llvm
/
torch-mlir
mirror of https://github.com/llvm/torch-mlir
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity
torch-mlir/test/RefBackend/munge-calling-conventions.mlir

80 lines
4.0 KiB
MLIR
Raw Normal View History

[cleanup] Make diagnostics better Also remove some unused imports.
2022-11-15 23:25:39 +08:00
// RUN: torch-mlir-opt %s -refback-munge-calling-conventions -split-input-file -verify-diagnostics | FileCheck %s
Bring up new RefBackend. `tools/torchscript_e2e_test.sh` is all green. This needs a few passes I put into torch-mlir/lib/RefBackend (not to be confused with `npcomp/lib/RefBackend`, which will soon be deleted). For the sake of review, since this brings together a lot of things, I split this into its own commit. I temporarily commented out some "list" stuff that we are going to remove as part of the torch-mlir refocus.
2021-09-23 00:55:09 +08:00
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
// CHECK-LABEL: func.func @f(
Bring up new RefBackend. `tools/torchscript_e2e_test.sh` is all green. This needs a few passes I put into torch-mlir/lib/RefBackend (not to be confused with `npcomp/lib/RefBackend`, which will soon be deleted). For the sake of review, since this brings together a lot of things, I split this into its own commit. I temporarily commented out some "list" stuff that we are going to remove as part of the torch-mlir refocus.
2021-09-23 00:55:09 +08:00
// CHECK-SAME: %[[ARG0:.*]]: memref<*xf32>) attributes {llvm.emit_c_interface} {
// CHECK: %[[VAL:.*]] = memref.cast %[[ARG0]] : memref<*xf32> to memref<?xf32>
// CHECK: %[[RESULT:.*]] = memref.cast %[[VAL]] : memref<?xf32> to memref<*xf32>
Add support for multiple return values This change is to unblock the work of some backprop ops returning more than one tensors. We will need to think of a more scalable approach in the future if more flexible return types combinations are needed.
2021-11-08 23:56:40 +08:00
// CHECK: call @refbackend_consume_func_return_mrf32(%[[RESULT]]) : (memref<*xf32>) -> ()
Bring up new RefBackend. `tools/torchscript_e2e_test.sh` is all green. This needs a few passes I put into torch-mlir/lib/RefBackend (not to be confused with `npcomp/lib/RefBackend`, which will soon be deleted). For the sake of review, since this brings together a lot of things, I split this into its own commit. I temporarily commented out some "list" stuff that we are going to remove as part of the torch-mlir refocus.
2021-09-23 00:55:09 +08:00
// CHECK: return
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func @f(%arg0: memref<?xf32>) -> memref<?xf32> {
Bring up new RefBackend. `tools/torchscript_e2e_test.sh` is all green. This needs a few passes I put into torch-mlir/lib/RefBackend (not to be confused with `npcomp/lib/RefBackend`, which will soon be deleted). For the sake of review, since this brings together a lot of things, I split this into its own commit. I temporarily commented out some "list" stuff that we are going to remove as part of the torch-mlir refocus.
2021-09-23 00:55:09 +08:00
return %arg0 : memref<?xf32>
}
add i64 support to refbackend
2021-10-05 10:06:59 +08:00
// -----
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
// CHECK-LABEL: func.func @i(
add i64 support to refbackend
2021-10-05 10:06:59 +08:00
// CHECK-SAME: %[[ARG0:.*]]: memref<*xi64>) attributes {llvm.emit_c_interface} {
// CHECK: %[[VAL:.*]] = memref.cast %[[ARG0]] : memref<*xi64> to memref<?xi64>
// CHECK: %[[RESULT:.*]] = memref.cast %[[VAL]] : memref<?xi64> to memref<*xi64>
Add support for multiple return values This change is to unblock the work of some backprop ops returning more than one tensors. We will need to think of a more scalable approach in the future if more flexible return types combinations are needed.
2021-11-08 23:56:40 +08:00
// CHECK: call @refbackend_consume_func_return_mri64(%[[RESULT]]) : (memref<*xi64>) -> ()
add i64 support to refbackend
2021-10-05 10:06:59 +08:00
// CHECK: return
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func @i(%arg0: memref<?xi64>) -> memref<?xi64> {
add i64 support to refbackend
2021-10-05 10:06:59 +08:00
return %arg0 : memref<?xi64>
}
Adding support for returning elemental types. Support for returning elemental types. Previously, only memref types as returning types was supported. All the hacky ways to write tests which return elemental types should be taken care of. Signed-off-by: Prashant Kumar <prashant@nod-labs.com>
2021-11-07 03:25:06 +08:00
// -----
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
// CHECK-LABEL: func.func @elemental_type(
Adding support for returning elemental types. Support for returning elemental types. Previously, only memref types as returning types was supported. All the hacky ways to write tests which return elemental types should be taken care of. Signed-off-by: Prashant Kumar <prashant@nod-labs.com>
2021-11-07 03:25:06 +08:00
// CHECK-SAME: %[[ARG0:.*]]: memref<*xi64>) attributes {llvm.emit_c_interface} {
// CHECK: %[[VAL:.*]] = memref.cast %[[ARG0]] : memref<*xi64> to memref<i64>
// CHECK: %[[RESULT:.*]] = memref.load %[[VAL]][] : memref<i64>
Add support for multiple return values This change is to unblock the work of some backprop ops returning more than one tensors. We will need to think of a more scalable approach in the future if more flexible return types combinations are needed.
