derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
75 lines
4.6 KiB
MLIR
75 lines
4.6 KiB
MLIR
// RUN: mlir-opt -convert-to-emitc %s | FileCheck %s
|
|
|
|
/// NOTE: This test intentionally uses `-convert-to-emitc` only.
|
|
///
|
|
/// The `-convert-memref-to-emitc` pass introduces
|
|
/// `builtin.unrealized_conversion_cast` operations when lowering
|
|
/// `memref.alloc` results (which are lowered to `emitc.ptr`) to the canonical
|
|
/// memref representation used by the type converter (`emitc.array`).
|
|
/// These casts are expected at that stage of the pipeline.
|
|
///
|
|
/// The purpose of this test is to verify the final lowering produced by
|
|
/// `-convert-to-emitc`, where `memref.load` and `memref.store` conversions now
|
|
/// handle pointer-backed buffers directly and eliminate the intermediate
|
|
/// `unrealized_conversion_cast`.
|
|
/// Therefore, the test must run the full EmitC conversion pipeline.
|
|
|
|
/// AllocOp conversion always returns a ptr
|
|
// CHECK-LABEL: emitc.func private @memref_alloc_store(
|
|
// CHECK-SAME: %[[VAL:.*]]: f32,
|
|
// CHECK-SAME: %[[ARG_I:.*]]: !emitc.size_t,
|
|
// CHECK-SAME: %[[ARG_J:.*]]: !emitc.size_t)
|
|
func.func private @memref_alloc_store(%v : f32, %i: index, %j: index) {
|
|
/// Allocation size computation
|
|
// CHECK: %[[SIZEOF_F32:.*]] = call_opaque "sizeof"() <{args = [f32]}> : () -> !emitc.size_t
|
|
// CHECK: %[[NUM_ELEMS:.*]] = "emitc.constant"() <{value = 32 : index}> : () -> index
|
|
// CHECK: %[[TOTAL_BYTES:.*]] = mul %[[SIZEOF_F32]], %[[NUM_ELEMS]] : (!emitc.size_t, index) -> !emitc.size_t
|
|
/// Alloc
|
|
// CHECK: %[[MALLOC_PTR:.*]] = call_opaque "malloc"(%[[TOTAL_BYTES]]) : (!emitc.size_t) -> !emitc.ptr<!emitc.opaque<"void">>
|
|
// CHECK: %[[ELEM_PTR:.*]] = cast %[[MALLOC_PTR]] : !emitc.ptr<!emitc.opaque<"void">> to !emitc.ptr<f32>
|
|
%alloc = memref.alloc() : memref<4x8xf32>
|
|
/// Store subscript computation
|
|
// CHECK: %[[ROW_STRIDE:.*]] = "emitc.constant"() <{value = 8 : index}> : () -> !emitc.size_t
|
|
// CHECK: %[[ROW_OFFSET:.*]] = mul %[[ARG_I]], %[[ROW_STRIDE]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
|
|
// CHECK: %[[LINEAR_INDEX:.*]] = add %[[ROW_OFFSET]], %[[ARG_J]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
|
|
// CHECK: %[[ELEM_LVALUE:.*]] = subscript %[[ELEM_PTR]]{{\[}}%[[LINEAR_INDEX]]] : (!emitc.ptr<f32>, !emitc.size_t) -> !emitc.lvalue<f32>
|
|
/// Store
|
|
// CHECK: assign %[[VAL]] : f32 to %[[ELEM_LVALUE]] : <f32>
|
|
memref.store %v, %alloc[%i, %j] : memref<4x8xf32>
|
|
return
|
|
}
|
|
// CHECK-LABEL: emitc.func private @memref_alloc_load(
|
|
// CHECK-SAME: %[[ARG_I:.*]]: !emitc.size_t,
|
|
// CHECK-SAME: %[[ARG_J:.*]]: !emitc.size_t) -> f32
|
|
func.func private @memref_alloc_load(%i: index, %j: index) -> f32 {
|
|
// CHECK: %[[SIZEOF_F32:.*]] = call_opaque "sizeof"() <{args = [f32]}> : () -> !emitc.size_t
|
|
// CHECK: %[[NUM_ELEMS:.*]] = "emitc.constant"() <{value = 32 : index}> : () -> index
|
|
// CHECK: %[[TOTAL_BYTES:.*]] = mul %[[SIZEOF_F32]], %[[NUM_ELEMS]] : (!emitc.size_t, index) -> !emitc.size_t
|
|
// CHECK: %[[MALLOC_PTR:.*]] = call_opaque "malloc"(%[[TOTAL_BYTES]]) : (!emitc.size_t) -> !emitc.ptr<!emitc.opaque<"void">>
|
|
// CHECK: %[[ELEM_PTR:.*]] = cast %[[MALLOC_PTR]] : !emitc.ptr<!emitc.opaque<"void">> to !emitc.ptr<f32>
|
|
%alloc = memref.alloc() : memref<4x8xf32>
|
|
// CHECK: %[[ROW_STRIDE:.*]] = "emitc.constant"() <{value = 8 : index}> : () -> !emitc.size_t
|
|
// CHECK: %[[ROW_OFFSET:.*]] = mul %[[ARG_I]], %[[ROW_STRIDE]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
|
|
// CHECK: %[[LINEAR_INDEX:.*]] = add %[[ROW_OFFSET]], %[[ARG_J]] : (!emitc.size_t, !emitc.size_t) -> !emitc.size_t
|
|
// CHECK: %[[ELEM_LVALUE:.*]] = subscript %[[ELEM_PTR]]{{\[}}%[[LINEAR_INDEX]]] : (!emitc.ptr<f32>, !emitc.size_t) -> !emitc.lvalue<f32>
|
|
// CHECK: %[[LOADED_VAL:.*]] = load %[[ELEM_LVALUE]] : <f32>
|
|
%v = memref.load %alloc[%i, %j] : memref<4x8xf32>
|
|
return %v : f32
|
|
}
|
|
|
|
/// LoadOp and StoreOp still compatible
|
|
/// Previous array paths still available
|
|
// CHECK-LABEL: emitc.func @memref_load_store(
|
|
// CHECK-SAME: %[[BUFF0:.*]]: !emitc.array<2xf32>,
|
|
// CHECK-SAME: %[[BUFF1:.*]]: !emitc.array<4x8xf32>,
|
|
func.func @memref_load_store(%buff0: memref<2xf32>,
|
|
%buff1: memref<4x8xf32>, %i : index, %j : index) {
|
|
// CHECK: %[[ELEM_LVALUE0:.*]] = subscript %[[BUFF0]]{{.*}} (!emitc.array<2xf32>, !emitc.size_t) -> !emitc.lvalue<f32>
|
|
// CHECK: %[[VAL:.*]] = load %[[ELEM_LVALUE0]] : <f32>
|
|
%v = memref.load %buff0[%i] : memref<2xf32>
|
|
// CHECK: %[[ELEM_LVALUE1:.*]] = subscript %[[BUFF1]]{{.*}} (!emitc.array<4x8xf32>, !emitc.size_t, !emitc.size_t) -> !emitc.lvalue<f32>
|
|
// CHECK: assign %[[VAL]] : f32 to %[[ELEM_LVALUE1]] : <f32>
|
|
memref.store %v, %buff1[%i, %j] : memref<4x8xf32>
|
|
return
|
|
}
|