Amin Sedaghat 767199fd9b [flex_attention] replace sliced BlockMask noop with helpful error (#164702) ...

Fixes part of #163314

After slicing BlockMask with `[]`, mask_mod was silently replaced with noop_mask. This caused silent incorrect results when users applied transformations to `sliced_mask.mask_mod`.

Replace noop with `_sliced_mask_mod_error` that raises RuntimeError with guidance to use `base_mask.mask_mod` instead.

Pull Request resolved: https://github.com/pytorch/pytorch/pull/164702
Approved by: https://github.com/drisspg, https://github.com/BoyuanFeng

2025-10-20 03:46:16 +00:00

..

cpp

Fix some CMake issues (#153686)

2025-05-19 00:31:34 +00:00

extension_backends

[BE][PYFMT] migrate PYFMT for test/inductor/ to ruff format (#148186)

2025-06-24 11:12:11 +00:00

__init__.py

…

CMakeLists.txt

[aoti] Add initial custom op support (#127034)

2024-07-24 20:29:55 +00:00

custom_inductor_config.py

[inductor] Allow backends to register their own custom config object (#158254)

2025-07-23 15:56:06 +00:00

custom_ops.cpp

integrate kernacle into inductor (#160121)

2025-08-08 02:14:44 +00:00

indirect_assert_helper.py

[Inductor][WIndows] Fix Windows test case failure. (#161497)

2025-08-28 12:40:42 +00:00

minifier_smoke.py

[BE][Easy][12/19] enforce style for empty lines in import segments in test/i*/ (#129763)

2024-07-18 07:49:19 +00:00

opinfo_harness.py

[BE][Easy][12/19] enforce style for empty lines in import segments in test/i*/ (#129763)

2024-07-18 07:49:19 +00:00

s429861_repro.py

[BE] format test/inductor/s429861_repro.py (#148554)

2025-03-21 20:39:28 +00:00

test_alignment.py

[inductor] handle offset in ReinterpretView for alignment (#151859)

2025-04-23 01:50:49 +00:00

test_analysis.py

Revert "Improve device info with new flops and bandwidth formula based on hardware libraries (#162245)"

2025-09-19 20:09:12 +00:00

test_aot_inductor_arrayref.py

[export] Fix custom ops in subgraphs (#160004)

2025-08-18 15:42:19 +00:00

test_aot_inductor_custom_ops.py

[ROCm] Enabling several UTs (#161715)

2025-09-09 15:49:21 +00:00

test_aot_inductor_package.py

Add an option to put store large mmap weights on disk (#164526)

2025-10-10 07:53:57 +00:00

test_aot_inductor_utils.py

Migrating some more callsites (#163580)

2025-10-19 15:52:17 +00:00

test_aot_inductor_windows.py

Add an option to put store large mmap weights on disk (#164526)

2025-10-10 07:53:57 +00:00

test_aot_inductor.py

Fix incorrect function signature in template (#165567)

2025-10-17 02:40:56 +00:00

test_aoti_cross_compile_windows.py

[Windows] Add AOTI cross-compilation CI (#165573)

2025-10-17 01:05:35 +00:00

test_async_compile.py

async_compile: Fix the wait method to actually wait (#161561)

2025-08-27 21:35:31 +00:00

test_augmented_graph_helper.py

Limit coll bucketing within node idxs (#164944)

2025-10-08 22:18:53 +00:00

test_auto_functionalize.py

[Dynamo] Fix arg ordering in tf modes (#159707)

2025-08-05 01:43:21 +00:00

test_autoheuristic.py

Revert "Fix skipIfXpu and skipIfHpu disables tests when used on class (#151315)"

2025-05-14 13:15:03 +00:00

test_b2b_gemm.py

Enable all SIM rules except disabled ones (#164645)

2025-10-17 07:27:11 +00:00

test_benchmark_fusion.py

Enable all SIM rules except disabled ones (#164645)

2025-10-17 07:27:11 +00:00

test_benchmarking.py

refactor benchmarking to use dynamo_timed (#144315)

2025-01-18 02:35:00 +00:00

test_best_config.py

[Inductor] Record Triton’s Base32 Cache Key in .best_config for Debugging (#154618)

2025-05-30 19:30:25 +00:00

test_binary_folding.py

Enable ruff rule E721 (#165162)

