[Beignet] [PATCH] Add memory fence before barrier to support global memory barrier.

[Zou, Nanhai]

>> I think we should limit the thread group size according to SLM size per half slice. HW will not automatically split them.


To be more specific, 
We need to set a limit of thread/work group size according to SLM usage, if kernel use SLM/barrier.
If kernel is not using SLM/barrier. Thread group size is unlimited.



_______________________________________________
Beignet mailing list
Beignet at lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/beignet