Development repository for the Triton language and compiler
This report summarizes scanner evidence for triton-lang/triton, a MLIR project with 20,340 GitHub stars. SynthScan examined 431,092 lines of code across 1314 source files, recording 4254 pattern matches distributed across 25 syntactic categories. The legacy adjusted pattern signal is 11.0 (Moderate pattern signal); it is not an authorship estimate.
Use the findings to ask whether maintainers can explain intent, verify dependencies, simplify unnecessary scaffolding, and focus review effort. Risk Payload v1 adds ownership, supply-chain, over-engineering, and review-effort vectors when available. These are prompts for human verification, not probabilities, security findings, or measured maintenance costs.
Use each vector to guide manual verification of the underlying findings.
Can maintainers explain intent, assumptions, and trade-offs? 3117 finding(s).
Are imports declared and package/API names verified? 628 finding(s).
Are wrappers, guards, and repeated structures necessary? 2345 finding(s).
Which changes need focused human review? 4254 finding(s).
This chart tracks the legacy adjusted pattern score across scan runs. Changes can reflect source changes, scan coverage, or detector/scoring versions; they do not measure AI adoption or prove that a team refactored code. Compare runs only when their scoring versions are compatible. The dashed line tracks the raw pattern-hit count.
Groups findings by configured severity weight. Severity is not a confidence estimate, security rating, or proof that a package is missing. Verify imports against the project's manifests and inspect each finding in its local design context.
This horizontal bar chart shows where configured pattern scores accumulate across directories. Treat the bars as locations to inspect, not as evidence of authorship or code quality. Use the file-level findings to decide whether a directory needs a focused ownership, dependency, complexity, or review check.
The scanner identified 4254 distinct pattern matches across 25 syntactic categories. Each entry below represents a discrete location in the source code where the engine recorded a detector match to inspect in context. Expand any category row to inspect the individual file paths, line numbers, code snippets, and the lexical context (CODE, COMMENT, or STRING) in which each match was detected.
Reading the findings table: The Severity column indicates the configured finding weight (CRITICAL / HIGH / MEDIUM / LOW). The Context column identifies whether the match occurred inside executable code, an inline comment, or a string literal — legacy comment-context matches receive a ×1.5 score weight. The ⚡ bolt icon marks clustered matches: three or more patterns within a 10-line window, each receiving an additional ×1.5 density multiplier under the legacy scoring model.
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | setup.py | 577 | def get_triton_version_suffix(): | CODE |
| LOW | python/conftest.py | 60 | def fresh_knobs_including_libraries(): | CODE |
| LOW | python/build_helpers.py | 76 | def find_therock_rocm_include_dir() -> Optional[str]: | CODE |
| LOW | python/build_helpers.py | 161 | def _download_file_with_urllib(url: str, path: str, label: str): | CODE |
| LOW | python/build_helpers.py | 454 | def get_amd_codegen_package_info(helper_args: BuildHelperArgs): | CODE |
| LOW | python/build_helpers.py | 490 | def _get_thirdparty_package_cmake_vars(package: Package, helper_args: BuildHelperArgs): | CODE |
| LOW | python/build_helpers.py | 524 | def get_thirdparty_cmake_vars(packages: list[str], helper_args: BuildHelperArgs): | CODE |
| LOW | python/build_helpers.py | 544 | def write_thirdparty_cmake_vars(output: str, packages: list[str], helper_args: BuildHelperArgs): | CODE |
| LOW | python/build_helpers.py | 641 | def download_and_copy_amd_codegen(helper_args: BuildHelperArgs): | CODE |
| LOW | python/build_helpers.py | 729 | def get_nvidia_toolchain_packages(): | CODE |
| LOW | python/build_helpers.py | 803 | def download_and_copy_dependencies(helper_args: BuildHelperArgs): | CODE |
| LOW | python/test/unit/test_utils.py | 12 | def test_cdiv_symbolic_numerator(y): | CODE |
| LOW | python/test/unit/test_utils.py | 54 | def test_validate_block_shape_rejects_zero(): | CODE |
| LOW | python/test/unit/test_utils.py | 63 | def test_validate_block_shape_accepts_powers_of_two(): | CODE |
| LOW | python/test/unit/test_utils.py | 80 | def test_get_dram_gbps_memory_data_rate(monkeypatch, backend, arch, properties, expected): | CODE |
| LOW | python/test/unit/test_debug.py | 26 | def _run_device_assert_barrier(device): | CODE |
| LOW | python/test/unit/test_debug.py | 40 | def _run_expect_zero_device_assert(device): | CODE |
| LOW | python/test/unit/test_debug.py | 56 | def test_expect_zero_device_assert(device): | CODE |
| LOW | python/test/unit/test_debug.py | 78 | def test_device_assert_barrier(device): | CODE |
| LOW | python/test/unit/test_debug.py | 157 | def test_sanitize_int_add_overflow(x, y, x_dtype, y_dtype, debug, should_overflow, device): | CODE |
| LOW | python/test/unit/test_debug.py | 173 | def test_sanitize_int_mul_overflow(x, y, x_dtype, y_dtype, debug, should_overflow, device): | CODE |
| LOW | python/test/unit/test_debug.py | 188 | def test_sanitize_int_sub_overflow(x, y, x_dtype, y_dtype, debug, should_overflow, device): | CODE |
| LOW | python/test/unit/test_perf_warning.py | 11 | def enable_diagnostics_context(value): | CODE |
| LOW | python/test/unit/test_perf_warning.py | 95 | def test_remark_vectorization(capfd, fresh_triton_cache): | CODE |
| LOW | python/test/unit/test_perf_warning.py | 159 | def test_remark_swp_op_before_operands(capfd, fresh_triton_cache): | CODE |
| LOW | python/test/unit/test_stages_inspection.py | 11 | def test_instrumentation_mode(): | CODE |
| LOW | python/test/unit/test_stages_inspection.py | 18 | def test_dynamic_instrumentation_and_dialect_loading(): | CODE |
| LOW | python/test/unit/test_knobs.py | 72 | def _register_pressure_scheduler_kernel(): | CODE |
| LOW | python/test/unit/test_knobs.py | 101 | def test_nvidia_register_pressure_scheduler_hook(link_hip_first, fresh_triton_cache): | CODE |
| LOW | python/test/unit/test_knobs.py | 115 | def test_nvidia_register_pressure_scheduler_concurrent(): | CODE |
| LOW | python/test/unit/test_knobs.py | 132 | def _nvidia_short_pointer_kernel(): | CODE |
| LOW | python/test/unit/test_knobs.py | 153 | def test_nvidia_short_pointer_option(link_hip_first, fresh_triton_cache): | STRING |
| LOW | python/test/unit/test_knobs.py | 175 | def _compile_empty_hip_kernel(arch, warp_size): | STRING |
| LOW | python/test/unit/test_knobs.py | 205 | def test_amd_llvm_options_concurrent(): | STRING |
| LOW | python/test/unit/test_knobs.py | 280 | def test_amd_codegen_path_override(fresh_knobs, monkeypatch): | STRING |
| LOW | python/test/unit/test_knobs.py | 292 | def test_amd_codegen_revision_invalidates_backend_hash(fresh_knobs, monkeypatch): | STRING |
| LOW | python/test/unit/test_knobs.py | 307 | def test_amd_codegen_inlines_functions_from_bitcode(fresh_knobs, monkeypatch): | STRING |
| LOW | python/test/unit/test_knobs.py | 347 | def test_amd_codegen_options_restored(option, capfd, fresh_knobs): | STRING |
| LOW | python/test/unit/test_knobs.py | 362 | def test_amd_codegen_reports_invalid_llvm_ir(fresh_knobs): | STRING |
| LOW | python/test/unit/test_knobs.py | 372 | def test_amd_codegen_assembles_object_with_private_llvm(fresh_knobs): | STRING |
| LOW | python/test/unit/test_knobs.py | 391 | def test_amd_codegen_respects_mir_dump_knob(enabled, fresh_knobs, monkeypatch, tmp_path): | STRING |
| LOW | python/test/unit/test_knobs.py | 414 | def test_amd_codegen_preserves_mir_replacement(fresh_knobs, monkeypatch): | STRING |
| LOW | python/test/unit/tools/test_aot.py | 326 | def compile_aot_kernel_no_specialization(dir, kernel_path, dtype, BM, BN, BK, target=None): | STRING |
| LOW | python/test/unit/tools/test_aot.py | 371 | def generate_matmul_test_data(dir, M, N, K): | STRING |
| LOW | python/test/unit/tools/test_aot.py | 396 | def test_linker_rejects_invalid_algo_id(): | STRING |
| LOW | python/test/unit/tools/test_aot.py | 462 | def test_compile_link_matmul_no_specialization(): | STRING |
| LOW | python/test/unit/tools/test_aot.py | 529 | def test_launcher_has_no_available_kernel(): | STRING |
| LOW | python/test/unit/tools/test_aot.py | 567 | def test_compile_link_autotune_matmul(): | STRING |
