I also recommend porting lp_create_builder, which might also improve things.
Marek
On Wed, Jan 10, 2018 at 10:31 PM, Samuel Pitoiset
<samuel.pitoi...@gmail.com> wrote:
> This is an optimisation that is recommended by Matt Arsenault,
> and used by RadeonSI.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoi...@gmail.com>
> ---
> src/amd/common/ac_nir_to_llvm.c | 5 +++++
> 1 file changed, 5 insertions(+)
>
> diff --git a/src/amd/common/ac_nir_to_llvm.c b/src/amd/common/ac_nir_to_llvm.c
> index 1520342d05..b474c29244 100644
> --- a/src/amd/common/ac_nir_to_llvm.c
> +++ b/src/amd/common/ac_nir_to_llvm.c
> @@ -338,6 +338,11 @@ create_llvm_function(LLVMContextRef ctx, LLVMModuleRef
> module,
>
> "amdgpu-max-work-group-size",
> max_workgroup_size);
> }
> +
> + LLVMAddTargetDependentFunctionAttr(main_function,
> + "no-signed-zeros-fp-math",
> + "true");
> +
> if (unsafe_math) {
> /* These were copied from some LLVM test. */
> LLVMAddTargetDependentFunctionAttr(main_function,
> --
> 2.15.1
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev@lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev
_______________________________________________
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
