- 12 Mar, 2025 3 commits
-
-
penguin_wwy authored
-
Lei Wang authored
* Optimize CMake build process with dynamic job count calculation - Modify build_csrc function to use 90% of available CPU cores - Ensure at least one job is used during compilation - Improve build performance by dynamically adjusting parallel job count * Optimize build_csrc function with multiprocessing module - Replace os.cpu_count() with multiprocessing.cpu_count() - Maintain existing 90% CPU utilization logic - Improve CPU core count calculation for build process * Add dynamic shape support with out_idx in Cython JIT kernel compilation - Implement `run_cython_dynamic_shape_with_out_idx` function in test_tilelang_jit_gemm_cython.py - Update Cython wrapper to handle dynamic symbolic shapes during tensor allocation - Add support for resolving dynamic shape dimensions using input tensor references - Enhance flexibility of JIT kernel compilation with symbolic shape handling * Enhance error reporting for dynamic symbolic shape resolution in Cython JIT kernel - Add detailed error message when a dynamic symbolic dimension is not found in dynamic_symbolic_map - Improve debugging by providing context about missing symbolic dimensions - Maintain existing dynamic shape resolution logic * Fix Copy operation handling for scalar and multi-dimensional tensors - Add special handling for scalar tensor copy operations - Enhance error reporting in MakeIndices method with more detailed diagnostic information - Improve SIMT loop generation to support zero-dimensional tensors - Add explicit check and handling for scalar tensor scenarios * Refactor Copy operation code formatting and improve readability - Improve code formatting in MakeIndices and MakeSIMTLoop methods - Add line breaks to enhance readability of complex ICHECK statements - Simplify code structure in scalar tensor handling - Remove unnecessary whitespace and improve code alignment * Simplify GEMM example with direct kernel compilation - Update copyright header to Tile-AI Corporation - Remove Profiler import and usage - Replace tilelang.lower() with tilelang.compile() - Simplify kernel execution workflow - Update kernel source retrieval method
-
Lei Wang authored
* Optimize CMake build process with dynamic job count calculation - Modify build_csrc function to use 90% of available CPU cores - Ensure at least one job is used during compilation - Improve build performance by dynamically adjusting parallel job count * Optimize build_csrc function with multiprocessing module - Replace os.cpu_count() with multiprocessing.cpu_count() - Maintain existing 90% CPU utilization logic - Improve CPU core count calculation for build process * Add dynamic shape support with out_idx in Cython JIT kernel compilation - Implement `run_cython_dynamic_shape_with_out_idx` function in test_tilelang_jit_gemm_cython.py - Update Cython wrapper to handle dynamic symbolic shapes during tensor allocation - Add support for resolving dynamic shape dimensions using input tensor references - Enhance flexibility of JIT kernel compilation with symbolic shape handling * Enhance error reporting for dynamic symbolic shape resolution in Cython JIT kernel - Add detailed error message when a dynamic symbolic dimension is not found in dynamic_symbolic_map - Improve debugging by providing context about missing symbolic dimensions - Maintain existing dynamic shape resolution logic * Fix Copy operation handling for scalar and multi-dimensional tensors - Add special handling for scalar tensor copy operations - Enhance error reporting in MakeIndices method with more detailed diagnostic information - Improve SIMT loop generation to support zero-dimensional tensors - Add explicit check and handling for scalar tensor scenarios * Refactor Copy operation code formatting and improve readability - Improve code formatting in MakeIndices and MakeSIMTLoop methods - Add line breaks to enhance readability of complex ICHECK statements - Simplify code structure in scalar tensor handling - Remove unnecessary whitespace and improve code alignment
-
- 11 Mar, 2025 4 commits
-
-
penguin_wwy authored
-
Yu Cheng authored
* [Dev][Bugfix] Add RMS Normalization Kernels and Fix Reduce Bug - Implement two RMS normalization implementations in TileLang: * `rms_norm_splitk`: Split-K reduction approach for large matrices * `rms_norm`: Full reduction kernel with simplified implementation - Add reference implementation using PyTorch for validation - Include performance benchmarking for both kernel variants - Demonstrate flexible block size and matrix size configurations * [Examples] Simplify RMS Normalization Kernel Compilation - Remove commented-out code for split-K RMS normalization - Simplify kernel compilation by removing explicit TMA lowering configuration - Update copyright header to Tile-AI Corporation - Streamline main script for RMS normalization example
-
Lei Wang authored
