mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-06-29 11:16:59 +00:00
c31fc4df52
[CK Tile Engine] Daily tier sampling for tile engine GEMM (#7311) Summary - Replace uniform random instance sampling (random.shuffle) with scrambled Sobol + Latin Hypercube + maximin space-filling sampling, per the Tile Engine Benchmark Sampling RFC - Add op-weighted budget allocation via new TILE_ENGINE_SAMPLING_TIER=daily CMake knob that auto-distributes 8,000 instances across ops proportional to registered weights in op_weights.json - Emit chosen_instances.json manifests for reproducibility tracking - Consolidate 5 copies of sampling logic into single _apply_sampling() method on the base class Jenkinsfile changes Replace per-op -D *_MAX_INSTANCES=250 with single -D TILE_ENGINE_SAMPLING_TIER=daily in gfx942/gfx950/gfx1201 stages. Budget auto-distributes (8000 total per GPU target). --------- Co-authored-by: Claude Sonnet 4 <noreply@anthropic.com>
69 lines
2.9 KiB
CMake
69 lines
2.9 KiB
CMake
# Copyright (c) Advanced Micro Devices, Inc., or its affiliates.
|
|
# SPDX-License-Identifier: MIT
|
|
|
|
# Budget allocation when TILE_ENGINE_SAMPLING_TIER is set
|
|
if(NOT "${TILE_ENGINE_SAMPLING_TIER}" STREQUAL "")
|
|
# Map tier name to budget
|
|
if("${TILE_ENGINE_SAMPLING_TIER}" STREQUAL "daily")
|
|
set(_te_budget 8000)
|
|
elseif("${TILE_ENGINE_SAMPLING_TIER}" STREQUAL "weekly")
|
|
set(_te_budget 0)
|
|
else()
|
|
set(_te_budget ${TILE_ENGINE_SAMPLING_TIER})
|
|
endif()
|
|
|
|
if(_te_budget GREATER 0)
|
|
# Detect active ops from their DATATYPE variables
|
|
set(_active_ops "")
|
|
foreach(_op gemm_universal gemm_multi_d gemm_preshuffle grouped_gemm)
|
|
string(TOUPPER ${_op} _OP_UPPER)
|
|
if(NOT "${${_OP_UPPER}_DATATYPE}" STREQUAL "")
|
|
list(APPEND _active_ops ${_op})
|
|
endif()
|
|
endforeach()
|
|
|
|
if(_active_ops)
|
|
string(REPLACE ";" "," _active_ops_csv "${_active_ops}")
|
|
set(_alloc_dir "${CMAKE_CURRENT_BINARY_DIR}/sampling_alloc")
|
|
file(MAKE_DIRECTORY ${_alloc_dir})
|
|
|
|
execute_process(
|
|
COMMAND ${CMAKE_COMMAND} -E env
|
|
"PYTHONPATH=${CMAKE_CURRENT_LIST_DIR}/../../"
|
|
${Python3_EXECUTABLE} -m sampling.allocate_budget
|
|
--total-budget ${_te_budget}
|
|
--active-ops "${_active_ops_csv}"
|
|
--output-dir ${_alloc_dir}
|
|
WORKING_DIRECTORY ${CMAKE_CURRENT_LIST_DIR}/../../
|
|
RESULT_VARIABLE _alloc_ret
|
|
OUTPUT_VARIABLE _alloc_output
|
|
ERROR_VARIABLE _alloc_error
|
|
)
|
|
if(NOT _alloc_ret EQUAL 0)
|
|
message(FATAL_ERROR "Budget allocation failed: ${_alloc_error}")
|
|
endif()
|
|
message(STATUS "Sampling budget allocation:\n${_alloc_output}")
|
|
|
|
# Read per-op allocations (only if not already overridden)
|
|
foreach(_op gemm_universal gemm_multi_d gemm_preshuffle grouped_gemm)
|
|
string(TOUPPER ${_op} _OP_UPPER)
|
|
if("${${_OP_UPPER}_MAX_INSTANCES}" STREQUAL "")
|
|
if(EXISTS "${_alloc_dir}/${_op}_budget.txt")
|
|
file(READ "${_alloc_dir}/${_op}_budget.txt" _budget_val)
|
|
string(STRIP "${_budget_val}" _budget_val)
|
|
set(${_OP_UPPER}_MAX_INSTANCES ${_budget_val})
|
|
message(STATUS " ${_op}: ${_budget_val} instances (from budget allocation)")
|
|
endif()
|
|
else()
|
|
message(STATUS " ${_op}: ${${_OP_UPPER}_MAX_INSTANCES} instances (explicit override)")
|
|
endif()
|
|
endforeach()
|
|
endif()
|
|
endif()
|
|
endif()
|
|
|
|
add_subdirectory(gemm_universal EXCLUDE_FROM_ALL)
|
|
add_subdirectory(gemm_multi_d EXCLUDE_FROM_ALL)
|
|
add_subdirectory(gemm_preshuffle EXCLUDE_FROM_ALL)
|
|
add_subdirectory(grouped_gemm EXCLUDE_FROM_ALL)
|