1; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tahiti < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s 2; RUN: llc -amdgpu-scalarize-global-loads=false -march=amdgcn -mcpu=tonga < %s | FileCheck -check-prefix=SI -check-prefix=FUNC %s 3; RUN: llc -amdgpu-scalarize-global-loads=false -march=r600 -mcpu=redwood < %s 4 5define amdgpu_kernel void @srem_i32(i32 addrspace(1)* %out, i32 addrspace(1)* %in) { 6 %den_ptr = getelementptr i32, i32 addrspace(1)* %in, i32 1 7 %num = load i32, i32 addrspace(1) * %in 8 %den = load i32, i32 addrspace(1) * %den_ptr 9 %result = srem i32 %num, %den 10 store i32 %result, i32 addrspace(1)* %out 11 ret void 12} 13 14define amdgpu_kernel void @srem_i32_4(i32 addrspace(1)* %out, i32 addrspace(1)* %in) { 15 %num = load i32, i32 addrspace(1) * %in 16 %result = srem i32 %num, 4 17 store i32 %result, i32 addrspace(1)* %out 18 ret void 19} 20 21; FUNC-LABEL: {{^}}srem_i32_7: 22; SI: v_mov_b32_e32 [[MAGIC:v[0-9]+]], 0x92492493 23; SI: v_mul_hi_i32 {{v[0-9]+}}, {{v[0-9]+}}, [[MAGIC]] 24; SI: v_mul_lo_i32 25; SI: v_sub_{{[iu]}}32 26; SI: s_endpgm 27define amdgpu_kernel void @srem_i32_7(i32 addrspace(1)* %out, i32 addrspace(1)* %in) { 28 %num = load i32, i32 addrspace(1) * %in 29 %result = srem i32 %num, 7 30 store i32 %result, i32 addrspace(1)* %out 31 ret void 32} 33 34define amdgpu_kernel void @srem_v2i32(<2 x i32> addrspace(1)* %out, <2 x i32> addrspace(1)* %in) { 35 %den_ptr = getelementptr <2 x i32>, <2 x i32> addrspace(1)* %in, i32 1 36 %num = load <2 x i32>, <2 x i32> addrspace(1) * %in 37 %den = load <2 x i32>, <2 x i32> addrspace(1) * %den_ptr 38 %result = srem <2 x i32> %num, %den 39 store <2 x i32> %result, <2 x i32> addrspace(1)* %out 40 ret void 41} 42 43define amdgpu_kernel void @srem_v2i32_4(<2 x i32> addrspace(1)* %out, <2 x i32> addrspace(1)* %in) { 44 %num = load <2 x i32>, <2 x i32> addrspace(1) * %in 45 %result = srem <2 x i32> %num, <i32 4, i32 4> 46 store <2 x i32> %result, <2 x i32> addrspace(1)* %out 47 ret void 48} 49 50define amdgpu_kernel void @srem_v4i32(<4 x i32> addrspace(1)* %out, <4 x i32> addrspace(1)* %in) { 51 %den_ptr = getelementptr <4 x i32>, <4 x i32> addrspace(1)* %in, i32 1 52 %num = load <4 x i32>, <4 x i32> addrspace(1) * %in 53 %den = load <4 x i32>, <4 x i32> addrspace(1) * %den_ptr 54 %result = srem <4 x i32> %num, %den 55 store <4 x i32> %result, <4 x i32> addrspace(1)* %out 56 ret void 57} 58 59define amdgpu_kernel void @srem_v4i32_4(<4 x i32> addrspace(1)* %out, <4 x i32> addrspace(1)* %in) { 60 %num = load <4 x i32>, <4 x i32> addrspace(1) * %in 61 %result = srem <4 x i32> %num, <i32 4, i32 4, i32 4, i32 4> 62 store <4 x i32> %result, <4 x i32> addrspace(1)* %out 63 ret void 64} 65 66define amdgpu_kernel void @srem_i64(i64 addrspace(1)* %out, i64 addrspace(1)* %in) { 67 %den_ptr = getelementptr i64, i64 addrspace(1)* %in, i64 1 68 %num = load i64, i64 addrspace(1) * %in 69 %den = load i64, i64 addrspace(1) * %den_ptr 70 %result = srem i64 %num, %den 71 store i64 %result, i64 addrspace(1)* %out 72 ret void 73} 74 75define amdgpu_kernel void @srem_i64_4(i64 addrspace(1)* %out, i64 addrspace(1)* %in) { 76 %num = load i64, i64 addrspace(1) * %in 77 %result = srem i64 %num, 4 78 store i64 %result, i64 addrspace(1)* %out 79 ret void 80} 81 82define amdgpu_kernel void @srem_v2i64(<2 x i64> addrspace(1)* %out, <2 x i64> addrspace(1)* %in) { 83 %den_ptr = getelementptr <2 x i64>, <2 x i64> addrspace(1)* %in, i64 1 84 %num = load <2 x i64>, <2 x i64> addrspace(1) * %in 85 %den = load <2 x i64>, <2 x i64> addrspace(1) * %den_ptr 86 %result = srem <2 x i64> %num, %den 87 store <2 x i64> %result, <2 x i64> addrspace(1)* %out 88 ret void 89} 90 91define amdgpu_kernel void @srem_v2i64_4(<2 x i64> addrspace(1)* %out, <2 x i64> addrspace(1)* %in) { 92 %num = load <2 x i64>, <2 x i64> addrspace(1) * %in 93 %result = srem <2 x i64> %num, <i64 4, i64 4> 94 store <2 x i64> %result, <2 x i64> addrspace(1)* %out 95 ret void 96} 97 98define amdgpu_kernel void @srem_v4i64(<4 x i64> addrspace(1)* %out, <4 x i64> addrspace(1)* %in) { 99 %den_ptr = getelementptr <4 x i64>, <4 x i64> addrspace(1)* %in, i64 1 100 %num = load <4 x i64>, <4 x i64> addrspace(1) * %in 101 %den = load <4 x i64>, <4 x i64> addrspace(1) * %den_ptr 102 %result = srem <4 x i64> %num, %den 103 store <4 x i64> %result, <4 x i64> addrspace(1)* %out 104 ret void 105} 106 107define amdgpu_kernel void @srem_v4i64_4(<4 x i64> addrspace(1)* %out, <4 x i64> addrspace(1)* %in) { 108 %num = load <4 x i64>, <4 x i64> addrspace(1) * %in 109 %result = srem <4 x i64> %num, <i64 4, i64 4, i64 4, i64 4> 110 store <4 x i64> %result, <4 x i64> addrspace(1)* %out 111 ret void 112} 113