mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2024-11-25 20:23:11 +01:00
119ff7dab3
The architecture doesn't really have any native v4f32 operations except v4f32->v2f64 and v2f64->v4f32 conversions, with only half of the v4f32 elements being used. Even so, using vector registers for <4 x float> and scalarising individual operations is much better than generating completely scalar code, since there's much less register pressure. It's also more efficient to do v4f32 comparisons by extending to 2 v2f64s, comparing those, then packing the result. This particularly helps with llvmpipe. Based on a patch by Richard Sandiford. llvm-svn: 236523
201 lines
6.6 KiB
LLVM
201 lines
6.6 KiB
LLVM
; Test replications of a scalar register value, represented as splats.
|
|
;
|
|
; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 | FileCheck %s
|
|
|
|
; Test v16i8 splat of the first element.
|
|
define <16 x i8> @f1(i8 %scalar) {
|
|
; CHECK-LABEL: f1:
|
|
; CHECK: vlvgp [[REG:%v[0-9]+]], %r2, %r2
|
|
; CHECK: vrepb %v24, [[REG]], 7
|
|
; CHECK: br %r14
|
|
%val = insertelement <16 x i8> undef, i8 %scalar, i32 0
|
|
%ret = shufflevector <16 x i8> %val, <16 x i8> undef,
|
|
<16 x i32> zeroinitializer
|
|
ret <16 x i8> %ret
|
|
}
|
|
|
|
; Test v16i8 splat of the last element.
|
|
define <16 x i8> @f2(i8 %scalar) {
|
|
; CHECK-LABEL: f2:
|
|
; CHECK: vlvgp [[REG:%v[0-9]+]], %r2, %r2
|
|
; CHECK: vrepb %v24, [[REG]], 7
|
|
; CHECK: br %r14
|
|
%val = insertelement <16 x i8> undef, i8 %scalar, i32 15
|
|
%ret = shufflevector <16 x i8> %val, <16 x i8> undef,
|
|
<16 x i32> <i32 15, i32 15, i32 15, i32 15,
|
|
i32 15, i32 15, i32 15, i32 15,
|
|
i32 15, i32 15, i32 15, i32 15,
|
|
i32 15, i32 15, i32 15, i32 15>
|
|
ret <16 x i8> %ret
|
|
}
|
|
|
|
; Test v16i8 splat of an arbitrary element, using the second operand of
|
|
; the shufflevector.
|
|
define <16 x i8> @f3(i8 %scalar) {
|
|
; CHECK-LABEL: f3:
|
|
; CHECK: vlvgp [[REG:%v[0-9]+]], %r2, %r2
|
|
; CHECK: vrepb %v24, [[REG]], 7
|
|
; CHECK: br %r14
|
|
%val = insertelement <16 x i8> undef, i8 %scalar, i32 4
|
|
%ret = shufflevector <16 x i8> undef, <16 x i8> %val,
|
|
<16 x i32> <i32 20, i32 20, i32 20, i32 20,
|
|
i32 20, i32 20, i32 20, i32 20,
|
|
i32 20, i32 20, i32 20, i32 20,
|
|
i32 20, i32 20, i32 20, i32 20>
|
|
ret <16 x i8> %ret
|
|
}
|
|
|
|
; Test v8i16 splat of the first element.
|
|
define <8 x i16> @f4(i16 %scalar) {
|
|
; CHECK-LABEL: f4:
|
|
; CHECK: vlvgp [[REG:%v[0-9]+]], %r2, %r2
|
|
; CHECK: vreph %v24, [[REG]], 3
|
|
; CHECK: br %r14
|
|
%val = insertelement <8 x i16> undef, i16 %scalar, i32 0
|
|
%ret = shufflevector <8 x i16> %val, <8 x i16> undef,
|
|
<8 x i32> zeroinitializer
|
|
ret <8 x i16> %ret
|
|
}
|
|
|
|
; Test v8i16 splat of the last element.
|
|
define <8 x i16> @f5(i16 %scalar) {
|
|
; CHECK-LABEL: f5:
|
|
; CHECK: vlvgp [[REG:%v[0-9]+]], %r2, %r2
|
|
; CHECK: vreph %v24, [[REG]], 3
|
|
; CHECK: br %r14
|
|
%val = insertelement <8 x i16> undef, i16 %scalar, i32 7
|
|
%ret = shufflevector <8 x i16> %val, <8 x i16> undef,
|
|
<8 x i32> <i32 7, i32 7, i32 7, i32 7,
|
|
i32 7, i32 7, i32 7, i32 7>
|
|
ret <8 x i16> %ret
|
|
}
|
|
|
|
; Test v8i16 splat of an arbitrary element, using the second operand of
|
|
; the shufflevector.
|
|
define <8 x i16> @f6(i16 %scalar) {
|
|
; CHECK-LABEL: f6:
|
|
; CHECK: vlvgp [[REG:%v[0-9]+]], %r2, %r2
|
|
; CHECK: vreph %v24, [[REG]], 3
|
|
; CHECK: br %r14
|
|
%val = insertelement <8 x i16> undef, i16 %scalar, i32 2
|
|
%ret = shufflevector <8 x i16> undef, <8 x i16> %val,
|
|
<8 x i32> <i32 10, i32 10, i32 10, i32 10,
|
|
i32 10, i32 10, i32 10, i32 10>
|
|
ret <8 x i16> %ret
|
|
}
|
|
|
|
; Test v4i32 splat of the first element.