* Optimize CMake build process with dynamic job count calculation - Modify build_csrc function to use 90% of available CPU cores - Ensure at least one job is used during compilation - Improve build performance by dynamically adjusting parallel job count * Optimize build_csrc function with multiprocessing module - Replace os.cpu_count() with multiprocessing.cpu_count() - Maintain existing 90% CPU utilization logic - Improve CPU core count calculation for build process * Add dynamic shape support with out_idx in Cython JIT kernel compilation - Implement `run_cython_dynamic_shape_with_out_idx` function in test_tilelang_jit_gemm_cython.py - Update Cython wrapper to handle dynamic symbolic shapes during tensor allocation - Add support for resolving dynamic shape dimensions using input tensor references - Enhance flexibility of JIT kernel compilation with symbolic shape handling * Enhance error reporting for dynamic symbolic shape resolution in Cython JIT kernel - Add detailed error message when a dynamic symbolic dimension is not found in dynamic_symbolic_map - Improve debugging by providing context about missing symbolic dimensions - Maintain existing dynamic shape resolution logic
-
Lei Wang authored
* Optimize CMake build process with dynamic job count calculation - Modify build_csrc function to use 90% of available CPU cores - Ensure at least one job is used during compilation - Improve build performance by dynamically adjusting parallel job count * Optimize build_csrc function with multiprocessing module - Replace os.cpu_count() with multiprocessing.cpu_count() - Maintain existing 90% CPU utilization logic - Improve CPU core count calculation for build process
-
- 10 Mar, 2025 3 commits
-
-
Lei Wang authored
* Update native sparse attention example with scale parameter handling - Add scale parameter processing in native_sparse_attention function - Modify example script to include custom scale value - Update function calls to pass scale parameter - Enhance flexibility of sparse attention implementation * Refactor Triton Native Sparse Attention Example - Improve code formatting and readability in example_triton_nsa_bwd.py - Standardize function and parameter alignment - Remove unnecessary whitespaces and optimize imports - Enhance code style consistency with previous commits
-
Lei Wang authored
* [Refactor] Improve Thread Variable Handling in Layout Inference - Update layout inference to handle thread variables more robustly - Add explicit size check between infer_list_ and thread_var_vec_ - Modify thread variable access to use per-iteration thread variable - Simplify thread predicate retrieval logic - Add minor code cleanup and return variable assignment * [Refactor] Update Layout Inference Copyright and Simplify Return Logic - Replace Apache License header with Microsoft Corporation copyright notice - Simplify LayoutInference function by directly returning substituted function - Remove unnecessary variable assignment in return statement * [Refactor] Update Layout Inference Copyright to Tile-AI Corporation - Change copyright notice from Microsoft Corporation to Tile-AI Corporation - Maintain existing file structure and licensing header
-
Lei Wang authored
- Introduce `CreateEnvThread` function to generate environment threads for GPU kernel launches - Modify `KernelLaunch` to use `CreateEnvThread` for block and thread indices - Improve thread variable naming with shorter, more descriptive identifiers (bx, by, bz, tx, ty, tz) - Ensure proper thread environment setup within PrimFunc context
-
- 09 Mar, 2025 4 commits
-
-
Lei Wang authored
* Add kernel caching mechanism to TileLang - Implement a new `cached` function in `tilelang/cache/__init__.py` to cache and reuse compiled kernels - Expose the `cached` function in the main `tilelang/__init__.py` - Add a test case for cached matrix multiplication in `testing/python/cache/test_tilelang_cache_matmul.py` - Provide a `clear_cache()` function to reset the kernel cache when needed * Refactor kernel caching test and implementation - Simplify the `cached` function in `tilelang/cache/__init__.py` - Update test script `test_tilelang_cache_matmul.py` to use `tilelang.testing.main()` - Remove unnecessary whitespace and improve code formatting * Update import for `cached` function in MHA examples - Modify import statement in `example_mha_bwd.py` and `test_tilelang_kernel_mha_bwd.py` - Change import from `tilelang.profiler import cached` to `tilelang import cached` - Align with recent refactoring of kernel caching mechanism * Refactor `cached` function signature in kernel caching - Update function signature to use keyword-only arguments for `target` and `target_host` - Improve parameter order and readability of the `cached` decorator - Maintain existing functionality while enhancing function definition
-
Lei Wang authored
