mirror of
https://github.com/RPCS3/llvm-mirror.git
synced 2025-01-31 20:51:52 +01:00
18451cc4a4
The motivation is that the update script has at least two deviations (`<...>@GOT`/`<...>@PLT`/ and not hiding pointer arithmetics) from what pretty much all the checklines were generated with, and most of the tests are still not updated, so each time one of the non-up-to-date tests is updated to see the effect of the code change, there is a lot of noise. Instead of having to deal with that each time, let's just deal with everything at once. This has been done via: ``` cd llvm-project/llvm/test/CodeGen/X86 grep -rl "; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py" | xargs -L1 <...>/llvm-project/llvm/utils/update_llc_test_checks.py --llc-binary <...>/llvm-project/build/bin/llc ``` Not all tests were regenerated, however.
436 lines
17 KiB
LLVM
436 lines
17 KiB
LLVM
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f | FileCheck %s --check-prefix=ALL
|
|
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512bw | FileCheck %s --check-prefix=ALL
|
|
|
|
define <16 x i32> @_inreg16xi32(i32 %a) {
|
|
; ALL-LABEL: _inreg16xi32:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vpbroadcastd %edi, %zmm0
|
|
; ALL-NEXT: retq
|
|
%b = insertelement <16 x i32> undef, i32 %a, i32 0
|
|
%c = shufflevector <16 x i32> %b, <16 x i32> undef, <16 x i32> zeroinitializer
|
|
ret <16 x i32> %c
|
|
}
|
|
|
|
define <8 x i64> @_inreg8xi64(i64 %a) {
|
|
; ALL-LABEL: _inreg8xi64:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vpbroadcastq %rdi, %zmm0
|
|
; ALL-NEXT: retq
|
|
%b = insertelement <8 x i64> undef, i64 %a, i32 0
|
|
%c = shufflevector <8 x i64> %b, <8 x i64> undef, <8 x i32> zeroinitializer
|
|
ret <8 x i64> %c
|
|
}
|
|
|
|
define <16 x float> @_ss16xfloat_v4(<4 x float> %a) {
|
|
; ALL-LABEL: _ss16xfloat_v4:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastss %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
%b = shufflevector <4 x float> %a, <4 x float> undef, <16 x i32> zeroinitializer
|
|
ret <16 x float> %b
|
|
}
|
|
|
|
define <16 x float> @_inreg16xfloat(float %a) {
|
|
; ALL-LABEL: _inreg16xfloat:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastss %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
%b = insertelement <16 x float> undef, float %a, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
|
|
ret <16 x float> %c
|
|
}
|
|
|
|
define <16 x float> @_ss16xfloat_mask(float %a, <16 x float> %i, <16 x i32> %mask1) {
|
|
; ALL-LABEL: _ss16xfloat_mask:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vptestmd %zmm2, %zmm2, %k1
|
|
; ALL-NEXT: vbroadcastss %xmm0, %zmm1 {%k1}
|
|
; ALL-NEXT: vmovaps %zmm1, %zmm0
|
|
; ALL-NEXT: retq
|
|
%mask = icmp ne <16 x i32> %mask1, zeroinitializer
|
|
%b = insertelement <16 x float> undef, float %a, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
|
|
%r = select <16 x i1> %mask, <16 x float> %c, <16 x float> %i
|
|
ret <16 x float> %r
|
|
}
|
|
|
|
define <16 x float> @_ss16xfloat_maskz(float %a, <16 x i32> %mask1) {
|
|
; ALL-LABEL: _ss16xfloat_maskz:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vptestmd %zmm1, %zmm1, %k1
|
|
; ALL-NEXT: vbroadcastss %xmm0, %zmm0 {%k1} {z}
