{ "ml": { "v": "4", "fs": [ { "id": 0, "prob": 0.9196195602416992, "class": 0 } ], "prob": 0.91961956, "class": 0, "version": "v16.16", "thresholds": [ 0.9815004, 0.99404883 ], "analyzed_at": "2026-04-30T03:24:41Z" }, "path": "collective_emitter.cc", "raw": { "v": "4", "fs": [ { "f": "H(Db)Md(Pt)", "x": 1, "id": 0, "is": [ "TensorValue", "CreateAllReduceEmitterContext", "op.getOperands.size", "op.getOperands", "InvalidArgumentError", "op.getOperand.getType", "op.getOperand", "ConvertMlirTypeToPrimitiveType", "op.getOperand.print", "op.getOperand", "InvalidArgumentError", "StrFormat", "InvalidArgumentError", "op.getNumOperands", "op.getNumOperands", "ctx.xtile_entry_fn.getNumArguments", "InvalidArgumentError", "StrCat", "ctx.xtile_entry_fn.getNumArguments", "op.getOperand", "ctx.input_tile.getDefiningOp", "ctx.input_tile.getDefiningOp.getNumOperands", "ctx.input_tile.getDefiningOp", "ctx.input_tile.getDefiningOp.getOperand.getDefiningOp", "ctx.input_tile.getDefiningOp.getOperand", "ctx.input_tile.getDefiningOp", "InvalidArgumentError", "ctx.input_extract.getSource.getType.getShape", "ctx.input_extract.getSource.getType", "ctx.input_extract.getSource", "Product", "divideCeil", "ctx.input_tile.getType", "GetAllReduceStrategy", "op.getReplicaGroups.getShapedType.getDimSize", "op.getReplicaGroups.getShapedType", "op.getReplicaGroups", "MaybeBuildAllReduceInfo", "all_reduce.GetModule.config.debug_options.xla_gpu_unsupported_use_all_reduce_one_shot_kernel", "all_reduce.GetModule.config.debug_options", "all_reduce.GetModule.config", "all_reduce.GetModule", "VLOG", "all_reduce.device_list.replica_groups.empty", "all_reduce.device_list.replica_groups", "all_reduce.device_list", "VLOG", "all_reduce.name", "all_reduce.device_list.num_devices_per_group", "all_reduce.device_list", "has_single_bit", "VLOG", "all_reduce.name", "MatchReductionComputation", "all_reduce.called_computations.front", "all_reduce.called_computations", "reduction_kind.has_value", "all_reduce.operand_count", "ElementsIn", "all_reduce.operand.shape", "all_reduce.operand", "all_reduce.operand.shape.element_type", "all_reduce.operand.shape", "all_reduce.operand", "ByteSizeOfPrimitiveType", "GetAllReduceStrategy", "GetMaxSupportedAllReduceSizeBytes", "VLOG", "IsAllReduceKernelSupported", "reduction_kind.value", "reduction_kind.value", "GetBlockLevelFusionConfigForAllReduce", "device_info.cuda_compute_capability", "VLOG", "MaybeBuildAllReduceInfo", "all_reduce_info.has_value", "VLOG", "all_reduce.name", "all_reduce.shape", "AllReduceLaunchDimensions", "block_level_config.set_num_warps", "launch_dims.num_threads_per_block", "WarpSize", "block_level_config.set_num_ctas", "block_level_config.set_num_stages", "block_level_config.add_output_tiles", "GreedyPowerOfTwoTiles", "launch_dims.num_blocks", "output_tile.mutable_sizes.Assign", "output_tile.mutable_sizes", "tile_sizes.begin", "tile_sizes.end", "Product", "VLOG", "VLOG", "all_reduce.name", "GetAllReduceUnmanagedKernelArguments", "all_reduce.device_list.num_devices_per_group", "all_reduce.device_list", "unmanaged_arguments.reserve", "computation.num_parameters", "unmanaged_arguments.push_back", "MakeShape", "unmanaged_arguments.push_back", "MakeShape", "unmanaged_arguments.push_back", "MakeShape", "InsertDimensionAtIndex", "instr.shape", "unmanaged_arguments.push_back", "TF_RET_CHECK", "unmanaged_arguments.size", "computation.num_parameters", "PopulateReductionComputation", "op.getComputation", "reduction_computation_region.getBlocks", "block_op.getNumOperands", "op.getLoc", "block_op.getNumResults", "op.getLoc", "block_op.getName", "state.addOperands", "state.addTypes", "accumulator.getType", "builder.create", "new_op.getResult", "rewriter.notifyMatchFailure", "op.getLoc", "success", "emitter.Initialize", "result.ok", "LOG", "result.message", "failure", "emitter.EmitOneShot", "emitter.EmitTwoShot", "emitter.rewriter_.notifyMatchFailure", "emitter.ctx_.op.getLoc", "failed", "PopulateReductionComputation", "InternalError", "ctx_.xtile_entry_fn.getArgument", "CHECK", "device_rank_.getType.isInteger", "device_rank_.getType", "ctx_.xtile_entry_fn.getArgument", "CHECK", "signal_value_.getType.isInteger", "signal_value_.getType", "ctx_.xtile_entry_fn.getArgument", "ctx_.xtile_entry_fn.getArgument", "getElementTypeOrSelf", "ctx_.input_tile.getType", "StorageType", "get", "builder_.getI64Type", "get", "TF_ASSIGN_OR_RETURN", "GetPermutationMinorToMajor", "ctx_.input_extract.getSource.getType", "ctx_.input_extract.getSource", "ctx_.input_tile.getType.getShape", "ctx_.input_tile.getType", "MakeShapeWithDenseLayout", "input_tile_shape_dims.begin", "input_tile_shape_dims.end", "GreedyPowerOfTwoTiles", "CHECK_EQ", "Product", "input_tile_shape.dimensions", "Product", "StrJoin", "create", "builder_.getI64Type", "create", "signal_value.getType", "create", "signal_value.getType", "builder_.getI32IntegerAttr", "create", "create", "builder_.getI64IntegerAttr", "ByteSizeOfPrimitiveType", "ByteSizeOfPrimitiveType", "create", "create", "builder_.getI64IntegerAttr", "OkStatus", "create", "create", "create", "create", "CHECK", "GetRemoteBufferPtr", "CreateTensorOfPointersAndMask", "ctx_.input_extract.getStrides", "Cast", "create", "builder_.getI64Type", "builder_.getI64IntegerAttr", "LoadTileForRank", "CHECK", "GetRemoteBufferPtr", "Cast", "CreateTensorOfPointersAndMask", "ctx_.input_extract.getStrides", "create", "success", "create", "builder_.getI32IntegerAttr", "success", "RankToLocalOffsets", "tile_dims.size", "dimensions", "rank_idx.getType.isInteger", "rank_idx.getType", "create", "builder_.getI64Type", "offsets", "create", "builder_.getI64Type", "builder_.getI64IntegerAttr", "create", "builder_.getI64Type", "builder_.getI64IntegerAttr", "create", "create", "CalculateSubtileOffsets", "RankToLocalOffsets", "ctx_.input_tile.getType.getShape", "ctx_.input_tile.getType", "ctx_.input_extract.getOffsets", "local_offsets.size", "create", "builder_.getI64Type", "create", "tile_type.getShape", "get", "create", "create", "get", "builder_.getI32Type", "create", "get", "builder_.getI64Type", "create", "create.getResult", "create", "Value", "Value", "get", "create", "get", "builder_.getI64Type", "create", "create", "get", "Product", "tile_type.getEncoding", "create", "get", "tile_type.getEncoding", "EmitGatherLoad", "ctx_.input_tile.getType", "tile_type.getShape", "CreateTensorOfRemoteBufferPtrs", "CreateTensorOfPointersAndMask", "ctx_.input_extract.getOffsets", "ctx_.input_extract.getFullTileShape", "ctx_.input_extract.getStrides", "create.getResult", "create", "Value", "CHECK", "failed", "EmitCopyToSymmetric", "ctx_.input_extract.getOffsets", "ctx_.input_tile.getType.getShape", "ctx_.input_tile.getType", "rewriter_.notifyMatchFailure", "failed", "EmitSync", "rewriter_.notifyMatchFailure", "ctx_.input_extract.getOffsets", "ctx_.input_tile.getType.getShape", "ctx_.input_tile.getType", "LoadTileForRank", "LoadTileForRank", "reduce_computation_emitter_", "rewriter_.replaceOp", "success", "CHECK", "failed", "EmitCopyToSymmetric", "ctx_.input_extract.getOffsets", "ctx_.input_tile.getType.getShape", "ctx_.input_tile.getType", "rewriter_.notifyMatchFailure", "failed", "EmitSync", "rewriter_.notifyMatchFailure", "CalculateSubtileOffsets", "LoadTileForRank", "LoadTileForRank", "reduce_computation_emitter_", "failed", "EmitCopyToSymmetric", "rewriter_.notifyMatchFailure", "create", "builder_.getI32Type", "builder_.getI32IntegerAttr", "failed", "EmitSync", "rewriter_.notifyMatchFailure", "EmitGatherLoad", "gathered_tensor.ok", "rewriter_.notifyMatchFailure", "StrCat", "gathered_tensor.status.message", "gathered_tensor.status", "rewriter_.replaceOp", "gathered_tensor.value", "success", "GreedyPowerOfTwoTiles", "CHECK_GT", "output_shape.dimensions.size", "output_shape.dimensions", "MinorToMajor", "tile_sizes", "output_shape.dimensions", "max", "bit_floor", "min", "CeilOfRatio", "FloorOfRatio", "GetCollectiveBlockLevelFusionConfig", "fusion_instr.fused_expression_root", "root.opcode", "GetBlockLevelFusionConfigForAllReduce", "TrySetGpuBackendConfigForCollective", "GetCollectiveBlockLevelFusionConfig", "block_config.has_value", "VLOG", "fusion_instr.ToString", "gpu_backend_config.mutable_fusion_backend_config.set_kind", "gpu_backend_config.mutable_fusion_backend_config", "gpu_backend_config.mutable_fusion_backend_config.mutable_block_level_fusion_config", "gpu_backend_config.mutable_fusion_backend_config", "move", "TF_RETURN_IF_ERROR", "fusion_instr.set_backend_config", "move", "GetCollectiveUnmanagedKernelArguments", "fusion.fused_instructions_computation", "computation.root_instruction", "root.opcode", "GetAllReduceUnmanagedKernelArguments", "AddCollectiveMetadataArguments", "fn_arg_types.push_back", "b.getI32Type", "fn_arg_types.push_back", "b.getI32Type", "fn_arg_types.push_back", "get", "get", "b.getI32Type", "p.shape.element_type", "p.shape", "if", "b.getI16Type", "b.getI4Type", "TF_ASSIGN_OR_RETURN", "PrimitiveTypeToMlirType", "fn_arg_types.push_back.num_parameters", "fn_arg_types.push_back", "get", "get", "StorageType", "op.getLoc", "CreateAllReduceEmitterContext", "maybe_context.ok", "VLOG", "maybe_context.status.message", "maybe_context.status", "rewriter.notifyMatchFailure", "StrCat", "maybe_context.status.message", "maybe_context.status", "VLOG", "Emit", "maybe_context.value" ], "ms": { "text": { "digit_ratio": 0.01, "space_count": 9962.0, "total_lines": 1069.0, "char_entropy": 4.84, "unique_chars": 89.0, "import_density": 36.58, "string_density": 0.23, "ascii_art_lines": 1.0, "avg_line_length": 44.47, "max_line_length": 80.0, "empty_line_ratio": 0.05, "last_line_length": 24.0, "most_common_char": "e", "whitespace_ratio": 0.23, "most_common_ratio": 0.08, "identifier_density": 1.38, "line_length_stddev": 24.08, "normalized_import_count": 11.96, "normalized_string_count": 7.46, "repeated_char_sequences": 155.0, "suspicious_string_ratio": 0.01, "suspicious_comment_ratio": 0.02, "max_inline_whitespace_run": 29.0, "normalized_function_count": 0.21, "imports_to_functions_ratio": 55.86, "strings_to_functions_ratio": 