v0.1.5

[LeiWang1999]

What's Changed

• [Release] Bump version from 0.1.3 into 0.1.4 by @LeiWang1999 in [Release] Bump version from 0.1.3 into 0.1.4 #375
• [Enhancement] Remove redundant recursive rewrite rule for FloorDiv in RewriteSimplifier by @LeiWang1999 in [Enhancement] Remove redundant recursive rewrite rule for FloorDiv in RewriteSimplifier #408
• [Docker] cu128 Support by @andyluo03 in [Docker] cu128 Support #410
• [Refactor] Phaseout python dependency attrs and decorator by @LeiWang1999 in [Refactor] Phaseout python dependency attrs and decorator #411
• [Language] make linter and type checker happy with mocking by @YouJiacheng in [Language] make linter and type checker happy with mocking #407
• [Bugfix] Support larger than 256 box size tma copy by @LeiWang1999 in [Bugfix] Support larger than 256 box size tma copy #413
• [Enhancement] Add get_nvcc_compiler function to retrieve nvcc path by @LeiWang1999 in [Enhancement] Add get_nvcc_compiler function to retrieve nvcc path #414
• Update lower.py to set default value for params by @Alex4210987 in Update lower.py to set default value for params #416
• [Enhancement] Support Auto Layout Inference and Parallelism with variable constraint by @LeiWang1999 in [Enhancement] Support Auto Layout Inference and Parallelism with variable constraint #417
• [Enhancement] Support to find Cython path more automatically by @FrozenGene in [Enhancement] Support to find Cython path more automatically #418
• [Refactor] Enhance layout inference logic in ParallelOp by @chengyupku in [Refactor] Enhance layout inference logic in ParallelOp #420
• [BugFix] Fix tvm simplify pass by @smallscientist1 in [BugFix] Fix tvm simplify pass #421
• [Enhancement] Add TMA+WS support in pipeline planning logic by @chengyupku in [Enhancement] Add TMA+WS support in pipeline planning logic #422
• [Language] Support tile operator T.cumsum by @LeiWang1999 in [Language] Support tile operator T.cumsum #423
• Delete testing/python/language/test_tilelang_language_reduce_sum.py by @LeiWang1999 in Delete testing/python/language/test_tilelang_language_reduce_sum.py #424
• [Bugfix] Fix a bug for simplifier by @LeiWang1999 in [Bugfix] Fix a bug for simplifier #425
• [Layout] Enhance layout inference pass by @LeiWang1999 in [Layout] Enhance layout inference pass #427
• [Enhancement] Remove DeReplicate during parallel loop layout inference by @LeiWang1999 in [Enhancement] Remove DeReplicate during parallel loop layout inference #430
• [Bugfix] Fix the test data distribution of cumsum by @LeiWang1999 in [Bugfix] Fix the test data distribution of cumsum #432
• [Enhancement] Support cute mma tile mxn8ky by @LeiWang1999 in [Enhancement] Support cute mma tile mxn8ky #434
• [Bugfix] Removed the behavior that treated global -> local as a copy operation. by @LeiWang1999 in [Bugfix] Removed the behavior that treated global -> local as a copy operation. #435
• [Language] Support accumulative T.reduce_sum by @LeiWang1999 in [Language] Support accumulative T.reduce_sum #436
• [Bugfix] fix the unexpected keyword error of autotune by @yyttt6 in [Bugfix] fix the unexpected keyword error of autotune #438
• [Testing] Add atomic add test by @LeiWang1999 in [Testing] Add atomic add test  #439
• [Typo] Rename warp_source to wrap_source by @lucifer1004 in [Typo] Rename warp_source to wrap_source #440
• [Refactor] Update KernelLaunch to clarify block name by @LeiWang1999 in [Refactor] Update KernelLaunch to clarify block name #441
• [Enhancement] Reduce CPU overhead during kernel execution by @Cunxiao2002 in [Enhancement] Reduce CPU overhead during kernel execution #437
• [Enhancement] Improve layout inference accuracy in ParallelOp by @LeiWang1999 in [Enhancement] Improve layout inference accuracy in ParallelOp #442
• [Bugfix] Fix layout inference for free fragment buffer by @LeiWang1999 in [Bugfix] Fix layout inference for free fragment buffer #443
• Bump transformers from 4.48.0 to 4.50.0 in /examples/bitnet-1.58b by @dependabot in Bump transformers from 4.48.0 to 4.50.0 in /examples/bitnet-1.58b #444
• [Language] Support explicit programming for identified warp groups by @LeiWang1999 in [Language] Support explicit programming for identified warp groups  #445