| LOW | python/test/unit/tools/test_aot.py | 652 | def test_aot_target_parsing_with_explicit_target(): | STRING |
| LOW | python/test/unit/tools/test_linear_layout.py | 36 | def test_operator_mul_identity(): | CODE |
| LOW | python/test/unit/tools/test_linear_layout.py | 42 | def test_operator_mul_disjoint_dims(): | CODE |
| LOW⚡ | python/test/unit/tools/test_linear_layout.py | 75 | def test_get_matrix_view_identity(): | CODE |
| LOW⚡ | python/test/unit/tools/test_linear_layout.py | 83 | def test_get_matrix_view_strided(): | CODE |
| LOW⚡ | python/test/unit/tools/test_linear_layout.py | 92 | def test_get_matrix_view_from_bases(): | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 45 | def test_slice_kernel_basic_module_slicing(tmp_path): | CODE |
| LOW⚡ | python/test/unit/tools/test_slice_kernel.py | 759 | def test_slice_kernel_treats_assign_targets_as_locals(tmp_path): | STRING |
| LOW | python/test/unit/tools/test_slice_kernel.py | 121 | def test_slice_kernel_does_not_treat_site_packages_as_stdlib(tmp_path, monkeypatch): | STRING |
| LOW | python/test/unit/tools/test_slice_kernel.py | 167 | def test_slice_kernel_does_not_treat_base_python_site_packages_as_stdlib_in_venv(tmp_path, monkeypatch): | STRING |
| LOW | python/test/unit/tools/test_slice_kernel.py | 215 | def test_slice_kernel_preserves_gluon_builtin_sugar(tmp_path): | STRING |
| LOW | python/test/unit/tools/test_slice_kernel.py | 239 | def test_slice_kernel_supports_injected_decorator_matchers(tmp_path): | STRING |
| 1738 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | python/test/unit/cuda/test_libdevice_cuda.py | 13 | # ----------------------- | COMMENT |
| MEDIUM | python/test/unit/cuda/test_libdevice_cuda.py | 15 | # ----------------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_standard.py | 12 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_standard.py | 14 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_standard.py | 84 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_standard.py | 86 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_standard.py | 124 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_standard.py | 126 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_standard.py | 224 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_standard.py | 226 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 374 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 376 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 590 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 656 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 658 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 684 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 686 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 716 | # ---------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 718 | # ---------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 765 | # ------------------ | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 767 | # ------------------ | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 809 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 811 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 959 | # ---------------------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 961 | # ---------------------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 975 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 977 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1069 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1071 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1103 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1105 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1426 | # ------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1428 | # ------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1515 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1517 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1562 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1564 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1596 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1598 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1623 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1625 | # ---------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1806 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 1808 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 2549 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 2551 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 3156 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 3158 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 3866 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 3868 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 3935 | # ------------------------ | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 3937 | # ------------------------ | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 4171 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 4173 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 4288 | # --------------- | COMMENT |
| MEDIUM | python/test/unit/language/test_core.py | 4290 | # --------------- | COMMENT |
| MEDIUM⚡ | python/test/unit/language/test_core.py | 5803 | # --------------- | STRING |
| MEDIUM⚡ | python/test/unit/language/test_core.py | 5805 | # --------------- | STRING |
| MEDIUM⚡ | python/test/unit/language/test_core.py | 6020 | # -------------------- | STRING |
| MEDIUM⚡ | python/test/unit/language/test_core.py | 6022 | # -------------------- | STRING |
| MEDIUM⚡ | python/test/unit/language/test_core.py | 6260 | # ------------- | STRING |
| 318 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | unittest/Tools/LLVMOptionsTest.cpp | 1 | #include "triton/Tools/LLVMOptions.h" | COMMENT |
| LOW | unittest/Analysis/UtilityTest.cpp | 1 | #include "triton/Dialect/Triton/IR/Utility.h" | COMMENT |
| LOW | unittest/Dialect/TritonGPU/SwizzleTest.cpp | 1 | #include "triton/Dialect/Triton/IR/Dialect.h" | COMMENT |
| LOW | unittest/Dialect/TritonGPU/SwizzleTest.cpp | 241 | int phaseIdx; | COMMENT |
| LOW | unittest/Dialect/TritonGPU/SwizzleTest.cpp | 261 | } else { | COMMENT |
| LOW | …test/Dialect/TritonGPU/LinearLayoutConversionsTest.cpp | 1 | #include "triton/Dialect/TritonGPU/IR/LinearLayoutConversions.h" | COMMENT |
| LOW | …test/Dialect/TritonGPU/LinearLayoutConversionsTest.cpp | 3581 | // | COMMENT |
| LOW | test/lib/Analysis/TestMembar.cpp | 1 | #include "../third_party/nvidia/include/TritonNVIDIAGPUToLLVM/Utility.h" | COMMENT |
| LOW | bin/triton-llvm-opt.cpp | 1 | /// Trimmed down clone of llvm opt to be able to test triton custom llvm ir | COMMENT |
| LOW | bin/triton-tensor-layout.cpp | 1 | #include "RegisterTritonDialects.h" | COMMENT |
| LOW | bin/triton-tensor-layout.cpp | 21 | // | COMMENT |
| LOW | bin/RegisterTritonDialects.h | 1 | #pragma once | COMMENT |
| LOW | bin/RegisterTritonDialects.h | 21 | // Below headers will allow registration to ROCm passes | COMMENT |
| LOW | python/test/unit/language/test_line_info.py | 301 | # CHECK: %offsets_1 = tt.splat %offset_0 : i32 -> tensor<16xi32> loc(#loc15) | COMMENT |
| LOW | python/test/unit/language/test_line_info.py | 521 | # CHECK: #loc = loc("{{.*}}":@DEF_LINE@:5) | COMMENT |
| LOW | python/test/unit/language/test_line_info.py | 541 | # CHECK: } loc(#loc) | COMMENT |
| LOW | python/test/gluon/test_frontend.py | 881 | _TENSOR_TUPLE_288 = TensorTuple((288, ), _TENSOR_TUPLE_LAYOUT, minimum_padding_size=64) | COMMENT |
| LOW | python/test/gluon/test_frontend.py | 901 | COMMENT | |
| LOW | python/test/gluon/test_frontend.py | 941 | # CHECK: [[VALUE:%.*]] = tt.make_range {end = 128 : i32, start = 0 : i32} : tensor<128xi32, [[LAYOUT]]> | COMMENT |
| LOW | python/test/gluon/test_frontend.py | 1081 | # CHECK-NEXT: [[A:%.*]] = tt.make_range {end = 1 : i32, start = 0 : i32} | COMMENT |
| LOW | python/test/gluon/test_frontend.py | 1141 | @filecheck_test | COMMENT |
| LOW | python/test/gluon/test_core.py | 1701 | r"mbarrier\.arrive\.shared::cluster(?:\.multicast::cluster::32b)?\.b64\s+_,\s*\[[^\]]+\](?:,\s*(\d+))?(?:,\s*%r\ | COMMENT |
| LOW | python/test/gluon/test_core.py | 3341 | # Check that we allocated only 128 columns of TMEM. | COMMENT |
| LOW | python/test/backend/test_mir_stage.py | 61 | # so we key instructions by (opcode, ordinal-within-function) instead -- a | COMMENT |
| LOW | python/test/backend/test_mir_stage.py | 321 | COMMENT | |
| LOW | python/triton/knobs.py | 561 | roctx_lib_path: env_opt_str = env_opt_str("TRITON_ROCTX_LIB_PATH") | COMMENT |