34.86, "suspicious_identifier_ratio": 0.02, "normalized_unique_identifiers": 29.72, "identifiers_to_functions_ratio": 42.71 }, "imports": { "total": 391.0, "unique_modules": 201.0, "third_party_count": 391.0, "third_party_ratio": 1.0 }, "strings": { "total": 244.0, "avg_length": 32.93, "max_length": 113.0, "avg_entropy": 3.72, "total_bytes": 8035.0, "entropy_stddev": 0.59, "embedded_code_candidates": 3.0 }, "comments": { "chars": 10560.0, "lines": 233.0, "total": 254.0, "todo_count": 1.0, "to_code_ratio": 0.28, "empty_comments": 34.0, "url_in_comments": 1.0, "base64_in_comments": 2.0, "high_entropy_comments": 4.0 }, "functions": { "total": 7.0, "one_liners": 3.0, "length_stddev": 106.45, "over_100_lines": 2.0, "avg_name_length": 43.71, "no_params_count": 7.0, "avg_length_lines": 69.43, "max_length_lines": 313.0, "min_length_lines": 1.0, "avg_nesting_depth": 0.43, "max_nesting_depth": 1.0, "high_entropy_names": 2.0, "density_per_100_lines": 0.65, "code_in_functions_ratio": 0.45 }, "identifiers": { "total": 1476.0, "avg_length": 12.41, "max_length": 37.0, "min_length": 1.0, "avg_entropy": 2.91, "reuse_ratio": 0.2, "unique_count": 299.0, "length_stddev": 7.56, "has_digit_ratio": 0.04, "sequential_names": 5.0, "single_char_count": 4.0, "single_char_ratio": 0.01, "high_entropy_count": 65.0, "high_entropy_ratio": 0.22, "all_lowercase_ratio": 0.54, "all_uppercase_ratio": 0.04, "numeric_suffix_count": 5.0 } }, "ss": [ [ 655, "xla/backends/gpu/codegen/triton/collective_emitter.h" ], [ 656, "xla/backends/gpu/codegen/triton/collective_emitter.h" ], [ 720, "\u003calgorithm\u003e" ], [ 741, "\u003ccstdint\u003e" ], [ 760, "\u003coptional\u003e" ], [ 780, "\u003cstring\u003e" ], [ 798, "\u003ctype_traits\u003e" ], [ 821, "\u003cutility\u003e" ], [ 840, "\u003cvector\u003e" ], [ 859, "absl/functional/any_invocable.h" ], [ 860, "absl/functional/any_invocable.h" ], [ 902, "absl/log/check.h" ], [ 903, "absl/log/check.h" ], [ 930, "absl/log/log.h" ], [ 931, "absl/log/log.h" ], [ 956, "absl/status/status.h" ], [ 957, "absl/status/status.h" ], [ 988, "absl/status/statusor.h" ], [ 989, "absl/status/statusor.h" ], [ 1022, "absl/strings/str_cat.h" ], [ 1023, "absl/strings/str_cat.h" ], [ 1056, "absl/strings/str_format.h" ], [ 1057, "absl/strings/str_format.h" ], [ 1093, "absl/strings/str_join.h" ], [ 1094, "absl/strings/str_join.h" ], [ 1128, "absl/types/span.h" ], [ 1129, "absl/types/span.h" ], [ 1157, "llvm/ADT/SmallVector.h" ], [ 1158, "llvm/ADT/SmallVector.h" ], [ 1191, "llvm/ADT/bit.h" ], [ 1192, "llvm/ADT/bit.h" ], [ 1217, "llvm/Support/Casting.h" ], [ 1218, "llvm/Support/Casting.h" ], [ 1251, "llvm/Support/MathExtras.h" ], [ 1252, "llvm/Support/MathExtras.h" ], [ 1288, "llvm/Support/raw_ostream.h" ], [ 1289, "llvm/Support/raw_ostream.h" ], [ 1326, "mlir/Dialect/Arith/IR/Arith.h" ], [ 1327, "mlir/Dialect/Arith/IR/Arith.h" ], [ 1367, "mlir/Dialect/LLVMIR/NVVMDialect.h" ], [ 1368, "mlir/Dialect/LLVMIR/NVVMDialect.h" ], [ 1412, "mlir/Dialect/Tensor/IR/Tensor.h" ], [ 1413, "mlir/Dialect/Tensor/IR/Tensor.h" ], [ 1455, "mlir/IR/Attributes.h" ], [ 1456, "mlir/IR/Attributes.h" ], [ 1487, "mlir/IR/Builders.h" ], [ 1488, "mlir/IR/Builders.h" ], [ 1517, "mlir/IR/BuiltinTypeInterfaces.h" ], [ 1518, "mlir/IR/BuiltinTypeInterfaces.h" ], [ 1560, "mlir/IR/BuiltinTypes.h" ], [ 1561, "mlir/IR/BuiltinTypes.h" ], [ 1594, "mlir/IR/Location.h" ], [ 