On 08/21/2013 11:56 PM, Chuck Zhao wrote: > At clEnqueueNDRangeKernel() time, all workgroup details are available. > If I have a local workgroup of (2, 1, 1), can I understand that the > kernel is replicated/unrolled 2 * 1 * 1 = 2 times, regardless other > global dimensions?
Yes. It generates a function that executes the kernel for a work-group. Then that function is called for all work-groups in the work-space. pthread device driver executes multiple WGs in parallel using multiple threads, basic driver in a single thread sequentially (a for loop), etc. -- --Pekka ------------------------------------------------------------------------------ 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