| LOW | python/triton/experimental/gsan/src/GSan.h | 1 | #pragma once | COMMENT |
| LOW | …iton/experimental/gluon/language/amd/cdna5/_layouts.py | 181 | assert block_m % slice_m == 0 and block_n % slice_n == 0, \ | COMMENT |
| LOW | …iton/experimental/gluon/language/amd/cdna5/_layouts.py | 201 | # M=2: w0 w1 <- first tile computed by w1 (reg=0) | COMMENT |
| LOW | …iton/experimental/gluon/language/amd/cdna5/_layouts.py | 221 | # efficient. | COMMENT |
| LOW | python/triton/tools/disasm.py | 1 | # MIT License | COMMENT |
| LOW | …triton/tools/triton_to_gluon_translator/ordered_set.py | 1 | """ | COMMENT |
| LOW | python/examples/gluon/06-overlapping-accumulator.py | 61 | # With 512 TMEM columns, double buffered BLOCK_N=256 accumulators fully consume TMEM | COMMENT |
| LOW | python/examples/gluon/06-overlapping-accumulator.py | 81 | # 3) New ping-pong style barrier guards the overlapped columns. | COMMENT |
| LOW | python/examples/gluon/06-overlapping-accumulator.py | 101 | # | COMMENT |
| LOW | python/examples/gluon/02-conv-dgrad.py | 61 | # | COMMENT |
| LOW | python/examples/gluon/02-conv-fprop.py | 61 | COMMENT | |
| LOW | python/examples/gluon/02-conv-fprop.py | 481 | output = torch.empty((N, out_h, out_w, Co), device=input_tensor.device, dtype=TORCH_GEMM_DTYPE) | COMMENT |
| LOW | python/examples/gluon/02-conv-wgrad.py | 61 | # grad_W[Co, R*S*Ci] = grad_out[M, Co]^T @ im2col(input)[M, R*S*Ci] | COMMENT |
| LOW | python/triton_kernels/tests/test_reduce.py | 221 | COMMENT | |
| LOW | python/triton_kernels/triton_kernels/matmul.py | 61 | COMMENT | |
| LOW | python/triton_kernels/triton_kernels/matmul.py | 221 | assert out_ndim >= storage.data.ndim | COMMENT |
| LOW | …_kernels/tensor_details/layout_details/hopper_value.py | 201 | sizes += [a, b, c, s, d] | COMMENT |
| LOW | …riton_kernels/tensor_details/layout_details/strided.py | 1 | from dataclasses import dataclass | COMMENT |
| LOW | …n_kernels/tensor_details/layout_details/torch_utils.py | 1 | import torch | COMMENT |
| LOW | …n_kernels/tensor_details/layout_details/torch_utils.py | 21 | # if not is_fp4: | COMMENT |
| LOW | python/tutorials/02-fused-softmax.py | 61 | COMMENT | |
| LOW | python/tutorials/02-fused-softmax.py | 181 | COMMENT | |
| LOW | python/tutorials/08-grouped-gemm.py | 1 | """ | COMMENT |
| LOW | python/tutorials/04-low-memory-dropout.py | 21 | # The *dropout* operator was first introduced in [SRIVASTAVA2014]_ as a way to improve the performance | COMMENT |
| LOW | python/tutorials/04-low-memory-dropout.py | 81 | output = dropout(x, x_keep=x_keep, p=p) | COMMENT |
| LOW | python/tutorials/04-low-memory-dropout.py | 161 | COMMENT | |
| LOW | python/tutorials/03-matrix-multiplication.py | 21 | # ----------- | COMMENT |
| LOW | python/tutorials/03-matrix-multiplication.py | 41 | # for k in range(0, K, BLOCK_SIZE_K): | COMMENT |
| LOW | python/tutorials/03-matrix-multiplication.py | 61 | # For a row-major 2D tensor :code:`X`, the memory location of :code:`X[i, j]` is given | COMMENT |
| LOW | python/tutorials/03-matrix-multiplication.py | 81 | # offs_k = tl.arange(0, BLOCK_SIZE_K) | COMMENT |
| LOW | python/tutorials/03-matrix-multiplication.py | 101 | # | COMMENT |
| LOW | python/tutorials/03-matrix-multiplication.py | 121 | # # Number of programs ids along the N axis | COMMENT |
| LOW | python/tutorials/03-matrix-multiplication.py | 281 | # We will advance this pointer as we move in the K direction | COMMENT |
| LOW | python/tutorials/05-layer-norm.py | 21 | # It takes a vector :math:`x` as input and produces a vector :math:`y` of the same shape as output. | COMMENT |
| 878 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | python/test/unit/runtime/test_build.py | 1 | CODE | |
| LOW | python/test/unit/language/test_annotations.py | 1 | CODE | |
| LOW | python/test/gsan/test_write_once.py | 1 | CODE | |
| LOW | python/test/gsan/test_allocator.py | 1 | CODE | |
| LOW | python/test/gsan/test_gsan_failures.py | 1 | CODE | |
| LOW | python/test/gsan/test_symmetric_memory.py | 1 | CODE | |
| LOW | python/test/gsan/test_gsan.py | 1 | CODE | |
| LOW | python/test/gsan/test_write_once_failures.py | 1 | CODE | |
| LOW | python/triton/_instrumentation.py | 1 | CODE | |
| LOW | python/triton/knobs.py | 1 | CODE | |
| LOW | python/triton/__init__.py | 8 | CODE | |
| LOW | python/triton/__init__.py | 8 | CODE | |
| LOW | python/triton/__init__.py | 8 | CODE | |
| LOW | python/triton/__init__.py | 8 | CODE | |
| LOW | python/triton/__init__.py | 8 | CODE | |
| LOW | python/triton/__init__.py | 8 | CODE | |
| LOW | python/triton/__init__.py | 8 | CODE | |
| LOW | python/triton/__init__.py | 8 | CODE | |
| LOW | python/triton/__init__.py | 8 | CODE | |
| LOW | python/triton/__init__.py | 8 | CODE | |
| LOW | python/triton/__init__.py | 20 | CODE | |
| LOW | python/triton/__init__.py | 21 | CODE | |
| LOW | python/triton/__init__.py | 21 | CODE | |
| LOW | python/triton/__init__.py | 22 | CODE | |
| LOW | python/triton/__init__.py | 22 | CODE | |
| LOW | python/triton/__init__.py | 23 | CODE | |
| LOW | python/triton/__init__.py | 24 | CODE | |
| LOW | python/triton/__init__.py | 27 | CODE | |
| LOW | python/triton/__init__.py | 28 | CODE | |
| LOW | python/triton/_utils.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/graph.py | 7 | CODE | |
| LOW | python/triton/experimental/gsan/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/_testing_utils.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/_allocator.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/symmetric_memory.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/_testing.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/_stream_sync.py | 1 | CODE | |
| LOW | python/triton/experimental/gsan/_utils.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/__init__.py | 2 | CODE | |
| LOW | python/triton/experimental/gluon/__init__.py | 2 | CODE | |
| LOW | python/triton/experimental/gluon/__init__.py | 3 | CODE | |
| LOW | python/triton/experimental/gluon/__init__.py | 4 | CODE | |
| LOW | python/triton/experimental/gluon/_runtime.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/_runtime.py | 4 | CODE | |
| LOW | python/triton/experimental/gluon/language/_semantic.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/language/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/language/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/language/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/language/__init__.py | 1 | CODE | |
| LOW | python/triton/experimental/gluon/language/__init__.py | 1 | CODE | |
| 527 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| CRITICAL | python/test/unit/plugins/test_plugin.py | 36 | triton._C.libtriton.passes.plugin.extend_with(LIB) | CODE |
| CRITICAL | python/test/unit/plugins/test_custom_ops.py | 67 | triton._C.libtriton.ir.builder.extend_with(LIB) | CODE |
| CRITICAL | python/test/unit/plugins/override_helper.py | 15 | triton._C.libtriton.passes.plugin.extend_with(LIB) | CODE |
| CRITICAL | python/test/unit/language/test_core.py | 5181 | if not is_interpreter() and triton.runtime.driver.active.utils.get_device_properties( | CODE |
| CRITICAL | python/test/unit/language/test_matmul.py | 131 | shared_mem_avail = triton.runtime.driver.active.utils.get_device_properties(0)["max_shared_mem"] | CODE |
| CRITICAL | python/test/gsan/test_gsan.py | 1962 | properties = triton.runtime.driver.active.utils.get_device_properties(torch.cuda.current_device()) | CODE |
| CRITICAL | python/test/gluon/test_frontend.py | 1730 | %0 = tt.call @triton.experimental.gluon.language.nvidia.ampere.mbarrier.allocate_mbarrier__cNone_cTrue() : () -> !tt | CODE |
| CRITICAL | python/test/gluon/test_frontend.py | 1735 | tt.func private @triton.experimental.gluon.language.nvidia.ampere.mbarrier.allocate_mbarrier__cNone_cTrue() -> !ttg.me | CODE |
| CRITICAL | python/test/gluon/test_frontend.py | 4840 | %13 = tt.call @triton.experimental.gluon.language._standard.zeros__Tc1T_cfp16_cAL() : () -> tensor<1xf16, #gluon.aut | CODE |
| CRITICAL | python/test/gluon/test_frontend.py | 4868 | tt.func private @triton.experimental.gluon.language._standard.zeros__Tc1T_cfp16_cAL() -> tensor<1xf16, #gluon.auto_enc | CODE |
| CRITICAL | python/test/gluon/test_frontend.py | 4899 | %1 = tt.call @triton.experimental.gluon.language._standard.zeros__Tc1T_cbf16_cAL() : () -> tensor<1xbf16, #gluon.aut | CODE |