• [Bugfix] Fix safe memory legalization for fragment store by @LeiWang1999 in [Bugfix] Fix safe memory legalization for fragment store #446
• [Refactor] Separate warp specialize rewriter and tma barrier injector pass by @LeiWang1999 in [Refactor] Separate warp specialize rewriter and tma barrier injector pass #447
• [Enhancement] Add new examples for warp specialization and TMA integration by @LeiWang1999 in [Enhancement] Add new examples for warp specialization and TMA integration  #448
• [Refactor] Phaseout torch>=2.2.0 dependency by @LeiWang1999 in [Refactor] Phaseout torch>=2.2.0 dependency #451
• [Feature] Add TILELANG_CHECK_LAST_ERROR macro for improved error handling in CUDA and HIP by @LeiWang1999 in [Feature] Add TILELANG_CHECK_LAST_ERROR macro for improved error handling in CUDA and HIP #450
• [Enhancement] Introduce pass_configs parameter for kernel Caching by @LeiWang1999 in [Enhancement] Introduce pass_configs parameter for kernel Caching #452
• [Feature] Add cache directory management functions in tilelang.cache by @LeiWang1999 in [Feature] Add cache directory management functions in tilelang.cache #453
• [Bugfix] Fix get_swizzle_layout implementation. by @cherichy in [Bugfix] Fix get_swizzle_layout implementation. #455
• [Refactor] Update barrier functions and add new example for GEMM with warp specialization by @LeiWang1999 in [Refactor] Update barrier functions and add new example for GEMM with warp specialization #456
• [Refactor] Include examples in CI by @LeiWang1999 in [Refactor] Include examples in CI #457
• docs: add llvm version info to installation.md. by @AsakusaRinne in docs: add llvm version info to installation.md. #459
• [CI] Add elementwise and gemv examples to CI. by @Cunxiao2002 in [CI] Add elementwise and gemv examples to CI. #458
• [Bugfix] Fix for T.copy with dynamic range by @LeiWang1999 in [Bugfix] Fix for T.copy with dynamic range  #462
• [Bugfix] Fix copy region automation for dynamic extent by @LeiWang1999 in [Bugfix] Fix copy region automation for dynamic extent #465
• [Feature] Implement fast integer power operation and related API by @LeiWang1999 in [Feature] Implement fast integer power operation and related API #466
• [Typo] Rename power_of_int with pow_of_int for consistency by @LeiWang1999 in [Typo] Rename power_of_int with pow_of_int for consistency #468
• [CI] Add BlocksparseGemm, Dynamic, and Cast examples to CI by @tzj-fxz in [CI] Add BlocksparseGemm, Dynamic, and Cast examples to CI #467
• [Refactor] Update set_compile_args to allow None for out_idx parameter by @LeiWang1999 in [Refactor] Update set_compile_args to allow None for out_idx parameter #469
• [Refactor] Simplify buffer_region_to_tile_region function in copy.py by @LeiWang1999 in [Refactor] Simplify buffer_region_to_tile_region function in copy.py #470
• [CI] Add Convolution example to CI by @xwhzz in [CI] Add Convolution example to CI #473
• [BugFix] Correct argparse for example_convolution test by @xwhzz in [BugFix] Correct argparse for example_convolution test #474
• [Refactor] set USE_LLVM to optional. by @hyx1999 in [Refactor] set USE_LLVM to optional. #476
• [CI] Add Analyzer and blocksparse_attention examples to CI by @yyttt6 in [CI] Add Analyzer and blocksparse_attention examples to CI #472
• [Refactor] Skip patchelf if not installed by @LeiWang1999 in [Refactor] Skip patchelf if not installed #477
• [Refactor] Improve layout equality checks and error messaging by @LeiWang1999 in [Refactor] Improve layout equality checks and error messaging #471
• [Doc] Update version retrieval in conf.py to read from VERSION file by @xwhzz in [Doc] Update version retrieval in conf.py to read from VERSION file #478
• Fix Device Consistency in Autotuner Threads and Add Manual Profiler Check by @yuanjypku in Fix Device Consistency in Autotuner Threads and Add Manual Profiler Check #481
• [Bugfix] Check CUDA target before checking for TMA by @gau-nernst in [Bugfix] Check CUDA target before checking for TMA #482
• [Bugfix] Use AutoTune cache_input_tensors properly by @yyttt6 in [Bugfix] Use AutoTune cache_input_tensors properly #483