* Add TMA lowering configuration option and update copyright notices This commit introduces a new configuration option to disable TMA (Tensor Memory Access) lowering and updates copyright notices across multiple files. Key changes include: - Add `kDisableTMALower` configuration option in builtin.h and builtin.cc - Update copyright notices from Microsoft Corporation to Tile-AI Corporation - Modify `LowerArgs` struct to include `disable_tma_lower` flag - Update JIT compilation interfaces to support pass configuration - Enhance error reporting in bulk copy lowering - Propagate pass configuration through various adapter layers * lint fix
-
Lei Wang authored
* Improve Autotuner and CUDA Compatibility for Tensor Core Policies - Enhance autotuner with robust parallel compilation and error handling - Add logging for better debugging during configuration compilation - Support SM90 compute capabilities in TensorCore and matmul analysis policies - Improve future handling and result tracking in autotuner - Add more flexible SM version checks for pipeline and async copy stages * Refactor Autotuner Parallel Compilation with Improved Error Handling - Enhance tqdm progress bar formatting for concurrent configuration compilation - Simplify exception handling in parallel compilation process - Remove unnecessary logging and improve code readability - Optimize thread pool shutdown and result processing
-
Lei Wang authored
* [Refactor] Update BitBLAS Benchmark with TileLang Carver Imports and Roller Hints Generation - Replace BitBLAS imports with TileLang Carver imports in benchmark_matmul.py - Modify roller hints generation using new TileLang Carver template and utility functions - Update get_roller_hints_from_func to handle None cases and improve return logic - Adjust DefaultPolicy to handle different codegen dictionary formats * [Refactor] Update Thread Binding and Import Statements in TileLang Kernels - Replace T.thread_binding() with T.get_thread_binding() across multiple kernel test files - Update import statements for MMA layout and macro generator in dequantize GEMM and FP8 examples - Move map_torch_type utility function to tilelang.utils.tensor - Remove unnecessary imports and improve code organization * Refactor Native Sparse Attention Example with Enhanced Triton Kernel - Update parallel_nsa_fwd_kernel to support more flexible sparse attention computation - Add support for block counts and offsets in the Triton kernel - Modify kernel grid and computation logic for improved performance - Update example script to use naive_nsa_simple reference implementation - Improve type hints and kernel configuration * Add Native Sparse Attention Examples with Tilelang and Triton Implementations - Introduce new example scripts for native sparse attention: * example_tilelang_nsa_fwd.py: Forward pass implementation using TileLang * example_tilelang_nsa_decode.py: Decoding-specific sparse attention implementation * example_triton_nsa_fwd.py: Triton-based sparse attention forward pass - Update reference.py with naive implementations for sparse attention - Support different sparse attention scenarios including forward pass and inference - Add comprehensive testing and validation against reference implementations * lint fix * Add Variable-Length Native Sparse Attention Examples for TileLang and Triton - Introduce new example scripts for variable-length native sparse attention: * example_tilelang_nsa_fwd_varlen.py: TileLang implementation with variable sequence lengths * example_triton_nsa_fwd_varlen.py: Triton implementation with variable sequence lengths - Update reference.py to support variable-length sparse attention scenarios - Enhance existing sparse attention implementations to handle variable-length inputs - Add comprehensive testing and validation for variable-length sparse attention * Refactor Native Sparse Attention Examples: Code Style and Formatting Improvements - Standardize function and parameter formatting across NSA example files - Improve code readability by adjusting indentation and line breaks - Enhance type hints and parameter alignment - Remove unnecessary whitespaces and optimize imports - Maintain consistent code style across TileLang and Triton implementations * Add debug logging and extend execution backend in JIT and loop vectorization - Add detailed logging in loop vectorization to help diagnose buffer shape handling - Extend JIT execution backend to include 'cython' option - Improve boundary condition checks in BufferLoadNode visit method * Remove debug logging in loop vectorization BufferLoadNode visit method - Remove unnecessary INFO log statements in VisitExpr_ method - Simplify code by eliminating redundant logging - Maintain core logic for handling buffer load node visits
-
- 07 Mar, 2025 7 commits
-
-
Lei Wang authored
