56fd68e7dd
CCCL (CUDA C++ Core Libraries) provides: - CUB: device/block/warp-level GPU primitives (reduce, scan, sort, topk) - Thrust: high-level parallel algorithms (transform_reduce, sort, scan) - libcudacxx: CUDA C++ standard library (atomics, barriers, memory) - cudax: experimental features (memory resources, allocators) - Tuning policies: per-SM hardware-specific algorithm parameters Competition optimization vectors mapped to CCCL: - Output TPS (83% weight): warp_reduce, block_reduce, device_topk - Input TPS (14% weight): device_scan, block_load, prefetch - Cache TPS (3% weight): prefix caching strategy patterns - Memory (0.9 util): pooled/cached/buddy allocators Source: https://github.com/NVIDIA/cccl (shallow clone, HEAD only) License: Apache-2.0
23 lines
1.1 KiB
CMake
23 lines
1.1 KiB
CMake
# There are extra ; in the extended lambda implementation
|
|
target_compile_options(
|
|
${test_target}
|
|
PRIVATE $<$<COMPILE_LANG_AND_ID:CUDA,NVIDIA>:--extended-lambda>
|
|
)
|
|
|
|
# This check is actually not correct, because we must check the host compiler, not the CXX compiler.
|
|
# We rely on these usually being the same ;)
|
|
if (
|
|
"Clang" STREQUAL "${CMAKE_CXX_COMPILER_ID}"
|
|
AND CMAKE_CXX_COMPILER_VERSION VERSION_GREATER_EQUAL 13
|
|
)
|
|
# When clang >= 13 is used as host compiler, we get the following warning:
|
|
# nvcc_internal_extended_lambda_implementation:312:22: error: definition of implicit copy constructor for '__nv_hdl_wrapper_t<false, true, false, __nv_dl_tag<void (*)(), &TestAddressStabilityLambda, 2>, int (const int &)>' is deprecated because it has a user-declared copy assignment operator [-Werror,-Wdeprecated-copy]
|
|
# 312 | __nv_hdl_wrapper_t & operator=(const __nv_hdl_wrapper_t &in) = delete;
|
|
# | ^
|
|
# Let's suppress it until NVBug 4980157 is resolved.
|
|
target_compile_options(
|
|
${test_target}
|
|
PRIVATE $<$<COMPILE_LANG_AND_ID:CUDA,NVIDIA>: -Wno-deprecated-copy>
|
|
)
|
|
endif()
|