mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-12 01:10:17 +00:00
920acd2c12
[CK] [CK_Tile] Add GroupConv to Kernel Dispatcher ## Motivation This PR adds CK Tile group convolution (forward, backward-data, backward-weight) support to the kernel dispatcher, matching and unifying with the existing dispatcher GEMM infrastructure in architecture and usability. The dispatcher provides a unified kernel dispatch system with both C++ and Python frontends, and until now only supported GEMM operations. This PR enables framework integrators to use the same declarative kernel workflow for convolutions as they do for GEMM: declare kernels, build a registry JIT, select kernels within the registry at runtime, and dispatch to GPU. Future PRs will include runtime kernel selection heuristics for autotuning of kernel parameters based on (problem, hardware arch). ## Technical Details Grouped convolution support has been added to the CK Tile Dispatcher with generated_conv_backend.hpp enabling dispatcher.run(in, wei, out, problem) for all 6 conv variants (fwd/bwdd/bwdw x 2D/3D), runtime heuristic kernel selection, and GroupedConvKernelKey with full ConvConfigBase fields. Python side adds parallel JIT via registry.build(max_workers) and heuristic registry.select(). Includes 7 C++ and 6 Python examples covering all directions with CPU reference validation, and shared infrastructure improvements (BaseRegistry CRTP, structured exceptions). As a sanity check, JIT compile times for a single kernel remains the same and for multiple kernels there is better parallelism: Kernels | 1 worker | 8 workers 1 | 7.7 s | 7.7 s 2 | 15.9 s | 8.2 s 4 | 33.4 s | 9.7 s 6 | 52.3 s | 10.2 s ## Test Plan 145 ephemeral unit tests have been added to test basic functionality. All 30 examples/integration tests run end-to-end on gfx950 (MI350): 7 C++ conv, 7 C++ GEMM, 6 Python conv, 10 Python GEMM. CPU reference validation for forward, backward-data, and backward-weight (2D) in both C++ and Python examples pass. ## Test Result 30 examples pass. Peak performance: 132 TFLOPS (Batch-32 forward 56x56), 53 TFLOPS (pointwise 1x1). CPU reference accuracy: max_abs_diff < 0.002 for all directions (fp16 vs fp32 reference). ## Submission Checklist - [x] Look over the contributing guidelines at https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
CK Tile Dispatcher Python Utilities
This directory contains Python utilities used by the dispatcher examples.
Contents
Shared Utilities (used by both GEMM and Grouped Conv)
dispatcher_common.py- Shared dispatcher infrastructure- Path helpers (
get_dispatcher_root,get_build_dir, etc.) ValidationResultBase- Structured validation feedbackvalidate_wave_config,validate_warp_tile_config,validate_trait_comboauto_correct_wave,auto_correct_trait- Auto-correction helpersColors- Cross-platform ANSI color supportprint_phase,print_success,print_error,print_info- Phased outputcleanup_generated_kernels- Cleanup helper
- Path helpers (
GEMM Utilities
ctypes_utils.py- Core ctypes utilities for GEMM Python examplesKernelConfig- Kernel configuration dataclasssetup_gemm_dispatcher()- Setup dispatcher with auto-correctioncleanup_gemm()- Cleanup dispatcher resourcesGemmRunner- GPU execution helper- Auto-correction and validation utilities
Grouped Convolution Utilities
grouped_conv_utils.py- Utilities for grouped convolutionGroupedConvValidationResult- Validation result (extendsValidationResultBase)validate_grouped_conv_config- Validate a grouped conv configauto_correct_grouped_conv_config- Auto-correct invalid configsget_grouped_conv_default_config- Get default config for a variantGroupedConvDataType- Data type enum (FP16, BF16, FP32, FP8, BF8, INT8)format_grouped_conv_summary- Human-readable config summary
Usage
GEMM Examples
The GEMM Python examples in dispatcher/examples/gemm/python/ import:
import sys
from pathlib import Path
sys.path.insert(0, str(Path(__file__).parent.parent.parent.parent / "python"))
from ctypes_utils import (
KernelConfig,
setup_gemm_dispatcher,
cleanup_gemm,
GemmRunner,
)
Grouped Conv Usage
import sys
from pathlib import Path
sys.path.insert(0, str(Path(__file__).parent.parent.parent.parent / "python"))
from grouped_conv_utils import (
validate_grouped_conv_config,
auto_correct_grouped_conv_config,
get_grouped_conv_default_config,
GroupedConvDataType,
)
# Get a default config
config = get_grouped_conv_default_config(variant="forward", arch="gfx942")
# Validate
result = validate_grouped_conv_config(config)
print(f"Valid: {result.is_valid}")
Requirements
- Python 3.8+
- NumPy
- HIP runtime (for GPU execution)