v0.0.5

What's Changed

• Rename linter/check_main.py -> scripts/lint_examples_main.py by @jansel in #124
• Improve error message for unpacking a tile by @jansel in #125
• Improve error message for overpacked tiles by @jansel in #126
• [BC breaking] Simplify block size configs by @jansel in #127
• Refactor reduction loop config spec by @jansel in #128
• Move BlockIdSequence to its own file by @jansel in #129
• Do not print output code durring autotuning by @jansel in #130
• Make helion.exc.TensorOperationInWrapper not fire on non-torch ops by @jansel in #131
• Add HELION_FORCE_AUTOTUNE=1 and update readme by @jansel in #132
• Correct units for time printouts by @jansel in #133
• Rename block_size_idx to block_id by @jansel in #134
• Rename block_indices to block_ids by @jansel in #135
• Add Pyre Pre-Commit Hook by @lolpack in #136
• Update .pre-commit-config.yaml by @oulgen in #137
• [Ready for review] Add hl.register_reduction_dim(); add support for matmul+layernorm example by @yf225 in #80
• Fix bug with errors on unreachable if branch by @jansel in #138
• [Error Message] Update block config size length mismatch by @drisspg in #139
• Increase atol/rtol for test_error_in_non_taken_branch by @jansel in #142
• Fix some typos by @jansel in #141
• More fair comparison by @drisspg in #146

New Contributors

• @lolpack made their first contribution in #136

Full Changelog: v0.0.4...v0.0.5

v0.0.4

What's Changed

• Beef up pre-commit checks by @oulgen in #106
• Run pre-commit as part of lint action by @oulgen in #108
• Add jagged_dense_add_2d example in generalize tensor indexing by @jansel in #105
• Update README.md with Helion logo by @oulgen in #100
• Optimization pass to remove unneeded masking by @jansel in #109
• Improve mask optimization to cover control flow and inductor ops by @jansel in #111
• Expand README.md by @jansel in #112
• Fix ImportError: cannot import name 'Never' from 'typing' by @jansel in #114
• Remove 'first_non_grid_index' for hl.grid index by @jansel in #113
• Pass to remove unnecessary hl.tile_index calls by @jansel in #115
• Replace torch.fx.GraphModule with torch.fx.Graph by @jansel in #116
• MoE matmul example by @yf225 in #110
• Add main() to moe_matmul_ogs by @yf225 in #118
• Add pre-commit hook to make sure examples have a main function by @oulgen in #119
• Add reduction example: Long sum by @joydddd in #92
• Make loop reordering work with register_block_size by @jansel in #117
• Temporarily disable unit test for moe_matmul_ogs example by @yf225 in #120
• Skip test_moe_matmul_ogs on older cards by @jansel in #121
• Make l2_grouping work with register_block_size by @jansel in #122
• Re-enable unit test for moe_matmul_ogs example; skip in fbcode by @yf225 in #123

New Contributors

• @joydddd made their first contribution in #92

Full Changelog: v0.0.3...v0.0.4

v0.0.3

What's Changed

• Minor fix to test file name by @yf225 in #1
• Add CI workflow by @yf225 in #2
• Allow direct running of add.py example by @yf225 in #6
• [CI] Use A10G (g5.4xlarge) machine type by @yf225 in #4
• Use site-package for torch in pyre_configuration by @jansel in #8
• Add use_default_config setting by @jansel in #9
• Add LICENSE/CONTRIBUTING.md/CODE_OF_CONDUCT.md by @jansel in #11
• Support persistent reductions by @jansel in #10
• Fix handling of block_ptr + reductions by @jansel in #12
• Support inductor lowerings that require multiple buffers by @jansel in #13
• Adjust rtol/atol for test_sum_keepdims by @yf225 in #14
• Support Python 3.10; Run lint in CI by @yf225 in #7
• Support looped reductions by @jansel in #15
• Compile in a subprocess to kill hangs by @jansel in #16
• Refactor autotuning logging by @jansel in #17
• Support view ops by @jansel in #18
• Support indirect loads by @jansel in #19
• Improve README.md by @jansel in #20
• Support if/else control flow by @jansel in #21
• Add hl.constexpr specialization by @jansel in #22
• Fix license file for PEP 621 by @oulgen in #23
• Use search-strategy: all for all site packages in pyre config by @stroxler in #25
• Add decorator check by @oulgen in #24
• Trigger CI on pull requests made by ghstack by @oulgen in #27
• Add hl.register_block_size and explicit tile sizes by @jansel in #30
• Update lint github workflow by @jansel in #31
• Add ../pytorch-nightly to Pyre optional_search_path by @yf225 in #36
• Fix TensorDescriptor handling in _find_device by @yf225 in #35
• Add HELION_USE_DEFAULT_CONFIG env var to force use default config by @yf225 in #37
• Add more pytorchbot utils by @oulgen in #43
• Add the core properties to Config object by @drisspg in #49
• Switch build system to Hatchling which has much better Language Server support by @drisspg in #55
• Add attention example and fix some bugs by @jansel in #56
• Fix bug where non-tensor variables are not exposed to inner loops by @jansel in #58
• Add hl.grid(...) support by @yf225 in #59
• Fix more unit tests by @oulgen in #64
• Fix test_matmul_tensor_descriptor unit test by @yf225 in #65
• Prototyping an hl.atomic opp by @drisspg in #63
• Add hl.specialize and improve reduction handling by @jansel in #72
• [test] Touch test/init.py to support more testing workflows by @danzimm in #73
• [reland without ghstack] handle PTXASError by @jansel in #79
• Support data-dependent loop bounds by @jansel in #81
• Add support for hl.tile(begin, end) and hl.tile(begin, end, block_size) by @jansel in #82
• Support user-defined minimum in hl.register_block_size by @jansel in #83
• Don't re-wrap exceptions in exc.TorchOpTracingError by @jansel in #84
• Add hl.tile_index() by @jansel in #89
• Add filecheck dependency by @jansel in #95
• Add env HELION_PRINT_OUTPUT_CODE=1 by @jansel in #93
• Add extra_mask arg to hl.load and hl.store by @jansel in #94
• Bump project version by @oulgen in #101
• Swap to using hatch vcs by @oulgen in #103
• Add publish to pypi workflow by @oulgen in #104

New Contributors

• @stroxler made their first contribution in #25
• @drisspg made their first contribution in #49
• @danzimm made their first contribution in #73

Full Changelog: https://github.com/pytorch-labs/helion/commits/v0.0.3