* [Refactor] Update BitBLAS Benchmark with TileLang Carver Imports and Roller Hints Generation - Replace BitBLAS imports with TileLang Carver imports in benchmark_matmul.py - Modify roller hints generation using new TileLang Carver template and utility functions - Update get_roller_hints_from_func to handle None cases and improve return logic - Adjust DefaultPolicy to handle different codegen dictionary formats * [Refactor] Update Thread Binding and Import Statements in TileLang Kernels - Replace T.thread_binding() with T.get_thread_binding() across multiple kernel test files - Update import statements for MMA layout and macro generator in dequantize GEMM and FP8 examples - Move map_torch_type utility function to tilelang.utils.tensor - Remove unnecessary imports and improve code organization * Refactor Native Sparse Attention Example with Enhanced Triton Kernel - Update parallel_nsa_fwd_kernel to support more flexible sparse attention computation - Add support for block counts and offsets in the Triton kernel - Modify kernel grid and computation logic for improved performance - Update example script to use naive_nsa_simple reference implementation - Improve type hints and kernel configuration * Add Native Sparse Attention Examples with Tilelang and Triton Implementations - Introduce new example scripts for native sparse attention: * example_tilelang_nsa_fwd.py: Forward pass implementation using TileLang * example_tilelang_nsa_decode.py: Decoding-specific sparse attention implementation * example_triton_nsa_fwd.py: Triton-based sparse attention forward pass - Update reference.py with naive implementations for sparse attention - Support different sparse attention scenarios including forward pass and inference - Add comprehensive testing and validation against reference implementations * lint fix * Add Variable-Length Native Sparse Attention Examples for TileLang and Triton - Introduce new example scripts for variable-length native sparse attention: * example_tilelang_nsa_fwd_varlen.py: TileLang implementation with variable sequence lengths * example_triton_nsa_fwd_varlen.py: Triton implementation with variable sequence lengths - Update reference.py to support variable-length sparse attention scenarios - Enhance existing sparse attention implementations to handle variable-length inputs - Add comprehensive testing and validation for variable-length sparse attention * Refactor Native Sparse Attention Examples: Code Style and Formatting Improvements - Standardize function and parameter formatting across NSA example files - Improve code readability by adjusting indentation and line breaks - Enhance type hints and parameter alignment - Remove unnecessary whitespaces and optimize imports - Maintain consistent code style across TileLang and Triton implementations
-
You Jiacheng authored
It's slightly faster than T.copy then RS-GEMM, and simpler.
-
Lei Wang authored
-
Lei Wang authored
* [Refactor] Update BitBLAS Benchmark with TileLang Carver Imports and Roller Hints Generation - Replace BitBLAS imports with TileLang Carver imports in benchmark_matmul.py - Modify roller hints generation using new TileLang Carver template and utility functions - Update get_roller_hints_from_func to handle None cases and improve return logic - Adjust DefaultPolicy to handle different codegen dictionary formats * [Refactor] Update Thread Binding and Import Statements in TileLang Kernels - Replace T.thread_binding() with T.get_thread_binding() across multiple kernel test files - Update import statements for MMA layout and macro generator in dequantize GEMM and FP8 examples - Move map_torch_type utility function to tilelang.utils.tensor - Remove unnecessary imports and improve code organization * Refactor Native Sparse Attention Example with Enhanced Triton Kernel - Update parallel_nsa_fwd_kernel to support more flexible sparse attention computation - Add support for block counts and offsets in the Triton kernel - Modify kernel grid and computation logic for improved performance - Update example script to use naive_nsa_simple reference implementation - Improve type hints and kernel configuration * Add Native Sparse Attention Examples with Tilelang and Triton Implementations - Introduce new example scripts for native sparse attention: * example_tilelang_nsa_fwd.py: Forward pass implementation using TileLang * example_tilelang_nsa_decode.py: Decoding-specific sparse attention implementation * example_triton_nsa_fwd.py: Triton-based sparse attention forward pass - Update reference.py with naive implementations for sparse attention - Support different sparse attention scenarios including forward pass and inference - Add comprehensive testing and validation against reference implementations * lint fix
-
Lei Wang authored
