8f9ee39c58
This change implements #70074 - `hlsl_intrinsics.h` - add the `rsqrt` api - `DXIL.td` add the llvm intrinsic to DXIL op lowering map. - `Builtins.td` - add an hlsl builtin for rsqrt. - `CGBuiltin.cpp` add the ir generation for the rsqrt intrinsic. - `SemaChecking.cpp` - reuse the one arg float only checks. - `IntrinsicsDirectX.td` -add an `rsqrt` intrinsic.
29 lines
883 B
LLVM
29 lines
883 B
LLVM
; RUN: opt -S -dxil-op-lower < %s | FileCheck %s
|
|
|
|
; Make sure dxil operation function calls for rsqrt are generated for float and half.
|
|
|
|
; CHECK-LABEL: rsqrt_float
|
|
; CHECK: call float @dx.op.unary.f32(i32 25, float %{{.*}})
|
|
define noundef float @rsqrt_float(float noundef %a) {
|
|
entry:
|
|
%a.addr = alloca float, align 4
|
|
store float %a, ptr %a.addr, align 4
|
|
%0 = load float, ptr %a.addr, align 4
|
|
%dx.rsqrt = call float @llvm.dx.rsqrt.f32(float %0)
|
|
ret float %dx.rsqrt
|
|
}
|
|
|
|
; CHECK-LABEL: rsqrt_half
|
|
; CHECK: call half @dx.op.unary.f16(i32 25, half %{{.*}})
|
|
define noundef half @rsqrt_half(half noundef %a) {
|
|
entry:
|
|
%a.addr = alloca half, align 2
|
|
store half %a, ptr %a.addr, align 2
|
|
%0 = load half, ptr %a.addr, align 2
|
|
%dx.rsqrt = call half @llvm.dx.rsqrt.f16(half %0)
|
|
ret half %dx.rsqrt
|
|
}
|
|
|
|
declare half @llvm.dx.rsqrt.f16(half)
|
|
declare float @llvm.dx.rsqrt.f32(float)
|