diff options
Diffstat (limited to 'test/CodeGen/R600/indirect-private-64.ll')
-rw-r--r-- | test/CodeGen/R600/indirect-private-64.ll | 24 |
1 files changed, 12 insertions, 12 deletions
diff --git a/test/CodeGen/R600/indirect-private-64.ll b/test/CodeGen/R600/indirect-private-64.ll index cb06d60..d63e1b6 100644 --- a/test/CodeGen/R600/indirect-private-64.ll +++ b/test/CodeGen/R600/indirect-private-64.ll @@ -14,12 +14,12 @@ declare void @llvm.AMDGPU.barrier.local() noduplicate nounwind ; SI-PROMOTE: ds_write_b64 ; SI-PROMOTE: ds_read_b64 define void @private_access_f64_alloca(double addrspace(1)* noalias %out, double addrspace(1)* noalias %in, i32 %b) nounwind { - %val = load double addrspace(1)* %in, align 8 + %val = load double, double addrspace(1)* %in, align 8 %array = alloca double, i32 16, align 8 - %ptr = getelementptr double* %array, i32 %b + %ptr = getelementptr double, double* %array, i32 %b store double %val, double* %ptr, align 8 call void @llvm.AMDGPU.barrier.local() noduplicate nounwind - %result = load double* %ptr, align 8 + %result = load double, double* %ptr, align 8 store double %result, double addrspace(1)* %out, align 8 ret void } @@ -38,12 +38,12 @@ define void @private_access_f64_alloca(double addrspace(1)* noalias %out, double ; SI-PROMOTE: ds_read_b32 ; SI-PROMOTE: ds_read_b32 define void @private_access_v2f64_alloca(<2 x double> addrspace(1)* noalias %out, <2 x double> addrspace(1)* noalias %in, i32 %b) nounwind { - %val = load <2 x double> addrspace(1)* %in, align 16 + %val = load <2 x double>, <2 x double> addrspace(1)* %in, align 16 %array = alloca <2 x double>, i32 16, align 16 - %ptr = getelementptr <2 x double>* %array, i32 %b + %ptr = getelementptr <2 x double>, <2 x double>* %array, i32 %b store <2 x double> %val, <2 x double>* %ptr, align 16 call void @llvm.AMDGPU.barrier.local() noduplicate nounwind - %result = load <2 x double>* %ptr, align 16 + %result = load <2 x double>, <2 x double>* %ptr, align 16 store <2 x double> %result, <2 x double> addrspace(1)* %out, align 16 ret void } @@ -56,12 +56,12 @@ define void @private_access_v2f64_alloca(<2 x double> addrspace(1)* noalias %out ; SI-PROMOTE: ds_write_b64 ; SI-PROMOTE: ds_read_b64 define void @private_access_i64_alloca(i64 addrspace(1)* noalias %out, i64 addrspace(1)* noalias %in, i32 %b) nounwind { - %val = load i64 addrspace(1)* %in, align 8 + %val = load i64, i64 addrspace(1)* %in, align 8 %array = alloca i64, i32 16, align 8 - %ptr = getelementptr i64* %array, i32 %b + %ptr = getelementptr i64, i64* %array, i32 %b store i64 %val, i64* %ptr, align 8 call void @llvm.AMDGPU.barrier.local() noduplicate nounwind - %result = load i64* %ptr, align 8 + %result = load i64, i64* %ptr, align 8 store i64 %result, i64 addrspace(1)* %out, align 8 ret void } @@ -80,12 +80,12 @@ define void @private_access_i64_alloca(i64 addrspace(1)* noalias %out, i64 addrs ; SI-PROMOTE: ds_read_b32 ; SI-PROMOTE: ds_read_b32 define void @private_access_v2i64_alloca(<2 x i64> addrspace(1)* noalias %out, <2 x i64> addrspace(1)* noalias %in, i32 %b) nounwind { - %val = load <2 x i64> addrspace(1)* %in, align 16 + %val = load <2 x i64>, <2 x i64> addrspace(1)* %in, align 16 %array = alloca <2 x i64>, i32 16, align 16 - %ptr = getelementptr <2 x i64>* %array, i32 %b + %ptr = getelementptr <2 x i64>, <2 x i64>* %array, i32 %b store <2 x i64> %val, <2 x i64>* %ptr, align 16 call void @llvm.AMDGPU.barrier.local() noduplicate nounwind - %result = load <2 x i64>* %ptr, align 16 + %result = load <2 x i64>, <2 x i64>* %ptr, align 16 store <2 x i64> %result, <2 x i64> addrspace(1)* %out, align 16 ret void } |