diff options
| author | Matt Arsenault <Matthew.Arsenault@amd.com> | 2016-08-20 00:01:21 +0000 |
|---|---|---|
| committer | Matt Arsenault <Matthew.Arsenault@amd.com> | 2016-08-20 00:01:21 +0000 |
| commit | f5b81cfc499b034ab79cdfe2bf4596438635203f (patch) | |
| tree | 1f1d6312603071c698b6cb56a8f9d5ff62ec83f0 | |
| parent | a28ef485c9e8b0f227bb1595ddd797a838bcedb4 (diff) | |
amdgcn: Fix get_local_size IR return type
git-svn-id: https://llvm.org/svn/llvm-project/libclc/trunk@279350 91177308-0d34-0410-b5e6-96231b3b80d8
| -rw-r--r-- | amdgcn/lib/SOURCES | 1 | ||||
| -rw-r--r-- | amdgcn/lib/workitem/get_local_size.ll | 21 | ||||
| -rw-r--r-- | amdgpu/lib/SOURCES | 1 | ||||
| -rw-r--r-- | r600/lib/SOURCES | 1 | ||||
| -rw-r--r-- | r600/lib/workitem/get_local_size.ll (renamed from amdgpu/lib/workitem/get_local_size.ll) | 8 |
5 files changed, 27 insertions, 5 deletions
diff --git a/amdgcn/lib/SOURCES b/amdgcn/lib/SOURCES index ca88b93..06ed228 100644 --- a/amdgcn/lib/SOURCES +++ b/amdgcn/lib/SOURCES @@ -3,4 +3,5 @@ synchronization/barrier_impl.ll workitem/get_global_offset.cl workitem/get_group_id.cl workitem/get_local_id.cl +workitem/get_local_size.ll workitem/get_work_dim.cl diff --git a/amdgcn/lib/workitem/get_local_size.ll b/amdgcn/lib/workitem/get_local_size.ll new file mode 100644 index 0000000..59dd913 --- /dev/null +++ b/amdgcn/lib/workitem/get_local_size.ll @@ -0,0 +1,21 @@ +declare i32 @llvm.r600.read.local.size.x() nounwind readnone +declare i32 @llvm.r600.read.local.size.y() nounwind readnone +declare i32 @llvm.r600.read.local.size.z() nounwind readnone + +define i64 @get_local_size(i32 %dim) nounwind readnone alwaysinline { + switch i32 %dim, label %default [i32 0, label %x_dim i32 1, label %y_dim i32 2, label %z_dim] +x_dim: + %x = call i32 @llvm.r600.read.local.size.x() + %x.ext = zext i32 %x to i64 + ret i64 %x.ext +y_dim: + %y = call i32 @llvm.r600.read.local.size.y() + %y.ext = zext i32 %y to i64 + ret i64 %y.ext +z_dim: + %z = call i32 @llvm.r600.read.local.size.z() + %z.ext = zext i32 %z to i64 + ret i64 %z.ext +default: + unreachable +} diff --git a/amdgpu/lib/SOURCES b/amdgpu/lib/SOURCES index 403e1e7..efee251 100644 --- a/amdgpu/lib/SOURCES +++ b/amdgpu/lib/SOURCES @@ -17,5 +17,4 @@ image/write_imagei.cl image/write_imageui.cl image/write_image_impl.ll workitem/get_num_groups.ll -workitem/get_local_size.ll workitem/get_global_size.ll diff --git a/r600/lib/SOURCES b/r600/lib/SOURCES index 33038f2..23da502 100644 --- a/r600/lib/SOURCES +++ b/r600/lib/SOURCES @@ -2,4 +2,5 @@ synchronization/barrier_impl.ll workitem/get_global_offset.cl workitem/get_group_id.cl workitem/get_local_id.cl +workitem/get_local_size.ll workitem/get_work_dim.cl diff --git a/amdgpu/lib/workitem/get_local_size.ll b/r600/lib/workitem/get_local_size.ll index 0a98de6..136c9c6 100644 --- a/amdgpu/lib/workitem/get_local_size.ll +++ b/r600/lib/workitem/get_local_size.ll @@ -5,14 +5,14 @@ declare i32 @llvm.r600.read.local.size.z() nounwind readnone define i32 @get_local_size(i32 %dim) nounwind readnone alwaysinline { switch i32 %dim, label %default [i32 0, label %x_dim i32 1, label %y_dim i32 2, label %z_dim] x_dim: - %x = call i32 @llvm.r600.read.local.size.x() nounwind readnone + %x = call i32 @llvm.r600.read.local.size.x() ret i32 %x y_dim: - %y = call i32 @llvm.r600.read.local.size.y() nounwind readnone + %y = call i32 @llvm.r600.read.local.size.y() ret i32 %y z_dim: - %z = call i32 @llvm.r600.read.local.size.z() nounwind readnone + %z = call i32 @llvm.r600.read.local.size.z() ret i32 %z default: - ret i32 0 + unreachable } |