2025-10-13 01:48:55 +00:00

test_block_analysis.py

[inductor][triton] Block ptr analysis fix assert on matched index expression (#148446)

2025-03-10 05:26:55 +00:00

test_cache.py

Enable ruff rule E721 (#165162)

2025-10-13 01:48:55 +00:00

test_caching.py

non-fb impls + unit tests (#164722)

2025-10-09 05:10:57 +00:00

test_ck_backend.py

[ROCm][Inductor][CK backend] Install rocm-composable-kernel python package on ROCm Linux CI docker images (#162288)

2025-09-10 19:33:40 +00:00

test_codecache.py

Enable all PIE rules on ruff (#165814)

2025-10-18 07:36:18 +00:00

test_codegen_triton.py

[inductor][3/N] triton support post-#5512, tt.divisibility format (#145575)

2025-01-27 21:48:58 +00:00

test_combo_kernels.py

[Inductor] Fix ComboKernels failing due to missing helper functions (#162759)

2025-09-12 20:01:06 +00:00

test_compile_subprocess.py

fix pickling for BitwiseFn (#163571)

2025-09-25 04:52:11 +00:00

test_compile_worker.py

[redone][pytorch] Moving torch.compile worker process logs to a dedicated rank based log directory (#160352)

2025-08-12 16:49:08 +00:00

test_compile.py

[inductor] add _get_inductor_debug_symbol_cflags for debug symbol control. (#159938)

2025-08-06 19:31:45 +00:00

test_compiled_autograd.py

Enable all PIE rules on ruff (#165814)

2025-10-18 07:36:18 +00:00

test_compiled_optimizers.py

Enable all SIM rules except disabled ones (#164645)

2025-10-17 07:27:11 +00:00

test_config.py

Use torch.testing.test_close instead of torch.testing.test_allclose (#164539)

2025-10-03 14:39:10 +00:00

test_control_deps.py

Add hop for additional control dependencies (#164568)

2025-10-06 15:47:55 +00:00

test_control_flow.py

[scan] materialize combine_fn in forward add more autograd tests (#161732)

2025-09-27 18:13:15 +00:00

test_cooperative_reductions.py

rename-HAS_CUDA-to-HAS_CUDA_AND_TRITON (#159883)

2025-08-08 15:44:52 +00:00

test_coordinate_descent_tuner.py

[Inductor] Refactor "r" reduction prefix to {"r0_", "r1_"}. (#142020)

2024-12-12 17:22:20 +00:00

test_cpp_wrapper_hipify.py

[AOTI] Embed cubin files into .so (#150739)

2025-05-19 01:11:46 +00:00

test_cpu_cpp_wrapper.py

[cpp_wrapper] Swap to new PyBind11 simple GIL header (#161063)

2025-08-27 21:15:01 +00:00

test_cpu_repro.py

[Inductor] support masked vectorization for the tail_loop for fp8 datatype (#163324)

2025-10-20 01:56:00 +00:00

test_cpu_select_algorithm.py

[Inductor][CPP] Fix the test case of test_linear_reuse_kernels (#163723)

2025-09-29 05:29:01 +00:00

test_cuda_repro.py

Gate division bitwise numerics under a flag (#165566)

2025-10-15 23:41:01 +00:00

test_cuda_select_algorithm.py

Disable failing test_int8_woq_mm_concat_cuda on slow grad check (#165331)

2025-10-13 17:08:00 +00:00

test_cudacodecache.py

Fix requires_cuda to requires_cuda_and_triton (#160222)

2025-08-10 07:05:52 +00:00

test_cudagraph_trees_expandable_segments.py

rename-HAS_CUDA-to-HAS_CUDA_AND_TRITON (#159883)

2025-08-08 15:44:52 +00:00

test_cudagraph_trees.py

Enable all SIM rules except disabled ones (#164645)

2025-10-17 07:27:11 +00:00

test_custom_lowering.py

[Inductor] Allow passing in custom lowering dict to register_lowering() (#154344)

2025-05-27 01:35:26 +00:00

test_custom_partitioner_fn.py

allow user to pass in custom partitioner function (#157580)

2025-09-05 22:49:39 +00:00

test_custom_post_grad_passes.py

Support caching if joint_custom_pre_pass/joint_custom_post_pass implement the proper interface (#157990)