|
|
; ALL-NEXT: retq
|
|
%mask = icmp ne <16 x i32> %mask1, zeroinitializer
|
|
%b = insertelement <16 x float> undef, float %a, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
|
|
%r = select <16 x i1> %mask, <16 x float> %c, <16 x float> zeroinitializer
|
|
ret <16 x float> %r
|
|
}
|
|
|
|
define <16 x float> @_ss16xfloat_load(float* %a.ptr) {
|
|
; ALL-LABEL: _ss16xfloat_load:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastss (%rdi), %zmm0
|
|
; ALL-NEXT: retq
|
|
%a = load float, float* %a.ptr
|
|
%b = insertelement <16 x float> undef, float %a, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
|
|
ret <16 x float> %c
|
|
}
|
|
|
|
define <16 x float> @_ss16xfloat_mask_load(float* %a.ptr, <16 x float> %i, <16 x i32> %mask1) {
|
|
; ALL-LABEL: _ss16xfloat_mask_load:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vptestmd %zmm1, %zmm1, %k1
|
|
; ALL-NEXT: vbroadcastss (%rdi), %zmm0 {%k1}
|
|
; ALL-NEXT: retq
|
|
%a = load float, float* %a.ptr
|
|
%mask = icmp ne <16 x i32> %mask1, zeroinitializer
|
|
%b = insertelement <16 x float> undef, float %a, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
|
|
%r = select <16 x i1> %mask, <16 x float> %c, <16 x float> %i
|
|
ret <16 x float> %r
|
|
}
|
|
|
|
define <16 x float> @_ss16xfloat_maskz_load(float* %a.ptr, <16 x i32> %mask1) {
|
|
; ALL-LABEL: _ss16xfloat_maskz_load:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vptestmd %zmm0, %zmm0, %k1
|
|
; ALL-NEXT: vbroadcastss (%rdi), %zmm0 {%k1} {z}
|
|
; ALL-NEXT: retq
|
|
%a = load float, float* %a.ptr
|
|
%mask = icmp ne <16 x i32> %mask1, zeroinitializer
|
|
%b = insertelement <16 x float> undef, float %a, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
|
|
%r = select <16 x i1> %mask, <16 x float> %c, <16 x float> zeroinitializer
|
|
ret <16 x float> %r
|
|
}
|
|
|
|
define <8 x double> @_inreg8xdouble(double %a) {
|
|
; ALL-LABEL: _inreg8xdouble:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastsd %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
%b = insertelement <8 x double> undef, double %a, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
|
|
ret <8 x double> %c
|
|
}
|
|
|
|
define <8 x double> @_sd8xdouble_mask(double %a, <8 x double> %i, <8 x i32> %mask1) {
|
|
; ALL-LABEL: _sd8xdouble_mask:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: # kill: def $ymm2 killed $ymm2 def $zmm2
|
|
; ALL-NEXT: vptestmd %zmm2, %zmm2, %k1
|
|
; ALL-NEXT: vbroadcastsd %xmm0, %zmm1 {%k1}
|
|
; ALL-NEXT: vmovapd %zmm1, %zmm0
|
|
; ALL-NEXT: retq
|
|
%mask = icmp ne <8 x i32> %mask1, zeroinitializer
|
|
%b = insertelement <8 x double> undef, double %a, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
|
|
%r = select <8 x i1> %mask, <8 x double> %c, <8 x double> %i
|
|
ret <8 x double> %r
|
|
}
|
|
|
|
define <8 x double> @_sd8xdouble_maskz(double %a, <8 x i32> %mask1) {
|
|
; ALL-LABEL: _sd8xdouble_maskz:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: # kill: def $ymm1 killed $ymm1 def $zmm1
|
|
; ALL-NEXT: vptestmd %zmm1, %zmm1, %k1
|
|
; ALL-NEXT: vbroadcastsd %xmm0, %zmm0 {%k1} {z}
|
|
; ALL-NEXT: retq
|
|
%mask = icmp ne <8 x i32> %mask1, zeroinitializer
|
|