| CRITICAL | python/test/gluon/test_frontend.py | 4917 | tt.func private @triton.experimental.gluon.language._standard.zeros__Tc1T_cbf16_cAL() -> tensor<1xbf16, #gluon.auto_en | CODE |
| CRITICAL | python/triton/_internal_testing.py | 88 | return triton.runtime.driver.active.utils.cuOccupancyMaxActiveClusters(kernel.function, kernel.metadata.shared, | CODE |
| CRITICAL | python/triton/language/core.py | 3432 | triton.runtime.driver.active.utils.set_printf_fifo_size(size_bytes) | STRING |
| CRITICAL | python/triton_kernels/triton_kernels/target_info.py | 80 | return triton.runtime.driver.active.utils.get_device_properties(0)["multiprocessor_count"] | CODE |
| CRITICAL | third_party/proton/proton/hooks/instrumentation.py | 135 | max_shared_mem = triton.runtime.driver.active.utils.get_device_properties(device)["max_shared_mem"] | CODE |
| CRITICAL | third_party/proton/proton/hooks/instrumentation.py | 160 | arch = triton.runtime.driver.active.utils.get_device_properties(device)["arch"].split(":")[0] | CODE |
| CRITICAL⚡ | third_party/amd/python/test/test_amd_codegen.py | 44 | declare void @llvm.amdgcn.s.barrier.signal.var(ptr addrspace(3), i32) | CODE |
| CRITICAL⚡ | third_party/amd/python/test/test_amd_codegen.py | 48 | call void @llvm.amdgcn.s.barrier.signal.var(ptr addrspace(3) @bar, i32 4) | CODE |
| CRITICAL⚡ | third_party/amd/python/test/test_amd_codegen.py | 69 | declare void @llvm.amdgcn.s.barrier.signal.var(ptr addrspace(3), i32) | CODE |
| CRITICAL⚡ | third_party/amd/python/test/test_amd_codegen.py | 70 | declare i32 @llvm.amdgcn.s.get.named.barrier.state(ptr addrspace(3)) | CODE |
| CRITICAL | third_party/amd/python/test/test_amd_codegen.py | 96 | call void @llvm.amdgcn.s.barrier.signal.var(ptr addrspace(3) %bar, i32 4) | STRING |
| CRITICAL⚡ | third_party/amd/python/test/test_amd_codegen.py | 120 | call void @llvm.amdgcn.s.barrier.signal.var(ptr addrspace(3) %bar, i32 4) | STRING |
| CRITICAL⚡ | third_party/amd/python/test/test_amd_codegen.py | 122 | %state = call i32 @llvm.amdgcn.s.get.named.barrier.state(ptr addrspace(3) %bar) | STRING |
| CRITICAL | third_party/amd/python/test/test_amd_codegen.py | 152 | call void @llvm.amdgcn.s.barrier.signal.var(ptr addrspace(3) %frozen, i32 4) | STRING |
| CRITICAL | third_party/nvidia/backend/driver.py | 225 | cu_tensor_map = triton.runtime.driver.active.utils.fill_tma_descriptor_im2col( | CODE |
| CRITICAL | third_party/nvidia/backend/driver.py | 240 | cu_tensor_map = triton.runtime.driver.active.utils.fill_tma_descriptor_tiled( | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | setup.py | 129 | CODE | |
| LOW | python/build_helpers.py | 393 | CODE | |
| LOW | python/test/unit/tools/test_triton_to_gluon.py | 564 | CODE | |
| LOW | python/test/unit/cuda/test_libdevice_cuda.py | 109 | CODE | |
| LOW | python/test/unit/runtime/test_specialize.py | 42 | CODE | |
| LOW | python/test/unit/language/test_conversions.py | 15 | CODE | |
| LOW | python/test/unit/language/test_conversions.py | 277 | CODE | |
| LOW | python/test/unit/language/test_frontend.py | 212 | CODE | |
| LOW | python/test/unit/language/test_frontend.py | 261 | CODE | |
| LOW | python/test/unit/language/test_frontend.py | 1961 | CODE | |
| LOW | python/test/unit/language/test_frontend.py | 215 | CODE | |
| LOW | python/test/unit/language/test_frontend.py | 264 | CODE | |
| LOW | python/test/unit/language/test_core.py | 392 | CODE | |
| LOW | python/test/unit/language/test_core.py | 3512 | CODE | |
| LOW | python/test/unit/language/test_core.py | 3694 | CODE | |
| LOW | python/test/unit/language/test_core.py | 4293 | CODE | |
| LOW | python/test/unit/language/test_core.py | 4474 | CODE | |
| LOW | python/test/unit/language/test_core.py | 5414 | CODE | |
| LOW | python/test/unit/language/test_core.py | 5639 | CODE | |
| LOW | python/test/unit/language/test_core.py | 5810 | CODE | |
| LOW | python/test/unit/language/test_core.py | 6270 | CODE | |
| LOW | python/test/unit/language/test_core.py | 6743 | CODE | |
| LOW | python/test/unit/language/test_core.py | 8208 | CODE | |
| LOW | python/test/unit/language/test_core.py | 3516 | CODE | |
| LOW | python/test/unit/language/test_core.py | 6273 | CODE | |
| LOW | python/test/unit/language/test_core.py | 6746 | CODE | |
| LOW | python/test/unit/language/test_core.py | 8211 | CODE | |
| LOW | python/test/unit/language/test_pipeliner.py | 223 | CODE | |
| LOW | python/test/unit/language/test_subprocess.py | 57 | CODE | |
| LOW | python/test/unit/language/test_matmul.py | 524 | CODE | |
| LOW | python/test/unit/language/test_matmul.py | 663 | CODE | |
| LOW | python/test/unit/language/test_matmul.py | 712 | CODE | |
| LOW | python/test/unit/language/test_line_info.py | 164 | CODE | |
| LOW | python/test/unit/language/test_line_info.py | 212 | CODE | |
| LOW | python/test/unit/language/test_annotations.py | 62 | CODE | |
| LOW | python/test/unit/language/print_helper.py | 107 | CODE | |
| LOW | python/test/unit/language/test_tensor_descriptor.py | 1565 | CODE | |
| LOW | python/test/unit/language/test_tensor_descriptor.py | 1574 | CODE | |
| LOW | python/test/gsan/test_gsan_failures.py | 172 | CODE | |
| LOW | python/test/gsan/test_gsan_failures.py | 779 | CODE | |
| LOW | python/test/gsan/test_symmetric_memory.py | 436 | CODE | |
| LOW | python/test/gsan/test_symmetric_memory.py | 535 | CODE | |
| LOW | python/test/gsan/test_gsan.py | 1615 | CODE | |
| LOW | python/test/gsan/test_write_once_failures.py | 31 | CODE | |
| LOW | python/test/gsan/test_write_once_failures.py | 70 | CODE | |
| LOW | python/test/gluon/test_lowerings.py | 25 | CODE | |
| LOW | python/test/gluon/test_lowerings.py | 982 | CODE | |
| LOW | python/test/gluon/test_lowerings.py | 1105 | CODE | |
| LOW | python/test/gluon/test_lowerings.py | 1126 | CODE | |
| LOW | python/test/gluon/test_frontend.py | 1320 | CODE | |
| LOW | python/test/gluon/test_frontend.py | 5661 | CODE | |
| LOW | python/test/gluon/test_frontend.py | 5877 | CODE | |
| LOW | python/test/gluon/test_frontend.py | 5664 | CODE | |
| LOW | python/test/gluon/test_core.py | 890 | CODE | |
| LOW | python/test/gluon/test_core.py | 2415 | CODE | |
| LOW | python/test/gluon/test_core.py | 4057 | CODE | |
| LOW | python/test/gluon/test_core.py | 5120 | CODE | |
| LOW | python/test/gluon/test_core.py | 5435 | CODE | |
| LOW | python/test/gluon/test_core.py | 5547 | CODE | |
| LOW | python/test/gluon/test_core.py | 2496 | CODE | |
| 127 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW⚡ | README.md | 263 | # Step 1: Run the kernel once to dump kernel's IRs and ptx/amdgcn in $TRITON_DUMP_DIR | COMMENT |
| LOW⚡ | README.md | 264 | # Step 2: Copy $TRITON_DUMP_DIR/<kernel_hash> to $TRITON_OVERRIDE_DIR | COMMENT |
| LOW⚡ | README.md | 265 | # Step 3: Delete the stages that you do not want to override and modify the stage you do want to override | COMMENT |
| LOW⚡ | README.md | 266 | # Step 4: Run the kernel again to see the overridden result | COMMENT |
| LOW⚡ | python/tutorials/gluon/13-conv-im2col.py | 149 | # Step 1: Extract patches for each output position | COMMENT |
| LOW⚡ | python/tutorials/gluon/13-conv-im2col.py | 156 | # Step 2: Stack patches into im2col matrix A (M=4, K=4) | COMMENT |
| LOW⚡ | python/tutorials/gluon/13-conv-im2col.py | 167 | # Step 3: Reshape filter into weight matrix W (Co=1, K=4) | COMMENT |
| LOW⚡ | python/tutorials/gluon/13-conv-im2col.py | 171 | # Step 4: Output = A @ W^T | COMMENT |
| LOW | …ude/triton/Conversion/TritonGPUToLLVM/TargetInfoBase.h | 120 | // NOTE: This function is used for backend debug. DO NOT DELETE. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 533 | // Step 1: Right shift for 8 lanes. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 536 | // Step 2: Right shift for 4 lanes. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 539 | // Step 3: Right shift for 2 lanes. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 542 | // Step 4: Right shift for 1 lane. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 545 | // Step 5: Broadcast lane 15 of each row to all the lanes of its next row. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 548 | // Step 6: Broadcast lane 31 to lane 32-63. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 553 | // Step 7: Read the reduction result from lane 63 and broadcast with | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1127 | // Step 1: extract EM bits for elements 0,2,4,6 and 1,3,5,7 respectively. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1134 | // Step 2: convert fp4 to fp8 using LUT | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1144 | // Step 3: extract sign bits | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1148 | // Step 4: assemble 4 packed fp8 values w/ sign | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1152 | // Step 5: convert fp8 to fp32 | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1217 | // Step 1: extract EM bits for elements 0,2,4,6 and 1,3,5,7 respectively. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1225 | // Step 2: extract S bit for elements 0,2,4,6 and 1,3,5,7 | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1233 | // Step 3: Upcast elements 0,2,4,6 to 4 16-bit elements | COMMENT |