• Revert "[Bugfix] Use AutoTune cache_input_tensors properly" by @LeiWang1999 in Revert "[Bugfix] Use AutoTune cache_input_tensors properly" #488
• [Enhancement] Support register input for gemm when trans_a or trans_b is true by @LeiWang1999 in [Enhancement] Support register input for gemm when trans_a or trans_b is true #490
• [CI] Add flash_decoding example to CI by @xuchangtolearn in [CI] Add flash_decoding example to CI #487
• [CI] Add Reminder Bot for pull request contributions by @xwhzz in [CI] Add Reminder Bot for pull request contributions #491
• [Refactor] Introduce quantize components of TileLang and add testing for dequant gemm exmaple by @LeiWang1999 in [Refactor] Introduce quantize components of TileLang and add testing for dequant gemm exmaple #494
• [Enhancement] Introduce flag to visualize shared memory merge plan by @LeiWang1999 in [Enhancement] Introduce flag to visualize shared memory merge plan #496
• [Refactor] Update main function structure in example scripts and add tests by @chengyupku in [Refactor] Update main function structure in example scripts and add tests #475
• [Bugfix] Fix Hopper GEMM layout for small tile size by @LeiWang1999 in [Bugfix] Fix Hopper GEMM layout for small tile size #497
• [Enhancement] Fallback transposed_ldmatrix into SM75_U16x4_LDSM_N when warp_n is 8 by @LeiWang1999 in [Enhancement] Fallback transposed_ldmatrix into SM75_U16x4_LDSM_N when warp_n is 8 #498
• [Bugfix] Rename SM75_U16x8_LDSM_N to SM75_U16x8_LDSM_T to reflect correct matrix type by @LeiWang1999 in [Bugfix] Rename SM75_U16x8_LDSM_N to SM75_U16x8_LDSM_T to reflect correct matrix type  #499
• [Refactor] Update GEMM layout and operand traits for improved CUDA compatibility by @LeiWang1999 in [Refactor] Update GEMM layout and operand traits for improved CUDA compatibility #500
• [Refactor] Update JIT kernel functions and streamline GEMM tests by @LeiWang1999 in [Refactor] Update JIT kernel functions and streamline GEMM tests #501
• Fix AMD Docker issues related to conda environment setup by @Hamerlate in Fix AMD Docker issues related to conda environment setup #503
• [Refactor] Refactor jit to _JitImplementation to support @tilelang.jit by @LeiWang1999 in [Refactor] Refactor jit to _JitImplementation to support @tilelang.jit #502
• [Refactor] Adjust in fragment GEMM layout by @LeiWang1999 in [Refactor] Adjust in fragment GEMM layout #504
• [Refactor] Update GlobalMemChecker to Detect Lower Bound illegal memory access automatically by @LeiWang1999 in [Refactor] Update GlobalMemChecker to Detect Lower Bound illegal memory access automatically #505
• [Enhancement] Enhance ReduceOp and JITKernel for improved dimension handling and initialization by @LeiWang1999 in [Enhancement] Enhance ReduceOp and JITKernel for improved dimension handling and initialization #507
• [Refactor] Update buffer handling in layout transformation to support layout on T.view by @LeiWang1999 in [Refactor] Update buffer handling in layout transformation to support layout on T.view  #509
• [Bugfix] Enhance smem copy selector for uncommon shape by @LeiWang1999 in [Bugfix] Enhance smem copy selector for uncommon shape #510
• [Enhancement] Introduce padding annotation and improve legalize safe memory access pass by @LeiWang1999 in [Enhancement] Introduce padding annotation and improve legalize safe memory access pass #511
• [Refactor] Enhance MergeSharedMemoryAllocations Pass for Improved Liveness Analysis and Scope Management by @LeiWang1999 in [Refactor] Enhance MergeSharedMemoryAllocations Pass for Improved Liveness Analysis and Scope Management #508
• [Dev] Add grouped GEMM example with TileLang and PyTorch integration by @chengyupku in [Dev] Add grouped GEMM example with TileLang and PyTorch integration #514
• [Dev] Add grouped GEMM backward example scripts by @chengyupku in [Dev] Add grouped GEMM backward example scripts #515
• Fix deepgemm exmaple by @benenzhu in Fix deepgemm exmaple  #513
• [Refactor] Support auto index bitwidth casting by @LeiWang1999 in [Refactor] Support auto index bitwidth casting #517
• [Enhancement] Support auto synchronization for global memory access by @LeiWang1999 in [Enhancement] Support auto synchronization for global memory access #519