%b = insertelement <8 x double> undef, double %a, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
|
|
%r = select <8 x i1> %mask, <8 x double> %c, <8 x double> zeroinitializer
|
|
ret <8 x double> %r
|
|
}
|
|
|
|
define <8 x double> @_sd8xdouble_load(double* %a.ptr) {
|
|
; ALL-LABEL: _sd8xdouble_load:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastsd (%rdi), %zmm0
|
|
; ALL-NEXT: retq
|
|
%a = load double, double* %a.ptr
|
|
%b = insertelement <8 x double> undef, double %a, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
|
|
ret <8 x double> %c
|
|
}
|
|
|
|
define <8 x double> @_sd8xdouble_mask_load(double* %a.ptr, <8 x double> %i, <8 x i32> %mask1) {
|
|
; ALL-LABEL: _sd8xdouble_mask_load:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: # kill: def $ymm1 killed $ymm1 def $zmm1
|
|
; ALL-NEXT: vptestmd %zmm1, %zmm1, %k1
|
|
; ALL-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1}
|
|
; ALL-NEXT: retq
|
|
%a = load double, double* %a.ptr
|
|
%mask = icmp ne <8 x i32> %mask1, zeroinitializer
|
|
%b = insertelement <8 x double> undef, double %a, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
|
|
%r = select <8 x i1> %mask, <8 x double> %c, <8 x double> %i
|
|
ret <8 x double> %r
|
|
}
|
|
|
|
define <8 x double> @_sd8xdouble_maskz_load(double* %a.ptr, <8 x i32> %mask1) {
|
|
; ALL-LABEL: _sd8xdouble_maskz_load:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0
|
|
; ALL-NEXT: vptestmd %zmm0, %zmm0, %k1
|
|
; ALL-NEXT: vbroadcastsd (%rdi), %zmm0 {%k1} {z}
|
|
; ALL-NEXT: retq
|
|
%a = load double, double* %a.ptr
|
|
%mask = icmp ne <8 x i32> %mask1, zeroinitializer
|
|
%b = insertelement <8 x double> undef, double %a, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
|
|
%r = select <8 x i1> %mask, <8 x double> %c, <8 x double> zeroinitializer
|
|
ret <8 x double> %r
|
|
}
|
|
|
|
define <16 x i32> @_xmm16xi32(<16 x i32> %a) {
|
|
; ALL-LABEL: _xmm16xi32:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastss %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
%b = shufflevector <16 x i32> %a, <16 x i32> undef, <16 x i32> zeroinitializer
|
|
ret <16 x i32> %b
|
|
}
|
|
|
|
define <16 x float> @_xmm16xfloat(<16 x float> %a) {
|
|
; ALL-LABEL: _xmm16xfloat:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastss %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
%b = shufflevector <16 x float> %a, <16 x float> undef, <16 x i32> zeroinitializer
|
|
ret <16 x float> %b
|
|
}
|
|
|
|
define <16 x i32> @test_vbroadcast(<16 x float> %a0) {
|
|
; ALL-LABEL: test_vbroadcast:
|
|
; ALL: # %bb.0: # %entry
|
|
; ALL-NEXT: vxorps %xmm1, %xmm1, %xmm1
|
|
; ALL-NEXT: vcmpunordps %zmm1, %zmm0, %k1
|
|
; ALL-NEXT: vpternlogd $255, %zmm0, %zmm0, %zmm0 {%k1} {z}
|
|
; ALL-NEXT: knotw %k1, %k1
|
|
; ALL-NEXT: vmovdqa32 %zmm0, %zmm0 {%k1} {z}
|
|
; ALL-NEXT: retq
|
|
entry:
|
|
%0 = sext <16 x i1> zeroinitializer to <16 x i32>
|
|
%1 = fcmp uno <16 x float> %a0, zeroinitializer
|
|
%2 = sext <16 x i1> %1 to <16 x i32>
|
|
%3 = select <16 x i1> %1, <16 x i32> %0, <16 x i32> %2
|
|
ret <16 x i32> %3
|
|
}
|
|
|
|
; We implement the set1 intrinsics with vector initializers. Verify that the
|
|
; IR generated will produce broadcasts at the end.
