Bug 5281036: Limit the register number of debug version for cdpAdvancedQuicksort

This commit is contained in:
shawnz 2025-05-20 14:47:44 +08:00
parent c1b03b9f81
commit 1141dd7af4

View File

@ -23,6 +23,7 @@ set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -Wno-deprecated-gpu-targets")
if(ENABLE_CUDA_DEBUG) if(ENABLE_CUDA_DEBUG)
set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -G") # enable cuda-gdb (may significantly affect performance on some targets) set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -G") # enable cuda-gdb (may significantly affect performance on some targets)
set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -maxrregcount=64") # Limit register usage to 64 for the 'big_bitonicsort kernel
else() else()
set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -lineinfo") # add line information to all builds for debug tools (exclusive to -G option) set(CMAKE_CUDA_FLAGS "${CMAKE_CUDA_FLAGS} -lineinfo") # add line information to all builds for debug tools (exclusive to -G option)
endif() endif()