* [Refactor] Update BitBLAS Benchmark with TileLang Carver Imports and Roller Hints Generation - Replace BitBLAS imports with TileLang Carver imports in benchmark_matmul.py - Modify roller hints generation using new TileLang Carver template and utility functions - Update get_roller_hints_from_func to handle None cases and improve return logic - Adjust DefaultPolicy to handle different codegen dictionary formats * [Refactor] Update Thread Binding and Import Statements in TileLang Kernels - Replace T.thread_binding() with T.get_thread_binding() across multiple kernel test files - Update import statements for MMA layout and macro generator in dequantize GEMM and FP8 examples - Move map_torch_type utility function to tilelang.utils.tensor - Remove unnecessary imports and improve code organization * Refactor Native Sparse Attention Example with Enhanced Triton Kernel - Update parallel_nsa_fwd_kernel to support more flexible sparse attention computation - Add support for block counts and offsets in the Triton kernel - Modify kernel grid and computation logic for improved performance - Update example script to use naive_nsa_simple reference implementation - Improve type hints and kernel configuration
-
Lei Wang authored
* [Refactor] Update BitBLAS Benchmark with TileLang Carver Imports and Roller Hints Generation - Replace BitBLAS imports with TileLang Carver imports in benchmark_matmul.py - Modify roller hints generation using new TileLang Carver template and utility functions - Update get_roller_hints_from_func to handle None cases and improve return logic - Adjust DefaultPolicy to handle different codegen dictionary formats * [Refactor] Update Thread Binding and Import Statements in TileLang Kernels - Replace T.thread_binding() with T.get_thread_binding() across multiple kernel test files - Update import statements for MMA layout and macro generator in dequantize GEMM and FP8 examples - Move map_torch_type utility function to tilelang.utils.tensor - Remove unnecessary imports and improve code organization
-
Wenhao Xie authored
* [Typo] Fix formatting in installation instructions in README.md * [Enhancement] Improve CUDA path detection and update configuration handling * fix typo * remove IS_WINDOWS constant * lint fix * Improve error messages for CUDA detection failure * lint fix * lint fix * Fix .gitignore to correctly include venv directory
-
- 06 Mar, 2025 8 commits
-
-
Lei Wang authored
Simplify the control flow in the MLA decode kernel by replacing TileLang's T.If construct with a standard Python if statement. This change improves code readability and maintains the existing logic for handling sequence length constraints during block-wise computation.
-
Chaofan Lin authored
* [Carver] Multi-Threads Compilation for Fast Auto Tuning * Add progress bar for compilation * lint
-
xs-keju authored
* Add cpu jit with backend ctypes * Resolve some lint issues * Apply PR feedback on head file and kernel example * Add test cases * Resolve formatting issues * Resolve formatting issues --------- Co-authored-by:xxw <1990389406@qq.con>
-
Lei Wang authored
Update Dockerfiles for CUDA 118, 120, 121, 123, 124, 125, and 126 to install libstdcxx-ng-12 from conda-forge, ensuring consistent standard library support across different CUDA versions
-
Lei Wang authored
* Remove Torch CPP backend and update execution backend options - Remove TorchCPPKernelAdapter and related code from JIT modules - Update execution backend options in jit/__init__.py, kernel.py, and adapter/__init__.py - Remove "torch_cpp" from supported execution backend literals - Simplify backend validation and remove unused torch_cpp-related code 。 * lint fix * Add block sparse attention implementations for TileLang and Triton - Implement block sparse attention kernels for TileLang and Triton - Add example scripts for block sparse attention with top-k and threshold-based masking - Include utility functions for generating sparse attention masks - Demonstrate causal attention with block-level sparsity - Add test cases to validate sparse attention implementations against PyTorch reference * Bump version to 0.1.1 * Bump version to 0.1.2
-
Yu Cheng authored
* [Dev] Adjust computation logic to avoid precision loss when casting acc_s from float to float16 - Remove redundant `acc_s_0` fragment in flash attention kernel - Simplify memory copy and reduction operations - Reorder memory copy and scaling steps for improved performance - Add Hopper-specific synchronization method in CUDA reduce template - Update reduce operation to use architecture-specific synchronization * [Dev] Add DeepSeek MLA Decoding (Paged+Varlen) kernel and Performance Benchmark Script - Implement comprehensive MLA (Multi-Head Latent Attention) decoding benchmark script - Add support for multiple implementations: Torch, TileLang, FlashMLA, FlashInfer, and Triton - Create flexible configuration for benchmarking different batch sizes, sequence lengths, and head configurations - Implement performance comparison and CSV output for detailed performance analysis - Add command-line argument support for targeted benchmarking and comparison * [Dev] Refactor MLA Paged Decoding Kernel with Improved Block Handling and Precision - Replace `d` parameter with `dv` to clarify value dimension in MLA decoding - Enhance block distribution logic for split KV processing - Improve handling of remaining blocks in split KV computation - Add initialization of `lse_max_local` to prevent potential precision issues - Optimize block start and range calculations for more accurate sequence processing * lint