|
|
define <8 x double> @test_set1_pd(double %d) #2 {
|
|
; ALL-LABEL: test_set1_pd:
|
|
; ALL: # %bb.0: # %entry
|
|
; ALL-NEXT: vbroadcastsd %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
entry:
|
|
%vecinit.i = insertelement <8 x double> undef, double %d, i32 0
|
|
%vecinit1.i = insertelement <8 x double> %vecinit.i, double %d, i32 1
|
|
%vecinit2.i = insertelement <8 x double> %vecinit1.i, double %d, i32 2
|
|
%vecinit3.i = insertelement <8 x double> %vecinit2.i, double %d, i32 3
|
|
%vecinit4.i = insertelement <8 x double> %vecinit3.i, double %d, i32 4
|
|
%vecinit5.i = insertelement <8 x double> %vecinit4.i, double %d, i32 5
|
|
%vecinit6.i = insertelement <8 x double> %vecinit5.i, double %d, i32 6
|
|
%vecinit7.i = insertelement <8 x double> %vecinit6.i, double %d, i32 7
|
|
ret <8 x double> %vecinit7.i
|
|
}
|
|
|
|
define <8 x i64> @test_set1_epi64(i64 %d) #2 {
|
|
; ALL-LABEL: test_set1_epi64:
|
|
; ALL: # %bb.0: # %entry
|
|
; ALL-NEXT: vpbroadcastq %rdi, %zmm0
|
|
; ALL-NEXT: retq
|
|
entry:
|
|
%vecinit.i = insertelement <8 x i64> undef, i64 %d, i32 0
|
|
%vecinit1.i = insertelement <8 x i64> %vecinit.i, i64 %d, i32 1
|
|
%vecinit2.i = insertelement <8 x i64> %vecinit1.i, i64 %d, i32 2
|
|
%vecinit3.i = insertelement <8 x i64> %vecinit2.i, i64 %d, i32 3
|
|
%vecinit4.i = insertelement <8 x i64> %vecinit3.i, i64 %d, i32 4
|
|
%vecinit5.i = insertelement <8 x i64> %vecinit4.i, i64 %d, i32 5
|
|
%vecinit6.i = insertelement <8 x i64> %vecinit5.i, i64 %d, i32 6
|
|
%vecinit7.i = insertelement <8 x i64> %vecinit6.i, i64 %d, i32 7
|
|
ret <8 x i64> %vecinit7.i
|
|
}
|
|
|
|
define <16 x float> @test_set1_ps(float %f) #2 {
|
|
; ALL-LABEL: test_set1_ps:
|
|
; ALL: # %bb.0: # %entry
|
|
; ALL-NEXT: vbroadcastss %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
entry:
|
|
%vecinit.i = insertelement <16 x float> undef, float %f, i32 0
|
|
%vecinit1.i = insertelement <16 x float> %vecinit.i, float %f, i32 1
|
|
%vecinit2.i = insertelement <16 x float> %vecinit1.i, float %f, i32 2
|
|
%vecinit3.i = insertelement <16 x float> %vecinit2.i, float %f, i32 3
|
|
%vecinit4.i = insertelement <16 x float> %vecinit3.i, float %f, i32 4
|
|
%vecinit5.i = insertelement <16 x float> %vecinit4.i, float %f, i32 5
|
|
%vecinit6.i = insertelement <16 x float> %vecinit5.i, float %f, i32 6
|
|
%vecinit7.i = insertelement <16 x float> %vecinit6.i, float %f, i32 7
|
|
%vecinit8.i = insertelement <16 x float> %vecinit7.i, float %f, i32 8
|
|
%vecinit9.i = insertelement <16 x float> %vecinit8.i, float %f, i32 9
|
|
%vecinit10.i = insertelement <16 x float> %vecinit9.i, float %f, i32 10
|
|
%vecinit11.i = insertelement <16 x float> %vecinit10.i, float %f, i32 11
|
|
%vecinit12.i = insertelement <16 x float> %vecinit11.i, float %f, i32 12
|
|
%vecinit13.i = insertelement <16 x float> %vecinit12.i, float %f, i32 13
|
|
%vecinit14.i = insertelement <16 x float> %vecinit13.i, float %f, i32 14
|
|
%vecinit15.i = insertelement <16 x float> %vecinit14.i, float %f, i32 15
|
|
ret <16 x float> %vecinit15.i
|
|
}
|
|
|
|