2021-11-08 23:56:40 +08:00
// CHECK: call @refbackend_consume_func_return_i64(%[[RESULT]]) : (i64) -> ()
Adding support for returning elemental types. Support for returning elemental types. Previously, only memref types as returning types was supported. All the hacky ways to write tests which return elemental types should be taken care of. Signed-off-by: Prashant Kumar <prashant@nod-labs.com>
2021-11-07 03:25:06 +08:00
// CHECK: return
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func @elemental_type(%arg0: memref<i64>) -> i64 {
Adding support for returning elemental types. Support for returning elemental types. Previously, only memref types as returning types was supported. All the hacky ways to write tests which return elemental types should be taken care of. Signed-off-by: Prashant Kumar <prashant@nod-labs.com>
2021-11-07 03:25:06 +08:00
%0 = memref.load %arg0[] : memref<i64>
return %0 : i64
}
Add support for multiple return values This change is to unblock the work of some backprop ops returning more than one tensors. We will need to think of a more scalable approach in the future if more flexible return types combinations are needed.
2021-11-08 23:56:40 +08:00
// -----
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
// CHECK-LABEL: func.func @multiple_return_values(
Add support for multiple return values This change is to unblock the work of some backprop ops returning more than one tensors. We will need to think of a more scalable approach in the future if more flexible return types combinations are needed.
2021-11-08 23:56:40 +08:00
// CHECK-SAME: %[[ARG0:.*]]: memref<*xf32>, %[[ARG1:.*]]: memref<*xf32>,
// CHECK-SAME: %[[ARG2:.*]]: memref<*xf32>) attributes {llvm.emit_c_interface} {
// CHECK: %[[VAL0:.*]] = memref.cast %[[ARG0]] : memref<*xf32> to memref<?xf32>
// CHECK: %[[VAL1:.*]] = memref.cast %[[ARG1]] : memref<*xf32> to memref<?xf32>
// CHECK: %[[VAL2:.*]] = memref.cast %[[ARG2]] : memref<*xf32> to memref<?xf32>
// CHECK: %[[RET0:.*]] = memref.cast %[[VAL0]] : memref<?xf32> to memref<*xf32>
// CHECK: %[[RET1:.*]] = memref.cast %[[VAL1]] : memref<?xf32> to memref<*xf32>
// CHECK: %[[RET2:.*]] = memref.cast %[[VAL2]] : memref<?xf32> to memref<*xf32>
// CHECK: call @refbackend_consume_func_return_mrf32_mrf32_mrf32(%[[RET0]], %[[RET1]], %[[RET2]])
// CHECK-SAME: : (memref<*xf32>, memref<*xf32>, memref<*xf32>) -> ()
// CHECK: return
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func @multiple_return_values(%arg0: memref<?xf32>, %arg1: memref<?xf32>, %arg2: memref<?xf32>) -> (memref<?xf32>, memref<?xf32>, memref<?xf32>) {
Add support for multiple return values This change is to unblock the work of some backprop ops returning more than one tensors. We will need to think of a more scalable approach in the future if more flexible return types combinations are needed.
2021-11-08 23:56:40 +08:00
return %arg0 ,%arg1, %arg2 : memref<?xf32>, memref<?xf32>, memref<?xf32>
}
Add support for two return values This commit adds support for two return values of type memref f32 and i64. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-01-10 21:24:08 +08:00
// -----
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
// CHECK-LABEL: func.func @two_return_values(
Add support for two return values This commit adds support for two return values of type memref f32 and i64. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-01-10 21:24:08 +08:00
// CHECK-SAME: %[[ARG0:.*]]: memref<*xf32>, %[[ARG1:.*]]: memref<*xi64>)
// CHECK-SAME: attributes {llvm.emit_c_interface} {
// CHECK: %[[VAL0:.*]] = memref.cast %[[ARG0]] : memref<*xf32> to memref<?xf32>
// CHECK: %[[VAL1:.*]] = memref.cast %[[ARG1]] : memref<*xi64> to memref<?xi64>
// CHECK: %[[RET0:.*]] = memref.cast %[[VAL0]] : memref<?xf32> to memref<*xf32>
// CHECK: %[[RET1:.*]] = memref.cast %[[VAL1]] : memref<?xi64> to memref<*xi64>
// CHECK: call @refbackend_consume_func_return_mrf32_mri64(%[[RET0]], %[[RET1]])
// CHECK-SAME: : (memref<*xf32>, memref<*xi64>) -> ()
// CHECK: return
mlir: bump llvm tag to 5380e3 (#856) In addition to updating the llvm-project submodule, this patch also: 1. updates shape functions and tests so that `func` and `call` operations refer to the `func` dialect 2. avoid duplicate registration of dialects
2022-05-17 03:54:35 +08:00
func.func @two_return_values(%arg0: memref<?xf32>, %arg1: memref<?xi64>) -> (memref<?xf32>, memref<?xi64>) {
Add support for two return values This commit adds support for two return values of type memref f32 and i64. Signed-Off By: Vivek Khandelwal <vivek@nod-labs.com>
2022-01-10 21:24:08 +08:00
return %arg0 ,%arg1 : memref<?xf32>, memref<?xi64>
}
[cleanup] Make diagnostics better Also remove some unused imports.
2022-11-15 23:25:39 +08:00
// -----
// expected-error-re @+1 {{argument must be a memref of {{.*}} but got 'tensor<?xf32>'}}
func.func @f(%arg0: tensor<?xf32>) {
return
}