On Wed, Aug 21, 2013 at 07:21:52PM +0300, Pekka Jääskeläinen wrote:
> On 08/21/2013 07:12 PM, Peter Colberg wrote:
> > In a few years this segmentation fault in the proprietary OpenCL
> > driver will be worked around, when we have production-ready free
> > software implementations for all CPU and GPU architectures…:-)
> 
> That makes me ask: Did you give up with your libcuda/NVPTX pocl
> driver experiment? :)

Yes, sorry about the silence after your comprehensive notes on how to
progress with the backend. It seemed a lot of work back then, and I
opted to develop my OpenCL application first using the proprietary
drivers. I worked out a few tricks to make the NVIDIA OpenCL driver
usable, since it is all too easy to end up with unforgivable gaps of
≥ 300µs in the execution sequence if one is not careful with the
choice of OpenCL calls.

Peter

------------------------------------------------------------------------------
Introducing Performance Central, a new site from SourceForge and 
AppDynamics. Performance Central is your source for news, insights, 
analysis and resources for efficient Application Performance Management. 
Visit us today!
http://pubads.g.doubleclick.net/gampad/clk?id=48897511&iu=/4140/ostg.clktrk
_______________________________________________
pocl-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/pocl-devel

Reply via email to