define <16 x i32> @test_set1_epi32(i32 %f) #2 {
|
|
; ALL-LABEL: test_set1_epi32:
|
|
; ALL: # %bb.0: # %entry
|
|
; ALL-NEXT: vpbroadcastd %edi, %zmm0
|
|
; ALL-NEXT: retq
|
|
entry:
|
|
%vecinit.i = insertelement <16 x i32> undef, i32 %f, i32 0
|
|
%vecinit1.i = insertelement <16 x i32> %vecinit.i, i32 %f, i32 1
|
|
%vecinit2.i = insertelement <16 x i32> %vecinit1.i, i32 %f, i32 2
|
|
%vecinit3.i = insertelement <16 x i32> %vecinit2.i, i32 %f, i32 3
|
|
%vecinit4.i = insertelement <16 x i32> %vecinit3.i, i32 %f, i32 4
|
|
%vecinit5.i = insertelement <16 x i32> %vecinit4.i, i32 %f, i32 5
|
|
%vecinit6.i = insertelement <16 x i32> %vecinit5.i, i32 %f, i32 6
|
|
%vecinit7.i = insertelement <16 x i32> %vecinit6.i, i32 %f, i32 7
|
|
%vecinit8.i = insertelement <16 x i32> %vecinit7.i, i32 %f, i32 8
|
|
%vecinit9.i = insertelement <16 x i32> %vecinit8.i, i32 %f, i32 9
|
|
%vecinit10.i = insertelement <16 x i32> %vecinit9.i, i32 %f, i32 10
|
|
%vecinit11.i = insertelement <16 x i32> %vecinit10.i, i32 %f, i32 11
|
|
%vecinit12.i = insertelement <16 x i32> %vecinit11.i, i32 %f, i32 12
|
|
%vecinit13.i = insertelement <16 x i32> %vecinit12.i, i32 %f, i32 13
|
|
%vecinit14.i = insertelement <16 x i32> %vecinit13.i, i32 %f, i32 14
|
|
%vecinit15.i = insertelement <16 x i32> %vecinit14.i, i32 %f, i32 15
|
|
ret <16 x i32> %vecinit15.i
|
|
}
|
|
|
|
; We implement the scalar broadcast intrinsics with vector initializers.
|
|
; Verify that the IR generated will produce the broadcast at the end.
|
|
define <8 x double> @test_mm512_broadcastsd_pd(<2 x double> %a) {
|
|
; ALL-LABEL: test_mm512_broadcastsd_pd:
|
|
; ALL: # %bb.0: # %entry
|
|
; ALL-NEXT: vbroadcastsd %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
entry:
|
|
%0 = extractelement <2 x double> %a, i32 0
|
|
%vecinit.i = insertelement <8 x double> undef, double %0, i32 0
|
|
%vecinit1.i = insertelement <8 x double> %vecinit.i, double %0, i32 1
|
|
%vecinit2.i = insertelement <8 x double> %vecinit1.i, double %0, i32 2
|
|
%vecinit3.i = insertelement <8 x double> %vecinit2.i, double %0, i32 3
|
|
%vecinit4.i = insertelement <8 x double> %vecinit3.i, double %0, i32 4
|
|
%vecinit5.i = insertelement <8 x double> %vecinit4.i, double %0, i32 5
|
|
%vecinit6.i = insertelement <8 x double> %vecinit5.i, double %0, i32 6
|
|
%vecinit7.i = insertelement <8 x double> %vecinit6.i, double %0, i32 7
|
|
ret <8 x double> %vecinit7.i
|
|
}
|
|
|
|
define <16 x float> @test1(<8 x float>%a) {
|
|
; ALL-LABEL: test1:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastss %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
%res = shufflevector <8 x float> %a, <8 x float> undef, <16 x i32> zeroinitializer
|
|
ret <16 x float>%res
|
|
}
|
|
|
|
define <8 x double> @test2(<4 x double>%a) {
|
|
; ALL-LABEL: test2:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastsd %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
%res = shufflevector <4 x double> %a, <4 x double> undef, <8 x i32> zeroinitializer
|
|
ret <8 x double>%res
|
|
}
|
|
|
|
define <64 x i8> @_invec32xi8(<32 x i8>%a) {