-
Lei Wang authored
* [Refactor] Consolidate GemmWarpPolicy Enum and Add Utility Method - Move GemmWarpPolicy from copy.py and gemm.py to primitives/gemm/base.py - Implement from_warp_partition class method to determine warp policy - Add docstring with examples for policy determination - Remove duplicate GemmWarpPolicy class definitions * [Enhancement] Add TensorCore Intrinsic Matrix Multiplication Benchmarks - Implement two new matrix multiplication benchmark scripts: 1. `benchmark_matmul_intrinsic.py`: Uses TensorCore intrinsics with advanced configuration 2. `benchmark_matmul.py`: Provides a more generic matrix multiplication benchmark - Add support for roller-based configuration generation in both benchmarks - Enhance MMA macro generator to handle 2D and 4D output buffer layouts - Implement flexible autotuning configurations with multiple parameters - Support different data types and accumulation modes - Add command-line arguments for matrix dimensions and roller configuration * lint fix * Fix roller hints generation in get_roller_hints_from_func - Simplify roller hints generation logic - Ensure policy-based configuration is always emitted when a policy is available - Remove redundant None check for roller hints * Add shared memory for matrix multiplication in benchmark and quickstart examples - Modify benchmark_matmul.py and quickstart.py to include C_shared allocation - Change accumulation dtype from float16 to float in benchmark_matmul.py - Update matrix multiplication kernels to use shared memory for result storage - Enable CUDA kernel source printing in quickstart example
-
Lei Wang authored
- Calculate 75% of available CPU cores for make jobs - Prevent system unresponsiveness during build - Dynamically adjust make job count based on system resources
-
- 05 Mar, 2025 6 commits
-
-
Chaofan Lin authored
-
Lei Wang authored
- Replace BitBLAS imports with TileLang Carver imports in benchmark_matmul.py - Modify roller hints generation using new TileLang Carver template and utility functions - Update get_roller_hints_from_func to handle None cases and improve return logic - Adjust DefaultPolicy to handle different codegen dictionary formats
-
Lei Wang authored
* Fix debug print buffer template for unsigned char type - Update debug_print_buffer_value template specialization for unsigned char - Modify test_tilelang_debug_print.py to include additional dtype tests - Add test case for uint8 dtype in debug print buffer function * Refactor debug print buffer template formatting for unsigned char - Improve code formatting for debug_print_buffer_value template specialization - Adjust line breaks and indentation for better readability - Maintain consistent code style with other template specializations * Extract map_torch_type utility function to tilelang.utils.tensor - Move map_torch_type function from multiple test files to a centralized location - Import map_torch_type from tilelang.utils.tensor in kernel test files - Improve code reusability by creating a shared utility function for type mapping * Add buffer dtype mapping for Cython kernel adapter - Introduce buffer_dtype_map in CythonKernelAdapter to track buffer variable dtypes - Add _process_buffer_dtype method to extract dtype information from TIR function - Update CythonKernelWrapper to support setting and validating buffer dtypes - Enhance type checking during kernel execution with dtype verification - Improve logging message for Cython JIT adapter compilation * Add static shape mapping for Cython kernel adapter - Introduce static_shape_map in CythonKernelAdapter to track buffer variable static shapes - Add _process_static_shape method to extract static shape information from TIR function - Update CythonKernelWrapper to support setting and validating static shapes - Enhance type checking during kernel execution with static shape verification * Add Multi-Head Attention (MHA) Backward Pass Test for TileLang Kernel - Implement comprehensive test for Multi-Head Attention backward pass - Support both causal and non-causal attention scenarios - Add reference implementation for comparing kernel outputs - Test different batch sizes, head counts, sequence lengths, and head dimensions - Verify forward and backward pass correctness using torch.testing.assert_close * Set random seed for MHA backward pass test - Add random seed initialization for consistent test reproducibility - Use tilelang.testing.set_random_seed(42) to ensure deterministic test results
-
Lei Wang authored
* Fix debug print buffer template for unsigned char type - Update debug_print_buffer_value template specialization for unsigned char - Modify test_tilelang_debug_print.py to include additional dtype tests - Add test case for uint8 dtype in debug print buffer function * Refactor debug print buffer template formatting for unsigned char - Improve code formatting for debug_print_buffer_value template specialization - Adjust line breaks and indentation for better readability - Maintain consistent code style with other template specializations