| LOW | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1262 | // Step 4: Upcast elements 1,3,5,7 to 4 16-bit elements | COMMENT |
| LOW | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1292 | // Step 5: Reorder 16-bit elements to be 0,1,2,3,4,5,6,7 | COMMENT |
| LOW | …_party/amd/lib/TritonAMDGPUToLLVM/DotOpToLLVM/MFMA.cpp | 559 | // Step 1: construct each kBase-element vector by | COMMENT |
| LOW | …_party/amd/lib/TritonAMDGPUToLLVM/DotOpToLLVM/MFMA.cpp | 571 | // Step 2: process rawElems based on element type | COMMENT |
| LOW | …_party/amd/lib/TritonAMDGPUToLLVM/DotOpToLLVM/MFMA.cpp | 591 | // Step 3: Insert the processed vals into the ValueTable | COMMENT |
| LOW | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1228 | // Step 1: collect all communications between producers and consumers. | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1239 | // Step 2: group channels | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1248 | // Step 3: reorder producer ops and the backward slices of the producer ops. | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1251 | // Step 4: find top-level ops that contain a channel, also create new ForOps | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1271 | // Step 5: Create buffers. An array of buffers for each channel. | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1279 | // Step 6: Lower the loads. Also add local copy ops for non-load | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1289 | // Step 7: Create tokens. A set of tokens for each group of channels for | COMMENT |
| LOW | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1300 | // Step 8: add async communication ops (ProducerAcquire etc). Also lower | COMMENT |
| LOW⚡ | …/hopper/lib/Transforms/WarpSpecialization/WSBuffer.cpp | 304 | // Step 1: Append accumCnts as forOp arguments. | COMMENT |
| LOW⚡ | …/hopper/lib/Transforms/WarpSpecialization/WSBuffer.cpp | 308 | // Step 2: Add accumCnts to yieldOp. | COMMENT |
| LOW⚡ | …/hopper/lib/Transforms/WarpSpecialization/WSBuffer.cpp | 317 | // Step 3: Create loop arguments for the new ForOp. | COMMENT |
| LOW⚡ | …/hopper/lib/Transforms/WarpSpecialization/WSBuffer.cpp | 326 | // Step 4: Create newForOp and take the region of the original forOp. | COMMENT |
| LOW⚡ | …/hopper/lib/Transforms/WarpSpecialization/WSBuffer.cpp | 332 | // Step 5: Replace forOp with newForOp. | COMMENT |
| LOW | lib/Dialect/TritonGPU/IR/LinearLayoutConversions.cpp | 1047 | // Step 3: Along the "register" dim, remove any all-zero bases. | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | python/triton/runtime/jit.py | 203 | # The purpose of this function is to visit everything in `arguments` | COMMENT |
| LOW⚡ | python/tutorials/gluon/13-conv-im2col.py | 149 | # Step 1: Extract patches for each output position | COMMENT |
| LOW⚡ | python/tutorials/gluon/13-conv-im2col.py | 156 | # Step 2: Stack patches into im2col matrix A (M=4, K=4) | COMMENT |
| LOW⚡ | python/tutorials/gluon/13-conv-im2col.py | 167 | # Step 3: Reshape filter into weight matrix W (Co=1, K=4) | COMMENT |
| LOW⚡ | python/tutorials/gluon/13-conv-im2col.py | 171 | # Step 4: Output = A @ W^T | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 533 | // Step 1: Right shift for 8 lanes. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 536 | // Step 2: Right shift for 4 lanes. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 539 | // Step 3: Right shift for 2 lanes. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 542 | // Step 4: Right shift for 1 lane. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 545 | // Step 5: Broadcast lane 15 of each row to all the lanes of its next row. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 548 | // Step 6: Broadcast lane 31 to lane 32-63. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/TargetInfo.cpp | 553 | // Step 7: Read the reduction result from lane 63 and broadcast with | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1127 | // Step 1: extract EM bits for elements 0,2,4,6 and 1,3,5,7 respectively. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1134 | // Step 2: convert fp4 to fp8 using LUT | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1144 | // Step 3: extract sign bits | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1148 | // Step 4: assemble 4 packed fp8 values w/ sign | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1152 | // Step 5: convert fp8 to fp32 | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1217 | // Step 1: extract EM bits for elements 0,2,4,6 and 1,3,5,7 respectively. | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1225 | // Step 2: extract S bit for elements 0,2,4,6 and 1,3,5,7 | COMMENT |
| LOW⚡ | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1233 | // Step 3: Upcast elements 0,2,4,6 to 4 16-bit elements | COMMENT |
| LOW | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1262 | // Step 4: Upcast elements 1,3,5,7 to 4 16-bit elements | COMMENT |
| LOW | third_party/amd/lib/TritonAMDGPUToLLVM/Utility.cpp | 1292 | // Step 5: Reorder 16-bit elements to be 0,1,2,3,4,5,6,7 | COMMENT |
| LOW | …_party/amd/lib/TritonAMDGPUToLLVM/DotOpToLLVM/MFMA.cpp | 559 | // Step 1: construct each kBase-element vector by | COMMENT |
| LOW | …_party/amd/lib/TritonAMDGPUToLLVM/DotOpToLLVM/MFMA.cpp | 571 | // Step 2: process rawElems based on element type | COMMENT |
| LOW | …_party/amd/lib/TritonAMDGPUToLLVM/DotOpToLLVM/MFMA.cpp | 591 | // Step 3: Insert the processed vals into the ValueTable | COMMENT |
| LOW | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1228 | // Step 1: collect all communications between producers and consumers. | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1239 | // Step 2: group channels | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1248 | // Step 3: reorder producer ops and the backward slices of the producer ops. | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1251 | // Step 4: find top-level ops that contain a channel, also create new ForOps | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1271 | // Step 5: Create buffers. An array of buffers for each channel. | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1279 | // Step 6: Lower the loads. Also add local copy ops for non-load | COMMENT |
| LOW⚡ | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1289 | // Step 7: Create tokens. A set of tokens for each group of channels for | COMMENT |
| LOW | …/lib/Transforms/WarpSpecialization/WSCodePartition.cpp | 1300 | // Step 8: add async communication ops (ProducerAcquire etc). Also lower | COMMENT |
| LOW⚡ | …/hopper/lib/Transforms/WarpSpecialization/WSBuffer.cpp | 304 | // Step 1: Append accumCnts as forOp arguments. | COMMENT |
| LOW⚡ | …/hopper/lib/Transforms/WarpSpecialization/WSBuffer.cpp | 308 | // Step 2: Add accumCnts to yieldOp. | COMMENT |
| LOW⚡ | …/hopper/lib/Transforms/WarpSpecialization/WSBuffer.cpp | 317 | // Step 3: Create loop arguments for the new ForOp. | COMMENT |
| LOW⚡ | …/hopper/lib/Transforms/WarpSpecialization/WSBuffer.cpp | 326 | // Step 4: Create newForOp and take the region of the original forOp. | COMMENT |
| LOW⚡ | …/hopper/lib/Transforms/WarpSpecialization/WSBuffer.cpp | 332 | // Step 5: Replace forOp with newForOp. | COMMENT |
| LOW | third_party/nvidia/lib/NVGPUToLLVM/NVGPUToLLVMPass.cpp | 630 | // TODO: pred will have a long liverange, we need to check if this is a | COMMENT |
| LOW | lib/Dialect/TritonGPU/IR/LinearLayoutConversions.cpp | 1047 | // Step 3: Along the "register" dim, remove any all-zero bases. | COMMENT |
| LOW | …t/TritonInstrument/Transforms/ConcurrencySanitizer.cpp | 1474 | // Op is writing to the buffer, we need to check if anything else | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | python/triton/__init__.py | 33 | __all__ = [ | CODE |
| LOW | python/triton/experimental/gsan/__init__.py | 3 | __all__ = [ | CODE |
