[Beignet] [PATCH] Remove global barrier assert.

[Zhigang Gong]

With this patch, you treat a global memory fence the same as a local memory fence.
I haven't check the bspec details, but I am doubt it. Need some time to investigate
the spec.

On Fri, Jun 14, 2013 at 11:01:09AM +0800, Yang Rong wrote:
> Per openCL spec, Global memory barrier is consistent across work-items in
> a single work group, which is match the bspec's barrier. So remove global
> barrier assert.
> 
> Signed-off-by: Yang Rong <rong.r.yang at intel.com>
> ---
>  backend/src/backend/gen_insn_selection.cpp |    4 ----
>  1 file changed, 4 deletions(-)
> 
> diff --git a/backend/src/backend/gen_insn_selection.cpp b/backend/src/backend/gen_insn_selection.cpp
> index 88f9e94..3a139ea 100644
> --- a/backend/src/backend/gen_insn_selection.cpp
> +++ b/backend/src/backend/gen_insn_selection.cpp
> @@ -1607,10 +1607,6 @@ namespace gbe
>      INLINE bool emitOne(Selection::Opaque &sel, const ir::SyncInstruction &insn) const
>      {
>        using namespace ir;
> -      const uint32_t params = insn.getParameters();
> -      GBE_ASSERTM(params == syncLocalBarrier,
> -                  "Only barrier(CLK_LOCAL_MEM_FENCE) is supported right now "
> -                  "for the synchronization primitives");
>        const ir::Register reg = sel.reg(FAMILY_DWORD);
>  
>        sel.push();
> -- 
> 1.7.10.4
> 
> _______________________________________________
> Beignet mailing list
> Beignet at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/beignet