1595, "mlir/IR/Location.h" ], [ 1624, "mlir/IR/OperationSupport.h" ], [ 1625, "mlir/IR/OperationSupport.h" ], [ 1662, "mlir/IR/PatternMatch.h" ], [ 1663, "mlir/IR/PatternMatch.h" ], [ 1696, "mlir/IR/TypeUtilities.h" ], [ 1697, "mlir/IR/TypeUtilities.h" ], [ 1731, "mlir/IR/Types.h" ], [ 1732, "mlir/IR/Types.h" ], [ 1758, "mlir/IR/Value.h" ], [ 1759, "mlir/IR/Value.h" ], [ 1785, "mlir/Support/LLVM.h" ], [ 1786, "mlir/Support/LLVM.h" ], [ 1816, "stablehlo/dialect/StablehloOps.h" ], [ 1817, "stablehlo/dialect/StablehloOps.h" ], [ 1860, "xla/backends/gpu/codegen/triton/ir/triton_xla_ops.h" ], [ 1861, "xla/backends/gpu/codegen/triton/ir/triton_xla_ops.h" ], [ 1923, "xla/backends/gpu/codegen/triton/lowering_util.h" ], [ 1924, "xla/backends/gpu/codegen/triton/lowering_util.h" ], [ 1982, "xla/backends/gpu/runtime/all_reduce.h" ], [ 1983, "xla/backends/gpu/runtime/all_reduce.h" ], [ 2031, "xla/codegen/emitters/ir/xla_ops.h" ], [ 2032, "xla/codegen/emitters/ir/xla_ops.h" ], [ 2098, "xla/codegen/xtile/codegen/emitter_helpers.h" ], [ 2099, "xla/codegen/xtile/codegen/emitter_helpers.h" ], [ 2153, "xla/codegen/xtile/ir/xtile_ops.h" ], [ 2154, "xla/codegen/xtile/ir/xtile_ops.h" ], [ 2197, "xla/core/collectives/reduction_kind.h" ], [ 2198, "xla/core/collectives/reduction_kind.h" ], [ 2246, "xla/hlo/ir/hlo_casting_utils.h" ], [ 2247, "xla/hlo/ir/hlo_casting_utils.h" ], [ 2288, "xla/hlo/ir/hlo_instruction.h" ], [ 2289, "xla/hlo/ir/hlo_instruction.h" ], [ 2328, "xla/hlo/ir/hlo_instructions.h" ], [ 2329, "xla/hlo/ir/hlo_instructions.h" ], [ 2369, "xla/hlo/ir/hlo_opcode.h" ], [ 2370, "xla/hlo/ir/hlo_opcode.h" ], [ 2404, "xla/layout_util.h" ], [ 2405, "xla/layout_util.h" ], [ 2433, "xla/mlir/utils/type_util.h" ], [ 2434, "xla/mlir/utils/type_util.h" ], [ 2471, "xla/service/collective_ops_utils.h" ], [ 2472, "xla/service/collective_ops_utils.h" ], [ 2517, "xla/service/gpu/backend_configs.pb.h" ], [ 2518, "xla/service/gpu/backend_configs.pb.h" ], [ 2565, "xla/service/gpu/gpu_constants.h" ], [ 2566, "xla/service/gpu/gpu_constants.h" ], [ 2608, "xla/service/gpu/ir_emission_utils.h" ], [ 2609, "xla/service/gpu/ir_emission_utils.h" ], [ 2655, "xla/service/gpu/launch_dimensions.h" ], [ 2656, "xla/service/gpu/launch_dimensions.h" ], [ 2702, "xla/shape.h" ], [ 2703, "xla/shape.h" ], [ 2725, "xla/shape_util.h" ], [ 2726, "xla/shape_util.h" ], [ 2753, "xla/status_macros.h" ], [ 2754, "xla/status_macros.h" ], [ 2784, "xla/stream_executor/device_description.h" ], [ 2785, "xla/stream_executor/device_description.h" ], [ 2836, "xla/stream_executor/gpu/all_reduce_kernel.h" ], [ 2837, "xla/stream_executor/gpu/all_reduce_kernel.h" ], [ 2891, "xla/tsl/platform/errors.h" ], [ 2892, "xla/tsl/platform/errors.h" ], [ 2928, "xla/tsl/platform/statusor.h" ], [ 2929, "xla/tsl/platform/statusor.h" ], [ 2967, "xla/util.h" ], [ 2968, "xla/util.h" ], [ 2989, "xla/xla_data.pb.h" ], [ 2990, "xla/xla_data.pb.h" ], [ 3018, "triton/Dialect/Triton/IR/Dialect.h" ], [ 3019, "triton/Dialect/Triton/IR/Dialect.h" ], [ 3064, "triton/Dialect/Triton/IR/Types.h" ], [ 3065, "triton/Dialect/Triton/IR/Types.h" ], [ 3108, "triton/Dialect/TritonGPU/IR/Attributes.h" ], [ 3109, "triton/Dialect/TritonGPU/IR/Attributes.h" ], [ 3160, "triton/Dialect/TritonGPU/IR/Dialect.h" ], [ 3161, "triton/Dialect/TritonGPU/IR/Dialect.h" ], [ 5098, "AllReduce op must have exactly one operand in order to be lowered \"\n \"to triton." ], [ 5098, "AllReduce op must have exactly one operand in order to be lowered " ], [ 5099, "AllReduce op must have exactly one operand in order to be lowered " ], [ 5175, "to triton." ], [ 5176, "to triton." ], [ 5657, "Could not convert operand type to a valid PrimitiveType.