| LOW | python/triton/experimental/gluon/__init__.py | 6 | __all__ = [ | CODE |
| LOW | python/triton/experimental/gluon/_runtime.py | 10 | __all__ = ["GluonASTSource", "GluonJITFunction", "constexpr_function", "jit"] | CODE |
| LOW | python/triton/experimental/gluon/language/_core.py | 55 | __all__ = [ | CODE |
| LOW | python/triton/experimental/gluon/language/_core.py | 378 | def _set_name(self, builder: ir.builder, name: str) -> None: | STRING |
| LOW | python/triton/experimental/gluon/language/amd/slice.py | 8 | __all__ = ["slice"] | CODE |
| LOW | …hon/triton/experimental/gluon/language/amd/__init__.py | 10 | __all__ = [ | CODE |
| LOW | python/triton/experimental/gluon/language/amd/hint.py | 3 | __all__ = ["sched_barrier"] | CODE |
| LOW | …hon/triton/experimental/gluon/language/amd/_layouts.py | 9 | __all__ = [ | CODE |
| LOW | …on/experimental/gluon/language/amd/cdna5/async_copy.py | 6 | __all__ = ["global_to_shared", "shared_to_global", "commit_group", "wait_group", "mbarrier_arrive"] | CODE |
| LOW | …iton/experimental/gluon/language/amd/cdna5/__init__.py | 15 | __all__ = [ | CODE |
| LOW | …on/triton/experimental/gluon/language/amd/cdna5/tdm.py | 14 | __all__ = [ | CODE |
| LOW | …on/triton/experimental/gluon/language/amd/cdna5/tdm.py | 66 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | …riton/experimental/gluon/language/amd/cdna5/cluster.py | 3 | __all__ = ["arrive", "wait"] | CODE |
| LOW | …iton/experimental/gluon/language/amd/cdna5/_layouts.py | 9 | __all__ = [ | CODE |
| LOW | …iton/experimental/gluon/language/amd/cdna5/mbarrier.py | 5 | __all__ = ["MBarrierLayout", "init", "wait", "arrive"] | CODE |
| LOW | …on/experimental/gluon/language/amd/cdna4/async_copy.py | 6 | __all__ = [ | CODE |
| LOW | …iton/experimental/gluon/language/amd/cdna4/__init__.py | 16 | __all__ = [ | CODE |
| LOW | …iton/experimental/gluon/language/amd/cdna3/__init__.py | 15 | __all__ = [ | CODE |
| LOW | …iton/experimental/gluon/language/amd/rdna4/__init__.py | 4 | __all__ = ["wmma"] | CODE |
| LOW | …iton/experimental/gluon/language/amd/rdna3/__init__.py | 4 | __all__ = ["wmma"] | CODE |
| LOW | …/triton/experimental/gluon/language/nvidia/__init__.py | 6 | __all__ = ["ampere", "blackwell", "hopper", "rubin"] | CODE |
| LOW | …n/experimental/gluon/language/nvidia/rubin/__init__.py | 38 | __all__ = [ | CODE |
| LOW | …n/experimental/gluon/language/nvidia/rubin/mbarrier.py | 11 | __all__ = [ | CODE |
| LOW | …/experimental/gluon/language/nvidia/hopper/__init__.py | 11 | __all__ = [ | CODE |
| LOW | …/experimental/gluon/language/nvidia/hopper/__init__.py | 94 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | …n/experimental/gluon/language/nvidia/hopper/cluster.py | 5 | __all__ = ["barrier"] | CODE |
| LOW | …riton/experimental/gluon/language/nvidia/hopper/tma.py | 13 | __all__ = [ | CODE |
| LOW | …riton/experimental/gluon/language/nvidia/hopper/tma.py | 122 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | …/experimental/gluon/language/nvidia/hopper/mbarrier.py | 4 | __all__ = [ | CODE |
| LOW | …perimental/gluon/language/nvidia/blackwell/__init__.py | 21 | __all__ = [ | CODE |
| LOW | …perimental/gluon/language/nvidia/blackwell/__init__.py | 327 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | …on/experimental/gluon/language/nvidia/blackwell/tma.py | 23 | __all__ = [ | CODE |
| LOW | …on/experimental/gluon/language/nvidia/blackwell/clc.py | 18 | __all__ = [ | CODE |
| LOW | …on/experimental/gluon/language/nvidia/blackwell/clc.py | 86 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | …xperimental/gluon/language/nvidia/ampere/async_copy.py | 5 | __all__ = [ | CODE |
| LOW | …/experimental/gluon/language/nvidia/ampere/__init__.py | 13 | __all__ = ["CachePolicy", "FractionalEvictionPolicy", "async_copy", "mbarrier", "mma_v2"] | CODE |
| LOW | …/experimental/gluon/language/nvidia/ampere/mbarrier.py | 8 | __all__ = ["allocate_mbarrier", "arrive", "init", "invalidate", "MBarrierLayout", "wait"] | CODE |
| LOW | …n/triton/experimental/gluon/language/extra/__init__.py | 3 | __all__ = ["libdevice"] | CODE |
| LOW | python/triton/experimental/gluon/amd/cdna5.py | 7 | __all__ = ["TensorDescriptor"] | CODE |
| LOW | python/triton/experimental/gluon/amd/__init__.py | 3 | __all__ = ["cdna5", "gfx1250"] | CODE |
| LOW | python/triton/experimental/gluon/nvidia/blackwell.py | 3 | __all__ = ["TensorDescriptor"] | CODE |
| LOW | python/triton/experimental/gluon/nvidia/__init__.py | 4 | __all__ = ["hopper", "blackwell"] | CODE |
| LOW | python/triton/experimental/gluon/nvidia/hopper.py | 7 | __all__ = ["TensorDescriptor", "TensorDescriptorIm2Col"] | CODE |
| LOW | …riton/tools/triton_to_gluon_translator/slice_kernel.py | 31 | logger = logging.getLogger(__name__) | CODE |
| LOW | python/triton/runtime/__init__.py | 7 | __all__ = [ | CODE |
| LOW | python/triton/runtime/interpreter.py | 1006 | def set_attr(self, obj: object, name: str, value: object) -> None: | CODE |
| LOW | python/triton/runtime/_allocation.py | 29 | def set_allocator(allocator: Allocator) -> None: | CODE |
| LOW | python/triton/runtime/_allocation.py | 59 | def set_profile_allocator(allocator: Optional[Allocator]) -> None: | CODE |
| LOW | python/triton/runtime/driver.py | 42 | def set_active(self, driver: DriverBase) -> None: | CODE |
| LOW | python/triton/language/__init__.py | 147 | __all__ = [ | CODE |
| LOW | python/triton/language/core.py | 161 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | python/triton/language/core.py | 238 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | python/triton/language/core.py | 904 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | python/triton/language/core.py | 1383 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | python/triton/language/core.py | 1450 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | python/triton/language/core.py | 1576 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | python/triton/language/core.py | 1754 | def _set_name(self, builder: ir.builder, name: str) -> None: | CODE |
| LOW | python/triton/language/target_info.py | 4 | __all__ = ["current_target"] | CODE |
| 8 more matches not shown… | ||||
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | setup.py | 555 | except Exception: | CODE |
| LOW | setup.py | 563 | except Exception: | CODE |
| MEDIUM | setup.py | 551 | def get_git_commit_hash(length=8): | CODE |
| MEDIUM | setup.py | 559 | def get_git_branch(): | CODE |
| LOW | python/test/unit/runtime/test_cache.py | 987 | except Exception as e: | STRING |
| LOW⚡ | python/test/unit/language/test_core.py | 6250 | except Exception as e: | STRING |
| LOW | python/test/unit/language/test_decorator.py | 27 | except Exception as e: | CODE |
| LOW | python/test/backend/test_mir_stage.py | 11 | except Exception: | CODE |
| MEDIUM | python/test/backend/test_mir_stage.py | 8 | def is_hip(): | CODE |
| LOW | python/triton/_compile_warmup_pool.py | 300 | except Exception as error: | CODE |
| LOW | python/triton/_compile_warmup_pool.py | 496 | except Exception as error: | CODE |
| LOW | python/triton/_compile_warmup_pool.py | 502 | except Exception as error: | CODE |
| MEDIUM | python/triton/_compile_warmup_pool.py | 425 | def serialize_specialization(name, signature, constants, attrs, options, key, ta | CODE |
| LOW | python/triton/_internal_testing.py | 326 | except Exception as e: | CODE |
| LOW | python/triton/experimental/gsan/symmetric_memory.py | 174 | except Exception: | CODE |
| LOW | python/triton/experimental/gsan/symmetric_memory.py | 285 | except Exception: | CODE |
| LOW | python/triton/experimental/gsan/symmetric_memory.py | 479 | except Exception: | CODE |
| MEDIUM | python/triton/experimental/gsan/symmetric_memory.py | 282 | def __del__(self) -> None: | CODE |
| LOW | python/triton/experimental/gsan/_utils.py | 113 | except Exception: | CODE |
| LOW | python/triton/runtime/cache.py | 81 | except Exception: | CODE |
| LOW | python/triton/runtime/cache.py | 227 | except Exception: | CODE |
| LOW | python/triton/runtime/_async_compile.py | 23 | except Exception: | CODE |
| LOW | python/triton/runtime/interpreter.py | 1528 | except Exception as e: | CODE |
| LOW | python/triton/runtime/interpreter.py | 1625 | except Exception: | CODE |
| LOW | python/triton/runtime/interpreter.py | 1718 | except Exception as e: | CODE |
| LOW | python/triton/runtime/autotuner.py | 174 | except Exception as e: | CODE |
| LOW | python/triton/compiler/compiler.py | 308 | except Exception as e: | CODE |