2025-07-10 19:17:11 +00:00

test_cutedsl_template.py

Updates to CuTe DSL template renderer (#161117)

2025-08-27 23:01:31 +00:00

test_cutlass_backend.py

Enable ruff rule E721 (#165162)

2025-10-13 01:48:55 +00:00

test_cutlass_evt.py

[BE][cutlass backend] BE changes post cutlass_cppgen name change (#164589)

2025-10-06 17:22:08 +00:00

test_debug_trace.py

[BE][PYFMT] migrate PYFMT for test/inductor/ to ruff format (#148186)

2025-06-24 11:12:11 +00:00

test_decompose_mem_bound_mm.py

rename-HAS_CUDA-to-HAS_CUDA_AND_TRITON (#159883)

2025-08-08 15:44:52 +00:00

test_dependencies.py

Fix unused Python variables in test/[e-z]* (#136964)

2024-12-18 23:02:30 +00:00

test_deterministic.py

[inductor] verify determinism with inductor benchmark script (#164904)

2025-10-12 00:03:42 +00:00

test_device_assert.py

[Inductor] Update DeviceAssert op to behave like store (#163696)

2025-09-24 23:35:56 +00:00

test_distributed_patterns.py

[ROCm] Enable and fix several FSDP + Inductor distributed unit tests (#165011)

2025-10-10 14:10:54 +00:00

test_efficient_conv_bn_eval.py

Enable ruff rule E721 (#165162)

2025-10-13 01:48:55 +00:00

test_extension_backend.py

[Inductor UT][Windows][XPU] Fix Inductor UT on XPU Windows. (#146481)

2025-02-22 02:53:16 +00:00

test_external_callables.py

[BE][4/6] fix typos in test/ (test/inductor/) (#157638)

2025-07-06 06:34:25 +00:00

test_flex_attention.py

[flex_attention] replace sliced BlockMask noop with helpful error (#164702)

2025-10-20 03:46:16 +00:00

test_flex_decoding.py

Enable all SIM rules except disabled ones (#164645)

2025-10-17 07:27:11 +00:00

test_flex_flash.py

Add Loads from fixed inputs (#162031)

2025-10-10 01:23:37 +00:00

test_foreach.py

Fix requires_cuda to requires_cuda_and_triton (#160222)

2025-08-10 07:05:52 +00:00

test_fp8.py

[Inductor][Triton][FP8] Refactor scaled_mm template to accept scaling mode (#164318)

2025-10-16 20:40:45 +00:00

test_fused_attention.py

Add SDPA patterns for T5 variants when batch size is 1 (#163252)

2025-09-26 08:50:06 +00:00

test_fuzzer.py

[export] Turn on install_free_tensors flag (#164691)

2025-10-14 15:33:50 +00:00

test_fx_fusion.py

[inductor] Replace torch.allclose with torch.testing.assert_close in test_fx_fusion (#130618)

2024-08-06 03:58:43 +00:00

test_fxir_backend.py

[Inductor-FX] Generalize FloorDiv conversion to handle more complex launch grids. Remove python_slow grid mode. (#163828)

2025-09-30 16:47:49 +00:00

test_gpu_cpp_wrapper.py

Fix redundant H2D/D2H memcpy in cpp_wrapper by creating scalar tensors on CPU (#160584)

2025-09-24 23:40:37 +00:00

test_graph_transform_observer.py

Enable all SIM rules except disabled ones (#164645)

2025-10-17 07:27:11 +00:00

test_group_batch_fusion.py

More ruff SIM fixes (#164695)

2025-10-09 03:24:50 +00:00

test_halide.py

Fix test_halide.py report invocation to re-run failed tests (#147640)

2025-02-26 06:32:22 +00:00

test_helion_kernels.py

Add Helion softmax test (#155976)

2025-06-14 05:53:21 +00:00

test_indexing.py

Fix FloorDiv should not generate non integer rationals (due to sympy bug) (#164398)

2025-10-02 22:51:03 +00:00

test_inductor_annotations.py

[Graph Partition] Pass all OSS unit tests (#154667)

2025-08-12 04:37:58 +00:00

test_inductor_freezing.py

[ez] Disable some failing periodic tests (#156731)

2025-06-24 23:02:21 +00:00

test_inductor_scheduler.py

Add torch compile force disable caches alias (#158072)