\"\n \"Operand Type: %s" ], [ 5657, "Could not convert operand type to a valid PrimitiveType." ], [ 5658, "Could not convert operand type to a valid PrimitiveType." ], [ 5724, "Operand Type: %s" ], [ 5725, "Operand Type: %s" ], [ 5913, "AllReduce op must be in an XTile entry function in order to be \"\n \"lowered to triton." ], [ 5913, "AllReduce op must be in an XTile entry function in order to be " ], [ 5914, "AllReduce op must be in an XTile entry function in order to be " ], [ 5987, "lowered to triton." ], [ 5988, "lowered to triton." ], [ 6606, "AllReduce op must have " ], [ 6607, "AllReduce op must have " ], [ 6673, " arguments in order to \"\n \"be lowered to triton, but it has " ], [ 6673, " arguments in order to " ], [ 6674, " arguments in order to " ], [ 6720, "be lowered to triton, but it has " ], [ 6721, "be lowered to triton, but it has " ], [ 7698, "AllReduce op must have an extract tile op as operand in order to be \"\n \"lowered to triton." ], [ 7698, "AllReduce op must have an extract tile op as operand in order to be " ], [ 7699, "AllReduce op must have an extract tile op as operand in order to be " ], [ 7777, "lowered to triton." ], [ 7778, "lowered to triton." ], [ 8880, "Skipping all-reduce codegen because \"\n \"xla_gpu_unsupported_use_all_reduce_one_shot_kernel is disabled." ], [ 8880, "Skipping all-reduce codegen because " ], [ 8881, "Skipping all-reduce codegen because " ], [ 8930, "xla_gpu_unsupported_use_all_reduce_one_shot_kernel is disabled." ], [ 8931, "xla_gpu_unsupported_use_all_reduce_one_shot_kernel is disabled." ], [ 9102, "Replica groups are empty for " ], [ 9103, "Replica groups are empty for " ], [ 9171, ". Codegen will not be supported." ], [ 9172, ". Codegen will not be supported." ], [ 9406, "Number of devices is not a power of 2 for " ], [ 9407, "Number of devices is not a power of 2 for " ], [ 9488, ". Codegen will not be supported." ], [ 9489, ". Codegen will not be supported." ], [ 10514, "Codegen forall-reduce is only supported for small inputs." ], [ 10515, "Codegen forall-reduce is only supported for small inputs." ], [ 10628, " \u003c" ], [ 10629, " \u003c" ], [ 11544, "Collective codegen requires compute capability greater than 9. \"\n \"Codegen will not be supported." ], [ 11544, "Collective codegen requires compute capability greater than 9. " ], [ 11545, "Collective codegen requires compute capability greater than 9. " ], [ 11625, "Codegen will not be supported." ], [ 11626, "Codegen will not be supported." ], [ 11839, "AllReduceInfo is not available for " ], [ 11840, "AllReduceInfo is not available for " ], [ 11914, ". Codegen will not be supported." ], [ 11915, ". Codegen will not be supported." ], [ 12990, "Two-shot all-reduce linear_tile_size(" ], [ 12991, "Two-shot all-reduce linear_tile_size(" ], [ 13065, ") % num_devices(" ], [ 13066, ") % num_devices(" ], [ 13131, ") != 0. Codegen will not be supported." ], [ 13132, ") != 0. Codegen will not be supported." ], [ 13215, "Block level fusion config for " ], [ 13216, "Block level fusion config