| LOW | python/triton/compiler/code_generator.py | 1470 | except Exception as e: | CODE |
| LOW | python/triton/compiler/code_generator.py | 1521 | except Exception as e: | CODE |
| LOW | python/triton/compiler/code_generator.py | 1683 | except Exception as e: | CODE |
| LOW | python/triton/compiler/code_generator.py | 1718 | except Exception as e: | CODE |
| LOW | python/examples/gluon/02-conv-dgrad.py | 980 | except Exception: | CODE |
| MEDIUM | python/examples/gluon/02-conv-dgrad.py | 937 | def _benchmark_dgrad_config( | CODE |
| LOW | python/examples/gluon/02-conv-wgrad.py | 735 | except Exception: | CODE |
| MEDIUM | python/examples/gluon/02-conv-wgrad.py | 692 | def _benchmark_wgrad_config( | CODE |
| LOW | third_party/proton/test/test_instrumentation.py | 1170 | except Exception as e: | CODE |
| MEDIUM | third_party/proton/test/test_instrumentation.py | 1167 | def run_kernel(): | CODE |
| LOW | third_party/proton/test/test_api.py | 98 | except Exception as e: | CODE |
| LOW | third_party/proton/test/test_api.py | 122 | except Exception as e: | CODE |
| LOW | third_party/proton/test/test_api.py | 135 | except Exception as e: | CODE |
| LOW | third_party/proton/test/test_api.py | 289 | except Exception as e: | CODE |
| LOW | third_party/proton/test/test_api.py | 406 | except Exception as e: | CODE |
| LOW | third_party/proton/test/test_api.py | 416 | except Exception as e: | CODE |
| LOW | …ton/test/reproducers/cupti_graph_replay_heap_growth.py | 275 | except Exception: | CODE |
| LOW | third_party/proton/proton/proton.py | 46 | except Exception as e: | STRING |
| MEDIUM | third_party/proton/proton/proton.py | 47 | print("An error occurred while executing the script:") | STRING |
| LOW⚡ | third_party/amd/python/test/test_gluon_cdna5.py | 2371 | except Exception as e: | CODE |
| LOW | third_party/nvidia/backend/driver.py | 66 | except Exception: | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | setup.py | 82 | except subprocess.CalledProcessError: | CODE |
| HIGH | setup.py | 84 | except FileNotFoundError: | CODE |
| HIGH | python/test/unit/language/test_annotations.py | 51 | except AttributeError: | CODE |
| HIGH | python/triton/experimental/gsan/symmetric_memory.py | 285 | except Exception: | CODE |
| HIGH | python/triton/tools/compile.py | 109 | except ValueError: | CODE |
| HIGH | python/triton/tools/compile.py | 114 | except ValueError: | CODE |
| HIGH | python/triton/language/core.py | 112 | except TypeError: | CODE |
| HIGH | third_party/proton/test/test_profile.py | 1115 | except ValueError: | CODE |
| HIGH | third_party/amd/backend/driver.py | 116 | except (subprocess.CalledProcessError, FileNotFoundError): | CODE |
| HIGH | third_party/nvidia/backend/driver.py | 66 | except Exception: | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | python/build_helpers.py | 439 | # Create a stable symlink that doesn't include revision | COMMENT |
| MEDIUM | python/test/gluon/test_core.py | 4906 | # Create a test matrix with known values | COMMENT |
| MEDIUM | python/test/gluon/test_core.py | 5835 | # Create a test matrix [N, M] | COMMENT |
| MEDIUM | python/test/gluon/test_core.py | 5996 | # Create a test tensor [N, M, P] | COMMENT |
| MEDIUM | python/triton/experimental/gluon/language/_semantic.py | 670 | # Create the warp specialize op. | COMMENT |
| MEDIUM | python/triton/experimental/gluon/language/_layouts.py | 617 | # Create a idendity mapping based on shape_per_cta + order | COMMENT |
| MEDIUM | python/triton/runtime/jit.py | 405 | # Create the function argument list and the dict entries for the return statement | COMMENT |
| MEDIUM | python/triton/runtime/jit.py | 1158 | # Create a bound function to support constexpr_function methods | STRING |
| MEDIUM | python/triton/language/core.py | 1706 | # Define the wrapped Triton value type. | COMMENT |
| MEDIUM | python/tutorials/02-fused-softmax.py | 173 | # Create a number of persistent programs. | COMMENT |
| MEDIUM | python/tutorials/01-vector-add.py | 46 | # Create a mask to guard memory operations against out-of-bounds accesses. | COMMENT |
| MEDIUM | docs/conf.py | 6 | # This file is execfile()d with the current directory set to its | COMMENT |
| MEDIUM | third_party/proton/proton/hooks/instrumentation.py | 39 | # Create the buffer | COMMENT |
| MEDIUM | third_party/proton/proton/hooks/instrumentation.py | 92 | # Create the appropriate mode instance | COMMENT |
| MEDIUM | third_party/amd/python/test/test_gluon_cdna5.py | 2589 | # Create a flattened view into the nD reference to unify the indexing logic over all dimensions | COMMENT |
| MEDIUM | third_party/amd/python/examples/gluon/mxfp_fa_cdna5.py | 3151 | # Create a pytest id | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | …iton/experimental/gluon/language/amd/cdna5/_layouts.py | 106 | """Create partitioned shared memory layouts and WMMA layout for a CDNA5 GEMM in order to avoid LDS partition conflic | STRING |
| MEDIUM | …riton/tools/triton_to_gluon_translator/slice_kernel.py | 116 | # HACK: An AST rewrite would be more robust, but this works for now. | COMMENT |
| MEDIUM | python/triton/runtime/cache.py | 117 | # use temp dir to be robust against program interruptions | COMMENT |
| LOW | python/triton/runtime/interpreter.py | 1622 | # e.g., dynamically generated functions, we cannot rewrite it so just return the original function | COMMENT |
| MEDIUM | python/triton_kernels/bench/bench_dense_matmul.py | 3 | # This is not meant to be a comprehensive benchmark of triton_kernels. | COMMENT |
| MEDIUM | python/triton_kernels/bench/bench_dense_matmul.py | 135 | # even for the fp16 benchmark path. Match that when reproducing the OpenAI harness. | COMMENT |
| MEDIUM | python/tutorials/09-persistent-matmul.py | 417 | # Enable warp specialization to leverage async warp scheduling in the GPU. | COMMENT |
| MEDIUM | python/tutorials/10-block-scaled-matmul.py | 405 | # With 256 elements we better utilize the L2 and don't require the TMA | COMMENT |
| MEDIUM | python/tutorials/gluon/11-tcgen05-mma-scaled.py | 654 | # elements. With 256 bytes along the inner dimension, we better utilize the | COMMENT |
| MEDIUM | python/tutorials/gluon/08-warp-specialization.py | 369 | # specialization is especially powerful when used to build intricate schedules | COMMENT |
| MEDIUM | python/tutorials/gluon/02-layouts.py | 868 | # Linear layouts are extremely poweful, and can be used in conjunction with | COMMENT |
| MEDIUM | python/tutorials/gluon/07-persistence.py | 841 | # work in conjunction with the GPU, getting the best of both worlds. This is | COMMENT |
| LOW | third_party/amd/python/test/test_gluon_cdna5.py | 3157 | # In practice, since we know that phase is initialized to 0, we can just set count = NUM_WARPS * WARP_SIZE and call | COMMENT |
| LOW | third_party/amd/python/test/test_gluon_cdna5.py | 3218 | # In practice, since we know that phase is initialized to 0, we can just set count = NUM_WARPS and call directly ttg | COMMENT |
| MEDIUM | …party/amd/lib/TritonAMDGPUTransforms/ScheduleLoops.cpp | 210 | /// which is not coalesced. To better utilize global memory bandwidth, | COMMENT |
| MEDIUM | …party/amd/lib/TritonAMDGPUToLLVM/LoadStoreOpToLLVM.cpp | 2290 | // In the case of unpaired f16 elements utilize dpp instructions to | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | python/test/unit/language/test_core.py | 224 | CODE | |
| LOW | python/triton/testing.py | 43 | CODE | |
| LOW | python/triton/experimental/gluon/_runtime.py | 81 | CODE | |
| LOW | python/triton/tools/ragged_tma.py | 76 | CODE | |
| LOW | …/tools/triton_to_gluon_translator/blackwell_helpers.py | 93 | CODE | |
| LOW | …/tools/triton_to_gluon_translator/blackwell_helpers.py | 205 | CODE | |
| LOW | …/tools/triton_to_gluon_translator/blackwell_helpers.py | 304 | CODE | |
| LOW | …triton/tools/triton_to_gluon_translator/amd_helpers.py | 302 | CODE | |
| LOW | …ton/tools/triton_to_gluon_translator/nvidia_helpers.py | 53 | CODE | |
| LOW | …ton/tools/triton_to_gluon_translator/nvidia_helpers.py | 133 | CODE | |
| LOW | …/triton/tools/triton_to_gluon_translator/translator.py | 257 | CODE | |
| LOW | python/triton/runtime/jit.py | 968 | CODE | |
| LOW | python/triton/runtime/autotuner.py | 424 | CODE | |
| LOW | python/triton/runtime/autotuner.py | 21 | CODE | |
| LOW | python/triton/language/standard.py | 443 | CODE | |
| LOW | python/triton/language/core.py | 3701 | CODE | |
| LOW | python/triton/compiler/code_generator.py | 296 | CODE | |
| LOW | python/examples/gluon/04-2cta-block-scale-matmul.py | 818 | CODE | |
