could write a little parsing-utility for making a one
> cl-string from multiple files. Maybe someone already has done it?)
> Any advise or reference would be helpful.

Include files are the way to go for now, I think. CL 1.2 isn't widely
implemented yet, and I haven't exactly figured out how the object-level
linking is going to work with PyOpenCL's compiler caching.

Andreas

_______________________________________________
PyOpenCL mailing list
[email protected]
http://lists.tiker.net/listinfo/pyopencl

Reply via email to