pashu123
/ new_ir.mlir
Created
June 27, 2024 16:17
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| util.func public @matmul_broad(%arg0: !hal.buffer_view, %arg1: !hal.buffer_view) -> !hal.buffer_view attributes {iree.abi.stub, iree.reflection = {iree.abi.declaration = "sync func @matmul_broad(%input0: tensor<?x?x3200xf32>, %input1: tensor<8640x3200xf16>) -> (%output0: tensor<?x?x8640xf32>)"}} { | |
| %cst = arith.constant 0.000000e+00 : f16 | |
| %c1 = arith.constant 1 : index | |
| %c0 = arith.constant 0 : index | |
| %cst_0 = arith.constant 0.000000e+00 : f32 | |
| %0 = hal.buffer_view.dim<%arg0 : !hal.buffer_view>[0] : index | |
| %1 = hal.buffer_view.dim<%arg0 : !hal.buffer_view>[1] : index | |
| %2 = hal.tensor.import %arg0 "input0" : !hal.buffer_view -> tensor<?x?x3200xf32>{%0, %1} | |
| %3 = hal.tensor.import %arg1 "input1" : !hal.buffer_view -> tensor<8640x3200xf16> | |
| %4 = tensor.empty(%0) : tensor<?x8640x3200xf16> |
pashu123
/ new_ir.mlir
Created
June 27, 2024 16:25
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| %46 = linalg.batch_mmt4d {lowering_config = #iree_codegen.lowering_config<tile_sizes = [[1, 1, 1, 0, 0, 0, 0], [1, 1, 1, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0, 0], [1, 1, 1, 0, 16, 16, 0], [0, 0, 0, 1, 0, 0, 1], [0, 0, 0, 0, 0, 0, 0]]>} ins(%41, %42 : tensor<?x?x3200x16x1xf32>, tensor<?x540x3200x16x1xf16>) outs(%45 : tensor<?x?x540x16x16xf32>) -> tensor<?x?x540x16x16xf32> | |
| util.func public @matmul_broad(%arg0: !hal.buffer_view, %arg1: !hal.buffer_view) -> !hal.buffer_view attributes {iree.abi.stub, iree.reflection = {iree.abi.declaration = "sync func @matmul_broad(%input0: tensor<?x?x3200xf32>, %input1: tensor<8640x3200xf16>) -> (%output0: tensor<?x?x8640xf32>)"}} { | |
| %cst = arith.constant 0.000000e+00 : f16 | |
| %c1 = arith.constant 1 : index | |
| %c0 = arith.constant 0 : index | |
| %cst_0 = arith.constant 0.000000e+00 : f32 | |
| %0 = hal.buffer_view.dim<%arg0 : !hal.buffer_view>[0] : index | |
| %1 = hal.buffer_view.dim<%arg0 : !hal.buffer_view>[1] : index | |
| %2 = hal.tensor.import %arg0 "input0" : !hal.buffer_view -> tensor<?x?x |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| #matmul_config = #iree_codegen.lowering_config<tile_sizes = [[1, 1, 0, 0, 0, 0], [1, 1, 0, 0, 0, 0], [0, 0, 0, 0, 0, 0], [1, 1, 0, 16, 16, 0], [0, 0, 1, 0, 0, 1], [0, 0, 0, 0, 0, 0]]> | |
| #executable_target_embedded_elf_x86_64_ = #hal.executable.target<"llvm-cpu", "embedded-elf-x86_64", {cpu = "znver4", cpu_features = "+avx512f", native_vector_size = 64 : index, target_triple = "x86_64-unknown-unknown-eabi-elf"}> | |
| func.func @mmt4d_bias_relu_fusion_dispatch_0_generic_DxDx16x16_f32() attributes {hal.executable.target = #executable_target_embedded_elf_x86_64_} { | |
| %c0 = arith.constant 0 : index | |
| %c32_i64 = arith.constant 32 : i64 | |
| %cst = arith.constant 0.000000e+00 : f32 | |
| %0 = hal.interface.constant.load[0] : i32 | |
| %1 = hal.interface.constant.load[1] : i32 | |
| %2 = hal.interface.constant.load[2] : i32 | |
| %3 = hal.interface.constant.load[3] : i32 |
pashu123
/ e2e_matmul_cpu_experimental_dt_uk_f16_f16_small_llvm-cpu_local-task_avx512_matmul.mlir
Created
August 9, 2024 18:42
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| func.func @matmul_accumulate_DYNxDYNxf16_times_DYNxDYNxf16_into_DYNxDYNxf16(%lhs: tensor<?x?xf16>, %rhs: tensor<?x?xf16>, %acc: tensor<?x?xf16>) -> tensor<?x?xf16> { | |
| %result = linalg.matmul ins(%lhs, %rhs: tensor<?x?xf16>, tensor<?x?xf16>) outs(%acc: tensor<?x?xf16>) -> tensor<?x?xf16> | |
| return %result: tensor<?x?xf16> | |
| } | |
| func.func @matmul_accumulate_1x1xf16_times_1x1xf16_into_1x1xf16(%lhs: tensor<1x1xf16>, %rhs: tensor<1x1xf16>, %acc: tensor<1x1xf16>) -> tensor<1x1xf16> { | |
| %result = linalg.matmul ins(%lhs, %rhs: tensor<1x1xf16>, tensor<1x1xf16>) outs(%acc: tensor<1x1xf16>) -> tensor<1x1xf16> | |
| return %result: tensor<1x1xf16> |
pashu123
/ double_generic.mlir
Created
August 12, 2024 12:34
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| #map = affine_map<(d0) -> (d0)> | |
| #map1 = affine_map<(d0, d1, d2) -> (d0, d1, d2)> | |
| module { | |