| LOW | python/triton_kernels/triton_kernels/matmul.py | 248 | CODE | |
| LOW | python/triton_kernels/triton_kernels/matmul.py | 893 | CODE | |
| LOW | python/triton_kernels/triton_kernels/reduce.py | 665 | CODE | |
| LOW | python/triton_kernels/triton_kernels/reduce.py | 1145 | CODE | |
| LOW | python/triton_kernels/triton_kernels/reduce.py | 1183 | CODE | |
| LOW | python/triton_kernels/triton_kernels/testing.py | 295 | CODE | |
| LOW | python/triton_kernels/triton_kernels/swiglu.py | 100 | CODE | |
| LOW | python/triton_kernels/triton_kernels/swiglu.py | 82 | CODE | |
| LOW | …n_kernels/tensor_details/layout_details/cdna4_scale.py | 77 | CODE | |
| LOW | …_kernels/tensor_details/layout_details/hopper_value.py | 288 | CODE | |
| LOW | …_kernels/tensor_details/layout_details/hopper_value.py | 252 | CODE | |
| LOW | …rnels/tensor_details/layout_details/blackwell_scale.py | 212 | CODE | |
| LOW | …_kernels/tensor_details/layout_details/hopper_scale.py | 147 | CODE | |
| LOW | …ton_kernels/triton_kernels/matmul_details/_p_matmul.py | 84 | CODE | |
| LOW | …riton_kernels/triton_kernels/matmul_details/_matmul.py | 38 | CODE | |
| LOW | python/tutorials/02-fused-softmax.py | 59 | CODE | |
| LOW | python/tutorials/gluon/14-multicta.py | 1066 | CODE | |
| LOW | …ird_party/amd/python/examples/gluon/mxfp_gemm_cdna5.py | 132 | CODE | |
| LOW | third_party/amd/python/examples/gluon/moe_cdna5.py | 1202 | CODE | |
| LOW | third_party/amd/python/examples/gluon/mxfp_fa_cdna5.py | 3006 | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW⚡ | python/test/unit/tools/test_slice_kernel.py | 749 | def helper() -> None: | CODE |
| LOW⚡ | python/test/unit/tools/test_slice_kernel.py | 765 | def helper() -> int: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 130 | def helper() -> int: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 157 | def helper() -> int: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 178 | def helper() -> int: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 205 | def helper() -> int: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 466 | def helper() -> None: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 481 | def helper() -> None: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 501 | def helper(): | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 516 | def helper(): | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 539 | def helper() -> None: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 558 | def helper() -> None: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 607 | def helper(): | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 622 | def helper(): | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 649 | def helper() -> None: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 668 | def helper() -> None: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 691 | def helper() -> None: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 710 | def helper() -> None: | CODE |
| LOW | python/test/unit/tools/test_slice_kernel.py | 733 | def helper() -> None: | CODE |
| LOW | python/test/unit/runtime/test_driver.py | 290 | def helper(value): | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | …iton/experimental/gluon/language/amd/cdna5/__init__.py | 0 | computes matrix multiplication ``a * b + acc`` using an amd wmma instruction. args: a (tensor): the operand a to be mult | STRING |
| HIGH | …iton/experimental/gluon/language/amd/rdna4/__init__.py | 0 | computes matrix multiplication ``a * b + acc`` using an amd wmma instruction. args: a (tensor): the operand a to be mult | STRING |
| HIGH | …iton/experimental/gluon/language/amd/rdna3/__init__.py | 0 | computes matrix multiplication ``a * b + acc`` using an amd wmma instruction. args: a (tensor): the operand a to be mult | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | python/build_helpers.py | 131 | # Set timeout to 300 seconds to prevent the request from hanging forever. | COMMENT |
| LOW | python/triton/compiler/code_generator.py | 151 | # Check if the given syntax node has an "early" return | COMMENT |
| LOW | utils/generate-test-checks.py | 177 | # Check if an existing variable exists for this name. | STRING |
| LOW | utils/generate-test-checks.py | 320 | # Check if this is an attribute definition and process it | STRING |
| LOW | third_party/proton/proton/hooks/hook.py | 93 | # Check if any other sessions rely on this hook | COMMENT |
| LOW | third_party/amd/python/test/address_sanitizer_helper.py | 23 | #Set access to go out of bounds for ASAN test | COMMENT |
| LOW | third_party/amd/backend/driver.py | 289 | # Check if cooperative groups are supported on the device. | COMMENT |
| LOW | .github/workflows/integration-tests-nvidia.yml | 45 | # Check if files exist before proceeding | COMMENT |
| LOW | .github/workflows/integration-tests-amd.yml | 91 | # Check if files exist before proceeding | COMMENT |
| LOW | .github/workflows/build-macos.yml | 38 | # Check if files exist before proceeding | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | python/test/unit/tools/test_aot.py | 147 | while (fscanf(file, "%hd,", &buffer[index]) != EOF && index < size) { | CODE |
| HIGH | python/triton/experimental/gsan/symmetric_memory.py | 431 | stack.push(listener) | CODE |
| HIGH | python/triton/tools/link.py | 218 | src += " if (algo_id < 0 || algo_id >= num_algos)\n" | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | python/test/unit/language/test_core.py | 7744 | # reuse the temp variable and make sure to check that it isn't creating incorrect IR. | STRING |
| LOW | python/tutorials/01-vector-add.py | 74 | # - Don't forget to pass meta-parameters as keywords arguments. | COMMENT |
| MEDIUM | python/tutorials/gluon/14-multicta.py | 357 | # It's worth noting that the pattern changes a bit once the TMA has to wait on the tcgen05_mma | COMMENT |
| MEDIUM | python/tutorials/gluon/02-layouts.py | 839 | # require reordering registers within a thread (which is free), you can use | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | python/tutorials/11-programmatic-dependent-launch.py | 26 | # In this example | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW⚡ | python/test/unit/language/test_line_info.py | 584 | run_filecheck("placeholder", h.asm["ttir"], check_template) | STRING |
| LOW | python/test/unit/language/test_line_info.py | 344 | run_filecheck("placeholder", h.asm["ttir"], check_template) | STRING |
| LOW | python/test/unit/language/test_line_info.py | 364 | run_filecheck("placeholder", h.asm["ttir"], check_template) | STRING |
| LOW | python/test/unit/language/test_line_info.py | 383 | run_filecheck("placeholder", h.asm["ttir"], check_template) | STRING |
| LOW | python/test/unit/language/test_line_info.py | 396 | run_filecheck("placeholder", h.asm["ttir"], check_template) | STRING |
| LOW | python/test/unit/language/test_line_info.py | 416 | run_filecheck("placeholder", h.asm["ttir"], check_template) | STRING |
| LOW | python/test/unit/language/test_line_info.py | 442 | run_filecheck("placeholder", h.asm["ttir"], check_template) | STRING |
| LOW | python/test/unit/language/test_line_info.py | 466 | run_filecheck("placeholder", h.asm["ttir"], check_template) | STRING |
| LOW | python/test/unit/language/test_line_info.py | 493 | run_filecheck("placeholder", h.asm["ttir"], check_template) | STRING |
| LOW | python/test/unit/language/test_line_info.py | 605 | run_filecheck("placeholder", h.asm["ttir"], check_template) | STRING |
| LOW | python/triton/_filecheck.py | 87 | run_filecheck("placeholder", mlir_module.str_nodebug(), check_template) | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| HIGH | …on/triton/experimental/gluon/language/amd/cdna5/tdm.py | 162 | Update selected fields of a TDM descriptor; return a new descriptor SSA value. Each parameter is independently opti | STRING |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | third_party/amd/backend/include/hip/hip_runtime_api.h | 2500 | // TODO: implement IPC apis | COMMENT |
| LOW | lib/Analysis/Alias.cpp | 85 | // TODO: implement | COMMENT |
| LOW | lib/Analysis/Alias.cpp | 91 | // TODO: implement | COMMENT |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| MEDIUM | python/test/unit/language/test_compile_errors.py | 172 | CODE | |
| MEDIUM | python/triton_kernels/triton_kernels/tensor.py | 98 | CODE |
| Severity | File | Line | Snippet | Context |
|---|---|---|---|---|
| LOW | bin/triton-tensor-layout.cpp | 23 | // Example usage: | COMMENT |