diff options
author | Matt Arsenault <Matthew.Arsenault@amd.com> | 2016-08-25 07:31:40 +0000 |
---|---|---|
committer | Matt Arsenault <Matthew.Arsenault@amd.com> | 2016-08-25 07:31:40 +0000 |
commit | 6c3f6de9017feea72e22be98d02460239fa10717 (patch) | |
tree | 801527111f54284b0bffbf512c164d0512b86478 | |
parent | 1b2f5e03bc09705d5da69e8e4144f65f758d305d (diff) |
amdgcn: Fix return type of get_num_groups
git-svn-id: https://llvm.org/svn/llvm-project/libclc/trunk@279723 91177308-0d34-0410-b5e6-96231b3b80d8
-rw-r--r-- | amdgcn/lib/SOURCES | 1 | ||||
-rw-r--r-- | amdgcn/lib/workitem/get_num_groups.ll | 21 | ||||
-rw-r--r-- | amdgpu/lib/SOURCES | 1 | ||||
-rw-r--r-- | r600/lib/SOURCES | 1 | ||||
-rw-r--r-- | r600/lib/workitem/get_num_groups.ll (renamed from amdgpu/lib/workitem/get_num_groups.ll) | 2 |
5 files changed, 24 insertions, 2 deletions
diff --git a/amdgcn/lib/SOURCES b/amdgcn/lib/SOURCES index f7e2671..1ff5fd1 100644 --- a/amdgcn/lib/SOURCES +++ b/amdgcn/lib/SOURCES @@ -5,4 +5,5 @@ workitem/get_group_id.cl workitem/get_global_size.ll workitem/get_local_id.cl workitem/get_local_size.ll +workitem/get_num_groups.ll workitem/get_work_dim.cl diff --git a/amdgcn/lib/workitem/get_num_groups.ll b/amdgcn/lib/workitem/get_num_groups.ll new file mode 100644 index 0000000..5b4b1cc --- /dev/null +++ b/amdgcn/lib/workitem/get_num_groups.ll @@ -0,0 +1,21 @@ +declare i32 @llvm.r600.read.ngroups.x() nounwind readnone +declare i32 @llvm.r600.read.ngroups.y() nounwind readnone +declare i32 @llvm.r600.read.ngroups.z() nounwind readnone + +define i64 @get_num_groups(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.ngroups.x() + %x.ext = zext i32 %x to i64 + ret i64 %x.ext +y_dim: + %y = call i32 @llvm.r600.read.ngroups.y() + %y.ext = zext i32 %y to i64 + ret i64 %y.ext +z_dim: + %z = call i32 @llvm.r600.read.ngroups.z() + %z.ext = zext i32 %z to i64 + ret i64 %z.ext +default: + ret i64 1 +} diff --git a/amdgpu/lib/SOURCES b/amdgpu/lib/SOURCES index fac2fc3..9379ad7 100644 --- a/amdgpu/lib/SOURCES +++ b/amdgpu/lib/SOURCES @@ -16,4 +16,3 @@ image/write_imagef.cl image/write_imagei.cl image/write_imageui.cl image/write_image_impl.ll -workitem/get_num_groups.ll diff --git a/r600/lib/SOURCES b/r600/lib/SOURCES index d55ff30..284052c 100644 --- a/r600/lib/SOURCES +++ b/r600/lib/SOURCES @@ -4,4 +4,5 @@ workitem/get_group_id.cl workitem/get_global_size.ll workitem/get_local_id.cl workitem/get_local_size.ll +workitem/get_num_groups.ll workitem/get_work_dim.cl diff --git a/amdgpu/lib/workitem/get_num_groups.ll b/r600/lib/workitem/get_num_groups.ll index a708f42..d85a797 100644 --- a/amdgpu/lib/workitem/get_num_groups.ll +++ b/r600/lib/workitem/get_num_groups.ll @@ -14,5 +14,5 @@ z_dim: %z = call i32 @llvm.r600.read.ngroups.z() nounwind readnone ret i32 %z default: - ret i32 0 + ret i32 1 } |