-
Lei Wang authored
* Change default log level from WARNING to INFO in TileLang initialization * Refactor Flash Attention Variable-Length MHA Example with Cython Backend Support - Update `example_mha_fwd_varlen.py` to use Cython backend for kernel compilation - Remove unused imports and simplify function signature - Modify `flashattn` function to handle max sequence length as a separate argument - Update kernel call to include max sequence length parameter - Improve code readability and remove commented-out code - Add print statement to confirm successful assertion * Refactor code formatting in TileLang lowering and example files - Improve line breaks and code formatting in `lower.py`, `wrapper.py`, and `tensor.py` - Simplify line breaks and reduce unnecessary whitespace - Enhance code readability by adjusting indentation and line breaks - Update example MHA forward pass script with cleaner tensor initialization * Update TileLang kernel test with import path changes for MMA layout and macro generator - Modify import statements in test_tilelang_kernel_dequantize_gemm.py - Replace bitblas imports with tilelang.intrinsics imports for MMA-related utilities - Update main function to use tilelang.testing.main() * Add Block Sparse Attention Examples for TileLang and Triton - Implement block sparse attention kernels for both TileLang and Triton - Add utility functions for generating sparse attention masks using top-k and threshold methods - Support causal and variable-length attention scenarios - Include test cases for different sequence length configurations - Demonstrate block-level sparse attention with configurable parameters * Refactor Block Sparse Attention Examples with Code Style Improvements - Improve code formatting in block_sparse_attn_tilelang.py and block_sparse_attn_triton.py - Enhance readability by adjusting line breaks and indentation - Simplify kernel and function calls with better formatting - Add whitespace and line break improvements for better code clarity * Enhance Layout Plotting with Multi-Replication and Dynamic Visualization - Update plot_layout function to support multiple replications in thread and value mapping - Improve thread and value mapping to handle replicated layouts - Dynamically adjust figure size and legend positioning - Add print statements for saved plot file paths - Modify example fragment_mma_load_a.py to uncomment and enable warp and block layout plotting * Refactor AtomicAdd functions in CUDA common header - Implement a generic template for AtomicAdd function - Specialize templates for half_t, bfloat16_t, and pointer types - Reorganize and clean up existing AtomicAdd implementations - Improve type handling and conversion in atomic operations * Remove unused import in MHA backward test file - Remove unnecessary argparse import from test_tilelang_kenrel_mha_bwd.py - Add blank line for improved code formatting - Minor code cleanup in test file * Add FP8 GEMM Example with TensorCore Intrinsics - Implement a new example for FP8 matrix multiplication using TensorCore intrinsics - Support E4M3 and E5M2 floating-point 8-bit data types - Add README with notes on current FP8 implementation limitations - Include correctness test for FP8 GEMM with different configurations - Demonstrate swizzle layout and pipeline optimizations for FP8 computation
-
Yu Cheng authored
[Dev] Adjust computation logic to avoid precision loss when casting acc_s from float to float16 (#141) - Remove redundant `acc_s_0` fragment in flash attention kernel - Simplify memory copy and reduction operations - Reorder memory copy and scaling steps for improved performance - Add Hopper-specific synchronization method in CUDA reduce template - Update reduce operation to use architecture-specific synchronization
-
- 04 Mar, 2025 3 commits
-
-
Yu Cheng authored
- Add non-split flash attention macro for more flexible kernel generation - Implement `main_no_split` function to handle single-split scenarios - Modify kernel selection logic to dynamically choose between split and non-split implementations
-
Lei Wang authored