2025-08-02 23:23:17 +00:00

test_inductor_utils.py

can't infer device on benchmarked function with no args or kwargs (#133290)

2024-08-13 20:13:44 +00:00

test_inplace_padding.py

[Linter] Expanding the scope of detecting device-bias code. (#159949)

2025-08-09 09:41:16 +00:00

test_inplacing_pass.py

Fix silent incorrectness arising from incorrect alias information (#152011)

2025-06-27 12:45:03 +00:00

test_kernel_benchmark.py

[Break XPU][Indutor UT] Fix failures introduced by community. (#159463)

2025-07-31 08:37:41 +00:00

test_kernel_optimization.py

[CI][testing] Use 3 processes for testing on sm89 and sm90 jobs (#158691)

2025-07-25 15:26:29 +00:00

test_layout_optim.py

[BE][4/6] fix typos in test/ (test/inductor/) (#157638)

2025-07-06 06:34:25 +00:00

test_loop_ordering.py

[inductor] don't try to reorder loops for template (#165601)

2025-10-16 19:05:21 +00:00

test_max_autotune.py

Enable all PIE rules on ruff (#165814)

2025-10-18 07:36:18 +00:00

test_mem_estimation.py

[Fix XPU CI] [Inductor UT] Fix test cases broken by community. (#165714)

2025-10-19 23:59:04 +00:00

test_memory_planning.py

Migrating some more callsites (#163580)

2025-10-19 15:52:17 +00:00

test_memory.py

add more restriction to fusion with large accumulate reads (#163163)

2025-09-19 01:20:30 +00:00

test_metrics.py

[BE][PYFMT] migrate PYFMT for test/inductor/ to ruff format (#148186)

2025-06-24 11:12:11 +00:00

test_minifier_isolate.py

[Inductor UT] Generalize inductor UT for intel GPU (Part 2) (#134556)

2024-09-13 05:16:28 +00:00

test_minifier_utils.py

shape guards (#161178)

2025-09-08 22:44:09 +00:00

test_minifier.py

shape guards (#161178)

2025-09-08 22:44:09 +00:00

test_mkldnn_pattern_matcher.py

Enable all SIM rules except disabled ones (#164645)

2025-10-17 07:27:11 +00:00

test_mmdecomp.py

[inductor] Fix mm decomposition evaluating symints (#158998)

2025-07-30 16:34:15 +00:00

test_move_constructors_to_cuda.py

rename-HAS_CUDA-to-HAS_CUDA_AND_TRITON (#159883)

2025-08-08 15:44:52 +00:00

test_mps_basic.py

AOTI MPS Shim Implementation (#163865)

2025-10-09 16:06:36 +00:00

test_multi_kernel.py

[multi-kernel] shape-similarity kernel selection (#163090)

2025-09-23 21:00:47 +00:00

test_native_matmul.py

[Fix XPU CI] [Inductor UT] Fix test cases broken by community. (#165714)

2025-10-19 23:59:04 +00:00

test_needs_exact_strides.py

rename-HAS_CUDA-to-HAS_CUDA_AND_TRITON (#159883)

2025-08-08 15:44:52 +00:00

test_online_softmax.py

[inductor] fix 3d tiled online softmax (#162341)

2025-09-09 02:59:52 +00:00

test_op_completeness.py

[inductor] Refactor op handlers part 5 (#146257)

2025-02-08 18:00:30 +00:00

test_op_dtype_prop.py

[inductor] add a runtime assert for triton shapes (#164242)

2025-10-01 18:55:33 +00:00

test_ordered_set.py

[BE] fix remaining flake8 v7 warnings (#159044)

2025-07-25 02:56:34 +00:00

test_pad_mm.py

Disable autocast when running joint graph passes (#162304)

2025-09-06 00:57:58 +00:00

test_padding.py

[ROCm] Remove HIPBLASLT_ALLOW_TF32 from codebase (#162998)

2025-09-18 13:53:48 +00:00

test_pattern_matcher.py

[inductor] Handle patterns where input/output nodes are the same (#163994)

2025-10-02 08:37:55 +00:00

test_perf.py

[Graph Partition] improve custom op output alias (#163227)

2025-09-19 17:01:36 +00:00

test_profiler.py

[1/N] Apply UP035 rule in tests (#163947)