• [Refactor] Replace default fp8 dtype with cute to perform fast cast by @LeiWang1999 in [Refactor] Replace default fp8 dtype with cute to perform fast cast #520
• [Enhancement] Add atomicAdd for FLOAT16x2 and FLOAT16x4 by @LeiWang1999 in [Enhancement] Add atomicAdd for FLOAT16x2 and FLOAT16x4 #522
• [Refactor] Reorganize Thread Synchronization Steps to make sure global synchronization can be correctly lowered by @LeiWang1999 in [Refactor] Reorganize Thread Synchronization Steps to make sure global synchronization can be correctly lowered #521
• [Enhancement] Add commit ID to versioning and improve logging initialization by @LeiWang1999 in [Enhancement] Add commit ID to versioning and improve logging initialization #524
• [Enhancement] Add warp specialization attribute handling in IR and rewriter by @chengyupku in [Enhancement] Add warp specialization attribute handling in IR and rewriter #518
• [CI] Add gemm and gemm_fp8 example to CI by @LeslinD in [CI] Add gemm and gemm_fp8 example to CI #516
• [Refactor] Refactor convolution example to streamline configuration and remove unused code by @LeiWang1999 in [Refactor] Refactor convolution example to streamline configuration and remove unused code #530
• [Autotune] Introduce cache mechanism for auto tuner by @LeiWang1999 in [Autotune] Introduce cache mechanism for auto tuner #527
• [Refactor]: add autotune example to convolution examples by @yyttt6 in [Refactor]: add autotune example to convolution examples #536
• [Refactor] Disable legacy vectorization for buffer allocation by @LeiWang1999 in [Refactor] Disable legacy vectorization for buffer allocation #535
• [Language] Support T.annotate_l2_hit_ratio via cudaStreamSetAttribute by @LeiWang1999 in [Language] Support T.annotate_l2_hit_ratio via cudaStreamSetAttribute #539
• [Bugfix] Fix a bug when simplifying warp combination for T.gemm by @LeiWang1999 in [Bugfix] Fix a bug when simplifying warp combination for T.gemm #540
• [AMD] Support float8 matrix core by @LeiWang1999 in [AMD] Support float8 matrix core #537
• [Doc] Include DeepWiki badge in README by @xwhzz in [Doc] Include DeepWiki badge in README #541
• [CI] Add hadamard example to CI by @Rachmanino in [CI] Add hadamard example to CI #549
• [chore] set default build type to release if not provided by @botbw in [chore] set default build type to release if not provided #548
• [Refactor] Include several examples into ci by @LeiWang1999 in [Refactor] Include several examples into ci #531
• [AMD][Enhancement] Add support for Vectorized FP8 DataPacking by @LeiWang1999 in [AMD][Enhancement] Add support for Vectorized FP8 DataPacking #542
• [Bugfix] Enhance layout inference pass for flexibility by @LeiWang1999 in [Bugfix] Enhance layout inference pass for flexibility  #550
• [Autotune] Remove the out_idx argument from the autotune cache by @LeiWang1999 in [Autotune] Remove the out_idx argument from the autotune cache #553
• [CI] Add linear attention examples to CI by @Rachmanino in [CI] Add linear attention examples to CI #552
• [CI]Add norm and layout_plot by @Alex4210987 in [CI]Add norm and layout_plot #534

New Contributors

• @FrozenGene made their first contribution in [Enhancement] Support to find Cython path more automatically #418
• @lucifer1004 made their first contribution in [Typo] Rename warp_source to wrap_source #440
• @AsakusaRinne made their first contribution in docs: add llvm version info to installation.md. #459
• @yuanjypku made their first contribution in Fix Device Consistency in Autotuner Threads and Add Manual Profiler Check #481
• @gau-nernst made their first contribution in [Bugfix] Check CUDA target before checking for TMA #482
• @xuchangtolearn made their first contribution in [CI] Add flash_decoding example to CI #487
• @Hamerlate made their first contribution in Fix AMD Docker issues related to conda environment setup #503
• @benenzhu made their first contribution in Fix deepgemm exmaple  #513
• @Rachmanino made their first contribution in [CI] Add hadamard example to CI #549

Full Changelog: v0.1.4...v0.1.5

---

This discussion was created from the release v0.1.5.