* Change default log level from WARNING to INFO in TileLang initialization * Refactor Flash Attention Variable-Length MHA Example with Cython Backend Support - Update `example_mha_fwd_varlen.py` to use Cython backend for kernel compilation - Remove unused imports and simplify function signature - Modify `flashattn` function to handle max sequence length as a separate argument - Update kernel call to include max sequence length parameter - Improve code readability and remove commented-out code - Add print statement to confirm successful assertion * Refactor code formatting in TileLang lowering and example files - Improve line breaks and code formatting in `lower.py`, `wrapper.py`, and `tensor.py` - Simplify line breaks and reduce unnecessary whitespace - Enhance code readability by adjusting indentation and line breaks - Update example MHA forward pass script with cleaner tensor initialization * Update TileLang kernel test with import path changes for MMA layout and macro generator - Modify import statements in test_tilelang_kernel_dequantize_gemm.py - Replace bitblas imports with tilelang.intrinsics imports for MMA-related utilities - Update main function to use tilelang.testing.main() * Add Block Sparse Attention Examples for TileLang and Triton - Implement block sparse attention kernels for both TileLang and Triton - Add utility functions for generating sparse attention masks using top-k and threshold methods - Support causal and variable-length attention scenarios - Include test cases for different sequence length configurations - Demonstrate block-level sparse attention with configurable parameters * Refactor Block Sparse Attention Examples with Code Style Improvements - Improve code formatting in block_sparse_attn_tilelang.py and block_sparse_attn_triton.py - Enhance readability by adjusting line breaks and indentation - Simplify kernel and function calls with better formatting - Add whitespace and line break improvements for better code clarity * Enhance Layout Plotting with Multi-Replication and Dynamic Visualization - Update plot_layout function to support multiple replications in thread and value mapping - Improve thread and value mapping to handle replicated layouts - Dynamically adjust figure size and legend positioning - Add print statements for saved plot file paths - Modify example fragment_mma_load_a.py to uncomment and enable warp and block layout plotting * Refactor AtomicAdd functions in CUDA common header - Implement a generic template for AtomicAdd function - Specialize templates for half_t, bfloat16_t, and pointer types - Reorganize and clean up existing AtomicAdd implementations - Improve type handling and conversion in atomic operations * Remove unused import in MHA backward test file - Remove unnecessary argparse import from test_tilelang_kenrel_mha_bwd.py - Add blank line for improved code formatting - Minor code cleanup in test file
-
Yu Cheng authored
- Update news and MLA performance benchmark in README.md - Move performance benchmark and layout images to a dedicated 'figures' directory - Improve code formatting and image references in documentation
-
- 03 Mar, 2025 2 commits
-
-
Lei Wang authored
* Change default log level from WARNING to INFO in TileLang initialization * Refactor Flash Attention Variable-Length MHA Example with Cython Backend Support - Update `example_mha_fwd_varlen.py` to use Cython backend for kernel compilation - Remove unused imports and simplify function signature - Modify `flashattn` function to handle max sequence length as a separate argument - Update kernel call to include max sequence length parameter - Improve code readability and remove commented-out code - Add print statement to confirm successful assertion * Refactor code formatting in TileLang lowering and example files - Improve line breaks and code formatting in `lower.py`, `wrapper.py`, and `tensor.py` - Simplify line breaks and reduce unnecessary whitespace - Enhance code readability by adjusting indentation and line breaks - Update example MHA forward pass script with cleaner tensor initialization * Update TileLang kernel test with import path changes for MMA layout and macro generator - Modify import statements in test_tilelang_kernel_dequantize_gemm.py - Replace bitblas imports with tilelang.intrinsics imports for MMA-related utilities - Update main function to use tilelang.testing.main() * Add Block Sparse Attention Examples for TileLang and Triton - Implement block sparse attention kernels for both TileLang and Triton - Add utility functions for generating sparse attention masks using top-k and threshold methods - Support causal and variable-length attention scenarios - Include test cases for different sequence length configurations - Demonstrate block-level sparse attention with configurable parameters * Refactor Block Sparse Attention Examples with Code Style Improvements - Improve code formatting in block_sparse_attn_tilelang.py and block_sparse_attn_triton.py - Enhance readability by adjusting line breaks and indentation - Simplify kernel and function calls with better formatting - Add whitespace and line break improvements for better code clarity * Enhance Layout Plotting with Multi-Replication and Dynamic Visualization - Update plot_layout function to support multiple replications in thread and value mapping - Improve thread and value mapping to handle replicated layouts - Dynamically adjust figure size and legend positioning - Add print statements for saved plot file paths - Modify example fragment_mma_load_a.py to uncomment and enable warp and block layout plotting
-
Yu Cheng authored
-