2025-09-29 01:42:01 +00:00

test_provenance_tracing.py

Revert "Add provenance to inductor IR nodes created after graph.run (#164255)"

2025-10-02 22:01:41 +00:00

test_quantization.py

[3/n][Optimus][Auto-AC][reland] Support any fp8 quantization type and set scaling as the default" (#154057)

2025-05-22 18:26:34 +00:00

test_remote_cache.py

pt2_remote_cache: Log sample for failures, and log the explicit reason we're faling. (#156874)

2025-07-18 20:28:27 +00:00

test_scatter_optimization.py

[BE][PYFMT] migrate PYFMT for test/inductor/ to ruff format (#148186)

2025-06-24 11:12:11 +00:00

test_segmented_tree.py

[inductor] dont reuse buffers if it affects peak (#145883) (#159530)

2025-08-19 19:02:56 +00:00

test_select_algorithm.py

[inductor][templates] Distinguish between kernel input nodes and codegen input nodes (#163752)

2025-10-08 14:12:14 +00:00

test_smoke.py

rename-HAS_CUDA-to-HAS_CUDA_AND_TRITON (#159883)

2025-08-08 15:44:52 +00:00

test_snode_runtime.py

[Reland][2/N]Port several test files under test/distributed to Intel GPU (#159473)

2025-09-17 06:42:27 +00:00

test_split_cat_fx_aten_passes.py

Fix requires_cuda to requires_cuda_and_triton (#160222)

2025-08-10 07:05:52 +00:00

test_split_cat_fx_passes.py

add torch.concat to normalization pass (#156574)

2025-06-27 06:07:26 +00:00

test_static_cuda_launcher.py

Fix requires_cuda to requires_cuda_and_triton (#160222)

2025-08-10 07:05:52 +00:00

test_subgraph_choice.py

[inductor][mm] restructure decompose k (#161026)

2025-08-28 20:14:41 +00:00

test_template_heuristics_registry.py

[inductor][heuristics registry] missing heuristic is not an error anymore, cross device heuristics (#161767)

2025-08-29 22:41:27 +00:00

test_torchbind.py

Add kernel information JSON generation for AOTI packages (#160540)

2025-08-20 02:33:45 +00:00

test_torchinductor_codegen_config_overrides.py

Add cse for make_block_ptr in Triton codegen (#163399)

2025-10-16 05:29:48 +00:00

test_torchinductor_codegen_dynamic_shapes.py

[Fix XPU CI] [Inductor UT] Fix test cases broken by community. (#165406)

2025-10-16 00:53:32 +00:00

test_torchinductor_dynamic_shapes.py

Revert "use sym_numel, to allow fake tensors to work (#163831)"

2025-10-14 05:10:56 +00:00

test_torchinductor_opinfo.py

[Fix XPU CI] [Inductor UT] Fix test cases broken by community. (#165714)

2025-10-19 23:59:04 +00:00

test_torchinductor_strided_blocks.py

Add cse for make_block_ptr in Triton codegen (#163399)

2025-10-16 05:29:48 +00:00

test_torchinductor.py

Enable all SIM rules except disabled ones (#164645)

2025-10-17 07:27:11 +00:00

test_triton_cpu_backend.py

enable test (#155342)

2025-06-09 19:26:05 +00:00

test_triton_extension_backend.py

cpp_wrapper: fix inductor triton tests (#146109)

2025-02-25 19:50:37 +00:00

test_triton_heuristics.py

[Reland][Inductor] Prune configs that require more shared memory than the hardware limit. (#161996)

2025-09-03 04:23:09 +00:00

test_triton_kernels.py

Enable all PIE rules on ruff (#165814)

2025-10-18 07:36:18 +00:00

test_triton_syntax.py

cpp_wrapper: fix inductor triton tests (#146109)

2025-02-25 19:50:37 +00:00

test_triton_wrapper.py

Fix unused Python variables in test/[e-z]* (#136964)

2024-12-18 23:02:30 +00:00

test_unbacked_symints.py

[inductor] fix issue for example value with unbacked strides (#163660)

2025-10-14 20:07:51 +00:00

test_utils.py

Enable ruff rule E721 (#165162)

2025-10-13 01:48:55 +00:00

test_xpu_basic.py

Renaming HAS_XPU to HAS_XPU_AND_TRITON (#159908)

2025-08-07 11:24:44 +00:00