[Example] Add GQA Example #118
Merged
+348
−168
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This commit introduces two new example scripts demonstrating advanced GEMM (matrix multiplication) techniques: - `example_tilelang_gemm_splitk.py`: Implements a Split-K GEMM kernel using TileLang - `example_tilelang_gemm_streamk.py`: Implements a Stream-K GEMM kernel using TileLang Both examples showcase different parallel computation strategies for matrix multiplication, with comprehensive testing using PyTorch reference implementations.
Clean up and improve code formatting for the SplitK and StreamK GEMM example scripts: - Remove unused import (Profiler) in splitk example - Simplify line breaks and improve code readability - Standardize indentation and remove unnecessary whitespace - Optimize atomic add and copy operations for better clarity
This commit introduces comprehensive block sparse attention benchmarks for different libraries: - TileLang block sparse FMHA implementation - Triton block sparse FMHA implementation - PyTorch reference block sparse FMHA implementation - FlashAttention dense FMHA reference implementation The benchmarks include: - Configurable benchmark parameters (batch size, heads, sequence length, etc.) - Sparse mask generation using top-k and threshold methods - Performance measurement for different sparse attention configurations - Utility functions for mask generation and benchmarking
- Add Ruff linter ignore comments to benchmark files - Improve code formatting and line breaks - Remove unused imports - Standardize print statement formatting - Enhance code readability across multiple library benchmarks
- Implement AtomicAdd functions for BFLOAT16 and BFLOAT16x2 in CUDA common header - Rename existing atomic add functions to use PascalCase (atomicAdd -> AtomicAdd) - Add a new __pack_nv_bfloat162 function for packing BFLOAT16 values - Update kernel and language customization to use new function names - Add return type annotations in profiler module
…Attention in TileLang This commit introduces a new example script `example_gqa_fwd_bshd.py` that demonstrates: - Group Query Attention (GQA) implementation - Flash Attention forward pass - Performance benchmarking - Configurable parameters for batch, heads, sequence length, and dimension - Autotuning support - Reference implementation comparison
This commit introduces a new module `phase.py` to modularize the IR lowering process by splitting the complex lowering pipeline into two distinct phases: - `LowerAndLegalize`: Handles initial IR legalization and transformation - `OptimizeForTarget`: Applies target-specific optimizations The changes simplify the lowering logic in multiple files by extracting the transformation steps into reusable functions, improving code readability and maintainability.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This pull request introduces a new example script for flash attention with grouped query and key-value attention (GQA) in the
examples/flash_attention/example_gqa_fwd_bshd.pyfile. The script includes functions for configuring and running the flash attention kernel, as well as comparing its performance to a reference implementation.Key changes include:
New example script for flash attention with GQA:
examples/flash_attention/example_gqa_fwd_bshd.pywhich includes the implementation of the flash attention kernel with grouped query and key-value attention.Kernel function and configuration:
flashattnfunction that sets up the kernel function, including macros for matrix multiplication (MMA), softmax, and rescaling operations.get_configsfunction to generate different configurations for autotuning the kernel.Reference implementation:
ref_programfunction to provide a reference implementation of the attention mechanism using PyTorch operations.Command-line interface:
mainfunction with argument parsing to allow users to run the script with different parameters such as batch size, number of heads, sequence length, dimension, causality, tuning, and groups. (examples/flash_attention/example_gqa_fwd_bshd.pyR1-R241