|
|
define <4 x i32> @f7(i32 %scalar) {
|
|
; CHECK-LABEL: f7:
|
|
; CHECK: vlvgp [[REG:%v[0-9]+]], %r2, %r2
|
|
; CHECK: vrepf %v24, [[REG]], 1
|
|
; CHECK: br %r14
|
|
%val = insertelement <4 x i32> undef, i32 %scalar, i32 0
|
|
%ret = shufflevector <4 x i32> %val, <4 x i32> undef,
|
|
<4 x i32> zeroinitializer
|
|
ret <4 x i32> %ret
|
|
}
|
|
|
|
; Test v4i32 splat of the last element.
|
|
define <4 x i32> @f8(i32 %scalar) {
|
|
; CHECK-LABEL: f8:
|
|
; CHECK: vlvgp [[REG:%v[0-9]+]], %r2, %r2
|
|
; CHECK: vrepf %v24, [[REG]], 1
|
|
; CHECK: br %r14
|
|
%val = insertelement <4 x i32> undef, i32 %scalar, i32 3
|
|
%ret = shufflevector <4 x i32> %val, <4 x i32> undef,
|
|
<4 x i32> <i32 3, i32 3, i32 3, i32 3>
|
|
ret <4 x i32> %ret
|
|
}
|
|
|
|
; Test v4i32 splat of an arbitrary element, using the second operand of
|
|
; the shufflevector.
|
|
define <4 x i32> @f9(i32 %scalar) {
|
|
; CHECK-LABEL: f9:
|
|
; CHECK: vlvgp [[REG:%v[0-9]+]], %r2, %r2
|
|
; CHECK: vrepf %v24, [[REG]], 1
|
|
; CHECK: br %r14
|
|
%val = insertelement <4 x i32> undef, i32 %scalar, i32 1
|
|
%ret = shufflevector <4 x i32> undef, <4 x i32> %val,
|
|
<4 x i32> <i32 5, i32 5, i32 5, i32 5>
|
|
ret <4 x i32> %ret
|
|
}
|
|
|
|
; Test v2i64 splat of the first element.
|
|
define <2 x i64> @f10(i64 %scalar) {
|
|
; CHECK-LABEL: f10:
|
|
; CHECK: vlvgp %v24, %r2, %r2
|
|
; CHECK: br %r14
|
|
%val = insertelement <2 x i64> undef, i64 %scalar, i32 0
|
|
%ret = shufflevector <2 x i64> %val, <2 x i64> undef,
|
|
<2 x i32> zeroinitializer
|
|
ret <2 x i64> %ret
|
|
}
|
|
|
|
; Test v2i64 splat of the last element.
|
|
define <2 x i64> @f11(i64 %scalar) {
|
|
; CHECK-LABEL: f11:
|
|
; CHECK: vlvgp %v24, %r2, %r2
|
|
; CHECK: br %r14
|
|
%val = insertelement <2 x i64> undef, i64 %scalar, i32 1
|
|
%ret = shufflevector <2 x i64> %val, <2 x i64> undef,
|
|
<2 x i32> <i32 1, i32 1>
|
|
ret <2 x i64> %ret
|
|
}
|
|
|
|
; Test v4f32 splat of the first element.
|
|
define <4 x float> @f12(float %scalar) {
|
|
; CHECK-LABEL: f12:
|
|
; CHECK: vrepf %v24, %v0, 0
|
|
; CHECK: br %r14
|
|
%val = insertelement <4 x float> undef, float %scalar, i32 0
|
|
%ret = shufflevector <4 x float> %val, <4 x float> undef,
|
|
<4 x i32> zeroinitializer
|
|
ret <4 x float> %ret
|
|
}
|
|
|
|
; Test v4f32 splat of the last element.
|
|
define <4 x float> @f13(float %scalar) {
|
|
; CHECK-LABEL: f13:
|
|
; CHECK: vrepf %v24, %v0, 0
|
|
; CHECK: br %r14
|
|
%val = insertelement <4 x float> undef, float %scalar, i32 3
|
|
%ret = shufflevector <4 x float> %val, <4 x float> undef,
|
|
<4 x i32> <i32 3, i32 3, i32 3, i32 3>
|
|
ret <4 x float> %ret
|
|
}
|
|
|
|
; Test v4f32 splat of an arbitrary element, using the second operand of
|
|
; the shufflevector.
|
|
define <4 x float> @f14(float %scalar) {
|
|
; CHECK-LABEL: f14:
|
|
; CHECK: vrepf %v24, %v0, 0
|
|
; CHECK: br %r14
|
|
%val = insertelement <4 x float> undef, float %scalar, i32 1
|
|
%ret = shufflevector <4 x float> undef, <4 x float> %val,
|
|
<4 x i32> <i32 5, i32 5, i32 5, i32 5>
|
|
ret <4 x float> %ret
|
|
}
|
|
|
|
; Test v2f64 splat of the first element.
|
|
define <2 x double> @f15(double %scalar) {
|
|
; CHECK-LABEL: f15:
|
|
; CHECK: vrepg %v24, %v0, 0
|
|
; CHECK: br %r14
|
|
%val = insertelement <2 x double> undef, double %scalar, i32 0
|
|
%ret = shufflevector <2 x double> %val, <2 x double> undef,
|
|
<2 x i32> zeroinitializer
|
|
ret <2 x double> %ret
|
|
}
|
|
|
|
; Test v2f64 splat of the last element.
|
|
define <2 x double> @f16(double %scalar) {
|
|
; CHECK-LABEL: f16:
|
|
; CHECK: vrepg %v24, %v0, 0
|
|
; CHECK: br %r14
|
|
%val = insertelement <2 x double> undef, double %scalar, i32 1
|
|
%ret = shufflevector <2 x double> %val, <2 x double> undef,
|
|
<2 x i32> <i32 1, i32 1>
|
|
ret <2 x double> %ret
|
|
}
|