There are some special cases introduced in gen_insn_selection stage
which will do more than one writing on the same virtual registers
with different stride or simd width with different suboffset.
For this type of registers, the current register spilling implementation
is broken. To avoid unecessar
LGTM, pushed, thanks.
On Thu, May 21, 2015 at 11:31:21AM +0800, Yang Rong wrote:
> If use arg as non add instruction's source directly, for example phi and
> selection, there is no add, just skip it.
>
> Signed-off-by: Yang Rong
> ---
> backend/src/ir/lowering.cpp | 1 +
> 1 file changed, 1 in
Hi,
For people who are interested Beignet compliant of OpenCL spec,
I am happy to announce that Beignet has passed Khronos OpenCL 1.2
conformance certification on BDW.
https://www.khronos.org/conformance/adopters/conformant-products
Thanks
Zou Nanhai
__