mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-12 01:10:17 +00:00
c1127a36f5
[CK][CK TILE]Autotuning heuristics infra for universal GEMM kernel selection (#5676) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit ## Motivation This PR adds ML-based kernel selection heuristics to the CK Tile dispatcher, enabling fast and accurate automatic kernel selection for Universal Gemm kernels. Instead of requiring exhaustive search through 4600+ kernel configurations (taking ~46 seconds per problem shape), the ML heuristic predicts optimal kernels in microseconds while achieving >98% of oracle-best performance. ## Technical Details **ML infrastructure** https://github.com/ROCm/rocm-libraries/tree/users/vanantha/ck/dispatcher-heuristics/projects/composablekernel/dispatcher/heuristics * Feature Engine ([feature_engine.py](https://github.com/ROCm/rocm-libraries/blob/users/vanantha/ck/dispatcher-heuristics/projects/composablekernel/dispatcher/heuristics/feature_engine.py)): 55-feature extraction including problem dimensions, kernel configuration, tile efficiency, and hardware profile * Training Pipeline ([train.py](https://github.com/ROCm/rocm-libraries/blob/users/vanantha/ck/dispatcher-heuristics/projects/composablekernel/dispatcher/heuristics/train.py)): LightGBM regression with log-transform, GroupKFold cross-validation, warm-start support * Predictor ([predict.py](https://github.com/ROCm/rocm-libraries/blob/users/vanantha/ck/dispatcher-heuristics/projects/composablekernel/dispatcher/heuristics/predict.py)): Kernel ranking and TFLOPS prediction for problem shapes * Evaluation ([evaluate.py](https://github.com/ROCm/rocm-libraries/blob/users/vanantha/ck/dispatcher-heuristics/projects/composablekernel/dispatcher/heuristics/evaluate.py)): Comprehensive metrics including efficiency, NDCG@k, shape family analysis **Data Generation Tools:** * [generate_benchmark_data.py](https://github.com/ROCm/rocm-libraries/blob/users/vanantha/ck/dispatcher-heuristics/projects/composablekernel/dispatcher/heuristics/generate_benchmark_data.py): Build and benchmark kernels across diverse problem shapes * [convert_json_to_parquet.py](https://github.com/ROCm/rocm-libraries/blob/users/vanantha/ck/dispatcher-heuristics/projects/composablekernel/dispatcher/heuristics/convert_json_to_parquet.py): Convert benchmark JSON to training-ready parquet format * [data_pipeline.py](https://github.com/ROCm/rocm-libraries/blob/users/vanantha/ck/dispatcher-heuristics/projects/composablekernel/dispatcher/heuristics/data_pipeline.py): Parse streaming benchmark logs into canonical datasets **Examples** * [09_ml_heuristic.cpp](https://github.com/ROCm/rocm-libraries/blob/users/vanantha/ck/dispatcher-heuristics/projects/composablekernel/dispatcher/examples/gemm/cpp/09_ml_heuristic.cpp): C++ example demonstrating ML-based kernel selection * [09_ml_heuristic.py](https://github.com/ROCm/rocm-libraries/blob/users/vanantha/ck/dispatcher-heuristics/projects/composablekernel/dispatcher/examples/gemm/python/09_ml_heuristic.py): Python example with validation **Pre-trained Models (projects/composablekernel/dispatcher/heuristics/models/):** * gemm_universal_fp8_gfx950/: fp8 RCR model (42K trees, 97.51% mean efficiency) * gemm_universal_fp16_gfx950/: fp16 RCR model (20K trees, 99.36% mean efficiency) ## Test Plan * Evaluated on 25 diverse shapes for fp16, 168 shapes for fp8 * All shape families tested: tiny M (M<8), small M, medium M, large M (M≥1024) * All pipeline types: compv3, compv4, mem ## Test Result **fp16 Model (gfx950, RCR layout)** * Mean Efficiency: 99.36% * P10 Efficiency: 98.05% (90th percentile of shapes achieve ≥98% of oracle best) * Min Efficiency: 95.45% **fp8 Model (gfx950, RCR layout)** * Mean Efficiency: 98.28% (original), 97.51% (wide coverage) * P10 Efficiency: 94.64% (original), 93.89% (wide coverage) * Min Efficiency: 84.5% ## Submission Checklist - [x ] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
28 lines
711 B
Plaintext
28 lines
711 B
Plaintext
# Core dependencies
|
|
numpy>=1.19.0
|
|
|
|
# ML Heuristic dependencies (OPTIONAL - large dependencies)
|
|
# For ML-based kernel selection, install separately:
|
|
# pip install -r ../requirements-ml.txt
|
|
# This avoids mandatory large dependencies (pyarrow, lightgbm) for users who don't need ML features
|
|
|
|
# Optional dependencies (install with pip install -e ".[torch]")
|
|
# torch>=2.0.0
|
|
|
|
# Development dependencies (install with pip install -e ".[dev]")
|
|
pytest>=6.0.0
|
|
# pytest-cov>=2.0.0
|
|
# black>=21.0
|
|
# flake8>=3.9.0
|
|
# mypy>=0.910
|
|
# isort>=5.0.0
|
|
|
|
# Visualization dependencies (install with pip install -e ".[viz]")
|
|
# matplotlib>=3.3.0
|
|
# seaborn>=0.11.0
|
|
|
|
# Documentation dependencies
|
|
# sphinx>=4.0.0
|
|
# sphinx-rtd-theme>=1.0.0
|
|
|