+; CHECK-LABEL: test_vaddpd_256
+; CHECK: vaddpd{{.*}}
+; CHECK: ret
+define <4 x double> @test_vaddpd_256(<4 x double> %y, <4 x double> %x) {
+entry:
+ %add.i = fadd <4 x double> %x, %y
+ ret <4 x double> %add.i
+}
+
+; CHECK-LABEL: test_fold_vaddpd_256
+; CHECK: vaddpd LCP{{.*}}(%rip){{.*}}
+; CHECK: ret
+define <4 x double> @test_fold_vaddpd_256(<4 x double> %y) {
+entry:
+ %add.i = fadd <4 x double> %y, <double 4.500000e+00, double 3.400000e+00, double 4.500000e+00, double 5.600000e+00>
+ ret <4 x double> %add.i
+}
+
+; CHECK-LABEL: test_broadcast_vaddpd_256
+; CHECK: LCP{{.*}}(%rip){1to8}, %ymm0, %ymm0
+; CHECK: ret
+define <8 x float> @test_broadcast_vaddpd_256(<8 x float> %a) nounwind {
+ %b = fadd <8 x float> %a, <float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000>
+ ret <8 x float> %b
+}
+
+; CHECK-LABEL: test_mask_vaddps_256
+; CHECK: vaddps {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <8 x float> @test_mask_vaddps_256(<8 x float> %dst, <8 x float> %i,
+ <8 x float> %j, <8 x i32> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <8 x i32> %mask1, zeroinitializer
+ %x = fadd <8 x float> %i, %j
+ %r = select <8 x i1> %mask, <8 x float> %x, <8 x float> %dst
+ ret <8 x float> %r
+}
+
+; CHECK-LABEL: test_mask_vmulps_256
+; CHECK: vmulps {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <8 x float> @test_mask_vmulps_256(<8 x float> %dst, <8 x float> %i,
+ <8 x float> %j, <8 x i32> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <8 x i32> %mask1, zeroinitializer
+ %x = fmul <8 x float> %i, %j
+ %r = select <8 x i1> %mask, <8 x float> %x, <8 x float> %dst
+ ret <8 x float> %r
+}
+
+; CHECK-LABEL: test_mask_vminps_256
+; CHECK: vminps {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <8 x float> @test_mask_vminps_256(<8 x float> %dst, <8 x float> %i,
+ <8 x float> %j, <8 x i32> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <8 x i32> %mask1, zeroinitializer
+ %cmp_res = fcmp olt <8 x float> %i, %j
+ %min = select <8 x i1> %cmp_res, <8 x float> %i, <8 x float> %j
+ %r = select <8 x i1> %mask, <8 x float> %min, <8 x float> %dst
+ ret <8 x float> %r
+}
+
+; CHECK-LABEL: test_mask_vmaxps_256
+; CHECK: vmaxps {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <8 x float> @test_mask_vmaxps_256(<8 x float> %dst, <8 x float> %i,
+ <8 x float> %j, <8 x i32> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <8 x i32> %mask1, zeroinitializer
+ %cmp_res = fcmp ogt <8 x float> %i, %j
+ %max = select <8 x i1> %cmp_res, <8 x float> %i, <8 x float> %j
+ %r = select <8 x i1> %mask, <8 x float> %max, <8 x float> %dst
+ ret <8 x float> %r
+}
+
+; CHECK-LABEL: test_mask_vsubps_256
+; CHECK: vsubps {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <8 x float> @test_mask_vsubps_256(<8 x float> %dst, <8 x float> %i,
+ <8 x float> %j, <8 x i32> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <8 x i32> %mask1, zeroinitializer
+ %x = fsub <8 x float> %i, %j
+ %r = select <8 x i1> %mask, <8 x float> %x, <8 x float> %dst
+ ret <8 x float> %r
+}
+
+; CHECK-LABEL: test_mask_vdivps_256
+; CHECK: vdivps {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <8 x float> @test_mask_vdivps_256(<8 x float> %dst, <8 x float> %i,
+ <8 x float> %j, <8 x i32> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <8 x i32> %mask1, zeroinitializer
+ %x = fdiv <8 x float> %i, %j
+ %r = select <8 x i1> %mask, <8 x float> %x, <8 x float> %dst
+ ret <8 x float> %r
+}
+
+; CHECK-LABEL: test_mask_vmulpd_256
+; CHECK: vmulpd {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <4 x double> @test_mask_vmulpd_256(<4 x double> %dst, <4 x double> %i,
+ <4 x double> %j, <4 x i64> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %x = fmul <4 x double> %i, %j
+ %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %dst
+ ret <4 x double> %r
+}
+
+; CHECK-LABEL: test_mask_vminpd_256
+; CHECK: vminpd {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <4 x double> @test_mask_vminpd_256(<4 x double> %dst, <4 x double> %i,
+ <4 x double> %j, <4 x i64> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %cmp_res = fcmp olt <4 x double> %i, %j
+ %min = select <4 x i1> %cmp_res, <4 x double> %i, <4 x double> %j
+ %r = select <4 x i1> %mask, <4 x double> %min, <4 x double> %dst
+ ret <4 x double> %r
+}
+
+; CHECK-LABEL: test_mask_vmaxpd_256
+; CHECK: vmaxpd {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <4 x double> @test_mask_vmaxpd_256(<4 x double> %dst, <4 x double> %i,
