diff options
Diffstat (limited to 'mlir/test/Dialect/GPU/transform-gpu.mlir')
| -rw-r--r-- | mlir/test/Dialect/GPU/transform-gpu.mlir | 44 |
1 files changed, 22 insertions, 22 deletions
diff --git a/mlir/test/Dialect/GPU/transform-gpu.mlir b/mlir/test/Dialect/GPU/transform-gpu.mlir index d9872a9666f6..64d80f95a6b3 100644 --- a/mlir/test/Dialect/GPU/transform-gpu.mlir +++ b/mlir/test/Dialect/GPU/transform-gpu.mlir @@ -31,9 +31,9 @@ func.func @saxpy2dblock(%x: !type, %y: !type, %t: !type1d, %alpha : f32, %stream } transform.sequence failures(propagate) { -^bb1(%arg0: !pdl.operation): - %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation - transform.gpu.map_forall_to_blocks %funcop grid_dims = [12, 9, 1] +^bb1(%arg0: !transform.any_op): + %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!transform.any_op) -> !transform.any_op + transform.gpu.map_forall_to_blocks %funcop grid_dims = [12, 9, 1] : (!transform.any_op) -> !transform.any_op } // ----- @@ -85,9 +85,9 @@ func.func @saxpy2d(%x: !type, %y: !type, %t: !type1d, %alpha : f32, %stream : !g } transform.sequence failures(propagate) { -^bb1(%arg0: !pdl.operation): - %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation - transform.gpu.map_nested_forall_to_threads %funcop block_dims = [12, 9, 1] +^bb1(%arg0: !transform.any_op): + %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!transform.any_op) -> !transform.any_op + transform.gpu.map_nested_forall_to_threads %funcop block_dims = [12, 9, 1] : (!transform.any_op) -> !transform.any_op } // ----- @@ -124,10 +124,10 @@ func.func @saxpy4d(%x: !type4d, %y: !type4d, %alpha : f32) -> !type4d { } transform.sequence failures(propagate) { -^bb1(%arg0: !pdl.operation): - %funcop = transform.structured.match ops{["func.func"]} in %arg0 : (!pdl.operation) -> !pdl.operation - %gpuLaunch = transform.gpu.map_forall_to_blocks %funcop { generate_gpu_launch } - transform.gpu.map_nested_forall_to_threads %gpuLaunch block_dims = [32, 4, 1] +^bb1(%arg0: !transform.any_op): + %funcop = transform.structured.match ops{["func.func"]} in %arg0 : (!transform.any_op) -> !transform.any_op + %gpuLaunch = transform.gpu.map_forall_to_blocks %funcop { generate_gpu_launch } : (!transform.any_op) -> !transform.any_op + transform.gpu.map_nested_forall_to_threads %gpuLaunch block_dims = [32, 4, 1] : (!transform.any_op) -> !transform.any_op } // ----- @@ -158,9 +158,9 @@ func.func @saxpy2d_no_barrier(%x: !type, %y: !type, %t: !type1d, %alpha : f32, % } transform.sequence failures(propagate) { -^bb1(%arg0: !pdl.operation): - %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation - transform.gpu.map_nested_forall_to_threads %funcop block_dims = [12, 9, 1] sync_after_distribute = false +^bb1(%arg0: !transform.any_op): + %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!transform.any_op) -> !transform.any_op + transform.gpu.map_nested_forall_to_threads %funcop block_dims = [12, 9, 1] sync_after_distribute = false : (!transform.any_op) -> !transform.any_op } // ----- @@ -190,9 +190,9 @@ func.func @saxpy2d_singleloop(%x: !type, %y: !type, %stream : !gpu.async.token) } transform.sequence failures(propagate) { -^bb1(%arg0: !pdl.operation): - %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation - transform.gpu.map_nested_forall_to_threads %funcop block_dims = [32, 1, 1] +^bb1(%arg0: !transform.any_op): + %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!transform.any_op) -> !transform.any_op + transform.gpu.map_nested_forall_to_threads %funcop block_dims = [32, 1, 1] : (!transform.any_op) -> !transform.any_op } // ----- @@ -226,9 +226,9 @@ func.func @saxpy3d_fold_id_z(%x: !type, %y: !type, %t: !type1d, %alpha : f32, %s } transform.sequence failures(propagate) { -^bb1(%arg0: !pdl.operation): - %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation - transform.gpu.map_nested_forall_to_threads %funcop block_dims = [12, 9, 1] sync_after_distribute = false +^bb1(%arg0: !transform.any_op): + %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!transform.any_op) -> !transform.any_op + transform.gpu.map_nested_forall_to_threads %funcop block_dims = [12, 9, 1] sync_after_distribute = false : (!transform.any_op) -> !transform.any_op } // ----- @@ -302,8 +302,8 @@ func.func @map_multi_level(%x: !type, %y: !type, %t: !type1d, %alpha : f32, %str } transform.sequence failures(propagate) { -^bb1(%arg0: !pdl.operation): - %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!pdl.operation) -> !pdl.operation +^bb1(%arg0: !transform.any_op): + %funcop = transform.structured.match ops{["gpu.launch"]} in %arg0 : (!transform.any_op) -> !transform.any_op transform.gpu.map_nested_forall_to_threads %funcop - block_dims = [12, 11, 1] warp_dims = [3, 2, 1] + block_dims = [12, 11, 1] warp_dims = [3, 2, 1] : (!transform.any_op) -> !transform.any_op } |