for " ], [ 13273, " " ], [ 13274, " " ], [ 15879, "AllReduce computation must only contain binary operations." ], [ 15880, "AllReduce computation must only contain binary operations." ], [ 16076, "AllReduce computation must only contain operations with one \"\n \"result." ], [ 16076, "AllReduce computation must only contain operations with one " ], [ 16077, "AllReduce computation must only contain operations with one " ], [ 16151, "result." ], [ 16152, "result." ], [ 16884, "AllReduce op must have exactly one relevant operation in order to \"\n \"be lowered to triton." ], [ 16884, "AllReduce op must have exactly one relevant operation in order to " ], [ 16885, "AllReduce op must have exactly one relevant operation in order to " ], [ 16961, "be lowered to triton." ], [ 16962, "be lowered to triton." ], [ 17323, "Failed to initialize AllReduceEmitter: " ], [ 17324, "Failed to initialize AllReduceEmitter: " ], [ 17777, "Multimem all-reduce is not yet supported for codegeneration." ], [ 17778, "Multimem all-reduce is not yet supported for codegeneration." ], [ 18457, "Failed to populate reduction computation." ], [ 18458, "Failed to populate reduction computation." ], [ 20600, "Input tile shape is not perfectly divisible by subtile shape." ], [ 20601, "Input tile shape is not perfectly divisible by subtile shape." ], [ 20677, "Input tile shape: " ], [ 20678, "Input tile shape: " ], [ 20731, "Subtile shape: " ], [ 20732, "Subtile shape: " ], [ 20782, "," ], [ 20783, "," ], [ 38761, "Failed to emit copy to symmetric" ], [ 38762, "Failed to emit copy to symmetric" ], [ 39020, "Failed to emit sync for one-shot" ], [ 39021, "Failed to emit sync for one-shot" ], [ 40051, "Failed to emit copy to symmetric" ], [ 40052, "Failed to emit copy to symmetric" ], [ 40294, "Failed to emit sync for shot1" ], [ 40295, "Failed to emit sync for shot1" ], [ 41095, "Failed to emit copy result to symmetric for two-shot" ], [ 41096, "Failed to emit copy result to symmetric for two-shot" ], [ 41590, "Failed to emit sync for shot2" ], [ 41591, "Failed to emit sync for shot2" ], [ 41861, "Failed to emit gathered load: " ], [ 41862, "Failed to emit gathered load: " ], [ 43358, "num_blocks must be positive. Was " ], [ 43359, "num_blocks must be positive. Was " ], [ 45445, "No block level fusion config calculated for collective: " ], [ 45446, "No block level fusion config calculated for collective: " ], [ 45559, ". Not using Triton collective fusion." ], [ 45560, ". Not using Triton collective fusion." ], [ 48153, "Failed to create AllReduceEmitterContext: " ], [ 48154, "Failed to create AllReduceEmitterContext: " ], [ 48313, "Failed to create AllReduceEmitterContext: " ], [ 48314, "Failed to create AllReduceEmitterContext: " ], [ 48438, "AllReduceEmitter::Emit using strategy: " ], [ 48439, "AllReduceEmitter::Emit using strategy: " ] ], "sz": 48612, "ts": [ { "c": 0.20000000298023224, "d": "output keyword", "e": [ "output" ], "i": "micro-behaviors/data/text/keywords::output", "l": 1 }, { "c": 0.75, "d": "Delphi assign file pattern", "e": [ "Assign(" ], "i": "metadata/lang/source::assignfile-pattern", "l": 2 } ], "sha": "4b3573da7322f502fc03d2195d635f6a84bb7fbf3fcf1c765e390ebd2fe7fdff", "path": "/data/samples/good/repos/tensorflow/third_party/xla/xla/backends/gpu/codegen/triton/collective_emitter.cc", "type": "c" } ], "tv": "f6eaa" } }