+ <4 x double> %j, <4 x i64> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %cmp_res = fcmp ogt <4 x double> %i, %j
+ %max = select <4 x i1> %cmp_res, <4 x double> %i, <4 x double> %j
+ %r = select <4 x i1> %mask, <4 x double> %max, <4 x double> %dst
+ ret <4 x double> %r
+}
+
+; CHECK-LABEL: test_mask_vsubpd_256
+; CHECK: vsubpd {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <4 x double> @test_mask_vsubpd_256(<4 x double> %dst, <4 x double> %i,
+ <4 x double> %j, <4 x i64> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %x = fsub <4 x double> %i, %j
+ %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %dst
+ ret <4 x double> %r
+}
+
+; CHECK-LABEL: test_mask_vdivpd_256
+; CHECK: vdivpd {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <4 x double> @test_mask_vdivpd_256(<4 x double> %dst, <4 x double> %i,
+ <4 x double> %j, <4 x i64> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %x = fdiv <4 x double> %i, %j
+ %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %dst
+ ret <4 x double> %r
+}
+
+; CHECK-LABEL: test_mask_vaddpd_256
+; CHECK: vaddpd {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}}}
+; CHECK: ret
+define <4 x double> @test_mask_vaddpd_256(<4 x double> %dst, <4 x double> %i,
+ <4 x double> %j, <4 x i64> %mask1)
+ nounwind readnone {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %x = fadd <4 x double> %i, %j
+ %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %dst
+ ret <4 x double> %r
+}
+
+; CHECK-LABEL: test_maskz_vaddpd_256
+; CHECK: vaddpd {{%ymm[0-9]{1,2}, %ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]} {z}}}
+; CHECK: ret
+define <4 x double> @test_maskz_vaddpd_256(<4 x double> %i, <4 x double> %j,
+ <4 x i64> %mask1) nounwind readnone {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %x = fadd <4 x double> %i, %j
+ %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> zeroinitializer
+ ret <4 x double> %r
+}
+
+; CHECK-LABEL: test_mask_fold_vaddpd_256
+; CHECK: vaddpd (%rdi), {{.*%ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]}.*}}
+; CHECK: ret
+define <4 x double> @test_mask_fold_vaddpd_256(<4 x double> %dst, <4 x double> %i,
+ <4 x double>* %j, <4 x i64> %mask1)
+ nounwind {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %tmp = load <4 x double>, <4 x double>* %j
+ %x = fadd <4 x double> %i, %tmp
+ %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %dst
+ ret <4 x double> %r
+}
+
+; CHECK-LABEL: test_maskz_fold_vaddpd_256
+; CHECK: vaddpd (%rdi), {{.*%ymm[0-9]{1,2}, %ymm[0-9]{1,2} {%k[1-7]} {z}.*}}
+; CHECK: ret
+define <4 x double> @test_maskz_fold_vaddpd_256(<4 x double> %i, <4 x double>* %j,
+ <4 x i64> %mask1) nounwind {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %tmp = load <4 x double>, <4 x double>* %j
+ %x = fadd <4 x double> %i, %tmp
+ %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> zeroinitializer
+ ret <4 x double> %r
+}
+
+; CHECK-LABEL: test_broadcast2_vaddpd_256
+; CHECK: vaddpd (%rdi){1to4}, %ymm{{.*}}
+; CHECK: ret
+define <4 x double> @test_broadcast2_vaddpd_256(<4 x double> %i, double* %j) nounwind {
+ %tmp = load double, double* %j
+ %b = insertelement <4 x double> undef, double %tmp, i32 0
+ %c = shufflevector <4 x double> %b, <4 x double> undef,
+ <4 x i32> zeroinitializer
+ %x = fadd <4 x double> %c, %i
+ ret <4 x double> %x
+}
+
+; CHECK-LABEL: test_mask_broadcast_vaddpd_256
+; CHECK: vaddpd (%rdi){1to4}, %ymm{{.*{%k[1-7]}.*}}
+; CHECK: ret
+define <4 x double> @test_mask_broadcast_vaddpd_256(<4 x double> %dst, <4 x double> %i,
+ double* %j, <4 x i64> %mask1) nounwind {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %tmp = load double, double* %j
+ %b = insertelement <4 x double> undef, double %tmp, i32 0
+ %c = shufflevector <4 x double> %b, <4 x double> undef,
+ <4 x i32> zeroinitializer
+ %x = fadd <4 x double> %c, %i
+ %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %i
+ ret <4 x double> %r
+}
+
+; CHECK-LABEL: test_maskz_broadcast_vaddpd_256
+; CHECK: vaddpd (%rdi){1to4}, %ymm{{.*{%k[1-7]} {z}.*}}
+; CHECK: ret
+define <4 x double> @test_maskz_broadcast_vaddpd_256(<4 x double> %i, double* %j,
+ <4 x i64> %mask1) nounwind {
+ %mask = icmp ne <4 x i64> %mask1, zeroinitializer
+ %tmp = load double, double* %j
+ %b = insertelement <4 x double> undef, double %tmp, i32 0
+ %c = shufflevector <4 x double> %b, <4 x double> undef,
+ <4 x i32> zeroinitializer
+ %x = fadd <4 x double> %c, %i
+ %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> zeroinitializer
+ ret <4 x double> %r
+}
+