| func.func @main_graph_dispatch_47_elementwise_64x56x56_f32(%arg0: tensor<200704xi8>, %arg1: tensor<64x56x56xf32>) -> tensor<64x56x56xf32> { | |
| %cst = arith.constant 0.000000e+00 : f32 | |
| %cst_0 = arith.constant -1.280000e+02 : f32 | |
| %cst_1 = arith.constant 1.270000e+02 : f32 | |
| %cst_2 = arith.constant 1.562500e-02 : f32 | |
| %0 = tensor.empty() : tensor<64x56x56xf32> | |
| %1 = tensor.empty() : tensor<200704xf32> |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| #map = affine_map<(d0, d1, d2) -> (d1, d2)> | |
| #map1 = affine_map<(d0, d1, d2) -> (d0, d1, d2)> | |
| #map2 = affine_map<(d0, d1, d2, d3) -> (d0, d1, d3)> | |
| #map3 = affine_map<(d0, d1, d2, d3) -> (d0, d2, d3)> | |
| #map4 = affine_map<(d0, d1, d2, d3) -> (d0, d1, d2)> | |
| module { | |
| util.func public @matmul_broad(%arg0: !hal.buffer_view, %arg1: !hal.buffer_view) -> !hal.buffer_view attributes {iree.abi.stub, iree.reflection = {iree.abi.declaration = "sync func @matmul_broad(%input0: tensor<?x?x3200xf32>, %input1: tensor<8640x3200xf16>) -> (%output0: tensor<?x?x8640xf32>)"}} { | |
| %cst = arith.constant 0.000000e+00 : f32 | |
| %0 = hal.buffer_view.dim<%arg0 : !hal.buffer_view>[0] : index | |
| %1 = hal.buffer_view.dim<%arg0 : !hal.buffer_view>[1] : index |
pashu123
/ 2ndbug.mlir
Created
August 19, 2024 16:46
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| func.func @matmul_broad_dispatch_1_set_encoding_LHS_DxDx3200() { | |
| %c0 = arith.constant 0 : index | |
| %c32_i64 = arith.constant 32 : i64 | |
| %0 = hal.interface.constant.load layout(<push_constants = 4, sets = [<0, bindings = [<0, storage_buffer, Indirect>], flags = Indirect>]>) ordinal(0) : i32 | |
| %1 = hal.interface.constant.load layout(<push_constants = 4, sets = [<0, bindings = [<0, storage_buffer, Indirect>], flags = Indirect>]>) ordinal(1) : i32 | |
| %2 = hal.interface.constant.load layout(<push_constants = 4, sets = [<0, bindings = [<0, storage_buffer, Indirect>], flags = Indirect>]>) ordinal(2) : i32 | |
| %3 = hal.interface.constant.load layout(<push_constants = 4, sets = [<0, bindings = [<0, storage_buffer, Indirect>], flags = Indirect>]>) ordinal(3) : i32 | |
| %4 = arith.extui %0 : i32 to i64 | |
| %5 = arith.extui %1 : i32 to i64 | |
| %6 = arith.shli %5, %c32_i64 : i64 |
This file has been truncated, but you can view the full file.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| // -----// IR Dump After AutoInputConversionPipeline (iree-auto-input-conversion) //----- // | |
| #map = affine_map<(d0, d1, d2) -> (d1, d2)> | |
| #map1 = affine_map<(d0, d1, d2) -> (d0, d1, d2)> | |
| module { | |
| util.func public @matmul_broad(%arg0: tensor<?x?x3200xf32>, %arg1: tensor<8640x3200xf16>) -> tensor<?x?x8640xf32> { | |
| %cst = arith.constant 0.000000e+00 : f32 | |
| %c0 = arith.constant 0 : index | |
| %c1 = arith.constant 1 : index | |
| %dim = tensor.dim %arg0, %c0 : tensor<?x?x3200xf32> | |
| %dim_0 = tensor.dim %arg0, %c1 : tensor<?x?x3200xf32> |
This file has been truncated, but you can view the full file.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| // -----// IR Dump After AutoInputConversionPipeline (iree-auto-input-conversion) //----- // | |
| #map = affine_map<(d0, d1, d2) -> (d1, d2)> | |
| #map1 = affine_map<(d0, d1, d2) -> (d0, d1, d2)> | |
| module { | |
| util.func public @matmul_broad(%arg0: tensor<?x?x3200xf32>, %arg1: tensor<8640x3200xf16>) -> tensor<?x?x8640xf32> { | |
| %cst = arith.constant 0.000000e+00 : f32 | |
| %c0 = arith.constant 0 : index | |
| %c1 = arith.constant 1 : index | |
| %dim = tensor.dim %arg0, %c0 : tensor<?x?x3200xf32> | |
| %dim_0 = tensor.dim %arg0, %c1 : tensor<?x?x3200xf32> |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| func.func @matmul_fusion_test(%arg0 : tensor<?x?xf32>, | |
| %arg1 : tensor<?x?xf32>) -> tensor<?x?xf32> { | |
| %c0 = arith.constant 0 : index | |
| %c1 = arith.constant 1 : index | |
| %c2 = arith.constant 2 : index | |
| %cst0 = arith.constant 0.0 : f32 | |
| %M = tensor.dim %arg0, %c0 : tensor<?x?xf32> | |
| %N = tensor.dim %arg1, %c1 : tensor<?x?xf32> | |
| %K = tensor.dim %arg0, %c1 : tensor<?x?xf32> | |
| %empty = tensor.empty(%M, %N) : tensor<?x?xf32> |