|
|
; AVX512F-LABEL: _invec32xi8:
|
|
; AVX512F: # %bb.0:
|
|
; AVX512F-NEXT: vpbroadcastb %xmm0, %ymm0
|
|
; AVX512F-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm0
|
|
; AVX512F-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: _invec32xi8:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vpbroadcastb %xmm0, %zmm0
|
|
; AVX512BW-NEXT: retq
|
|
%res = shufflevector <32 x i8> %a, <32 x i8> undef, <64 x i32> zeroinitializer
|
|
ret <64 x i8>%res
|
|
}
|
|
|
|
define <32 x i16> @_invec16xi16(<16 x i16>%a) {
|
|
; AVX512F-LABEL: _invec16xi16:
|
|
; AVX512F: # %bb.0:
|
|
; AVX512F-NEXT: vpbroadcastw %xmm0, %ymm0
|
|
; AVX512F-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm0
|
|
; AVX512F-NEXT: retq
|
|
;
|
|
; AVX512BW-LABEL: _invec16xi16:
|
|
; AVX512BW: # %bb.0:
|
|
; AVX512BW-NEXT: vpbroadcastw %xmm0, %zmm0
|
|
; AVX512BW-NEXT: retq
|
|
%res = shufflevector <16 x i16> %a, <16 x i16> undef, <32 x i32> zeroinitializer
|
|
ret <32 x i16>%res
|
|
}
|
|
|
|
define <16 x i32> @_invec8xi32(<8 x i32>%a) {
|
|
; ALL-LABEL: _invec8xi32:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastss %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
%res = shufflevector <8 x i32> %a, <8 x i32> undef, <16 x i32> zeroinitializer
|
|
ret <16 x i32>%res
|
|
}
|
|
|
|
define <8 x i64> @_invec4xi64(<4 x i64>%a) {
|
|
; ALL-LABEL: _invec4xi64:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: vbroadcastsd %xmm0, %zmm0
|
|
; ALL-NEXT: retq
|
|
%res = shufflevector <4 x i64> %a, <4 x i64> undef, <8 x i32> zeroinitializer
|
|
ret <8 x i64>%res
|
|
}
|
|
|
|
declare void @func_f32(float)
|
|
define <16 x float> @broadcast_ss_spill(float %x) {
|
|
; ALL-LABEL: broadcast_ss_spill:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: subq $24, %rsp
|
|
; ALL-NEXT: .cfi_def_cfa_offset 32
|
|
; ALL-NEXT: vaddss %xmm0, %xmm0, %xmm0
|
|
; ALL-NEXT: vmovaps %xmm0, (%rsp) # 16-byte Spill
|
|
; ALL-NEXT: callq func_f32@PLT
|
|
; ALL-NEXT: vbroadcastss (%rsp), %zmm0 # 16-byte Folded Reload
|
|
; ALL-NEXT: addq $24, %rsp
|
|
; ALL-NEXT: .cfi_def_cfa_offset 8
|
|
; ALL-NEXT: retq
|
|
%a = fadd float %x, %x
|
|
call void @func_f32(float %a)
|
|
%b = insertelement <16 x float> undef, float %a, i32 0
|
|
%c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
|
|
ret <16 x float> %c
|
|
}
|
|
|
|
declare void @func_f64(double)
|
|
define <8 x double> @broadcast_sd_spill(double %x) {
|
|
; ALL-LABEL: broadcast_sd_spill:
|
|
; ALL: # %bb.0:
|
|
; ALL-NEXT: subq $24, %rsp
|
|
; ALL-NEXT: .cfi_def_cfa_offset 32
|
|
; ALL-NEXT: vaddsd %xmm0, %xmm0, %xmm0
|
|
; ALL-NEXT: vmovapd %xmm0, (%rsp) # 16-byte Spill
|
|
; ALL-NEXT: callq func_f64@PLT
|
|
; ALL-NEXT: vbroadcastsd (%rsp), %zmm0 # 16-byte Folded Reload
|
|
; ALL-NEXT: addq $24, %rsp
|
|
; ALL-NEXT: .cfi_def_cfa_offset 8
|
|
; ALL-NEXT: retq
|
|
%a = fadd double %x, %x
|
|
call void @func_f64(double %a)
|
|
%b = insertelement <8 x double> undef, double %a, i32 0
|
|
%c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
|
|
ret <8 x double> %c
|
|
}
|