59 Pull requests merged by 13 people

• Fix internal shfl check

  #4282 merged Mar 27, 2025

• change version check in type_list.h so that *NO* clang-19.X compilers try to use pack indexing

  #4278 merged Mar 27, 2025

• fix clang portability issue in __rcvr_with_env_t and remove dead code

  #4277 merged Mar 27, 2025

• Drop _CCCL_NORETURN

  #4268 merged Mar 27, 2025

• Fix ratio plot

  #4099 merged Mar 27, 2025

• [Backport branch/3.0.x] Fix not_fn

  #4196 merged Mar 27, 2025

• Remove few remaining qualifiers _CCCL_NODISCARD

  #4274 merged Mar 26, 2025

• [BACKPORT] Allow rapids to avoid unrolling some loops in sort (#4253)

  #4254 merged Mar 26, 2025

• Cleanup libcu++ force_include.h test file

  #4262 merged Mar 26, 2025

• Drop _CCCL_FALLTHROUGH

  #4269 merged Mar 26, 2025

• Change to allow cccl/c/parallel/unique_by_key.h to compile by C compiler

  #4259 merged Mar 26, 2025

• Drop _CCCL_NO_INLINE_VARIABLES

  #4267 merged Mar 26, 2025

• Drop unused _CCCL_ALIAS_ATTRIBUTE macro

  #4266 merged Mar 26, 2025

• Drop _CCCL_NODISCARD

  #4265 merged Mar 26, 2025

• Restore CUB changelog

  #4263 merged Mar 26, 2025

• Implement __fp_neg

  #4257 merged Mar 26, 2025

• Allow rapids to avoid unrolling some loops in sort

  #4253 merged Mar 26, 2025

• [STF] fix type issues in the multi-GPU CG test

  #4260 merged Mar 26, 2025

• [STF] Enable NVHPC in CUDASTF CI

  #3857 merged Mar 26, 2025

• Implement fp constants

  #4256 merged Mar 26, 2025

• Drop old Readmes and other unused files

  #4199 merged Mar 26, 2025

• Add fixed size segmented reduce

  #3969 merged Mar 25, 2025

• [cudax] incorporate P3557 (constexpr completion signatures) into µstdex

  #3841 merged Mar 25, 2025

• Implement views::all

  #4244 merged Mar 25, 2025

• Refactor fp masks

  #4246 merged Mar 25, 2025

• Fix construction of host init value in test_reduce made incorrect after refactoring

  #4251 merged Mar 25, 2025

• Add new WarpReduce overloadings

  #3884 merged Mar 25, 2025

• Fix logic in test_segmented_reduce

  #4198 merged Mar 24, 2025

• Make cuda::stream_ref constructible on device

  #4243 merged Mar 24, 2025

• Exclude sm101 from RDC testing.

  #4247 merged Mar 24, 2025

• Drop duplicated system header blocks

  #4245 merged Mar 24, 2025

• Add transform c parallel implementation

  #4048 merged Mar 24, 2025

• Implement __cccl_fp class

  #4238 merged Mar 24, 2025

• Move implementation of cuda::std::array to libcu++

  #4239 merged Mar 24, 2025

• Use [[maybe_unused]] in whole cccl

  #4207 merged Mar 24, 2025

• Extend tuning verification docs

  #4236 merged Mar 24, 2025

• Implement __fp_is_subset trait

  #4230 merged Mar 24, 2025

• [Backport branch/3.0.x] Document deprecations from #4165

  #4242 merged Mar 24, 2025

• Document deprecations from #4165

  #4237 merged Mar 24, 2025

• Remove double underscore limit macros

  #4194 merged Mar 21, 2025

• Simplify Thrust assign_value

  #4227 merged Mar 21, 2025

• Fix typos in cuda mdspan documentation

  #4231 merged Mar 21, 2025

• Backport PR #4221 to branch/3.0.x — Remove python/cuda_cooperative/setup.py

  #4234 merged Mar 21, 2025

• Backport PR #4221 to branch/2.8.x — Remove python/cuda_cooperative/setup.py

  #4235 merged Mar 21, 2025

• Drop _CCCL_NO_VARIABLE_TEMPLATES

  #4229 merged Mar 21, 2025

• Allow cuda::par*.on() to take cuda::stream_ref

  #4225 merged Mar 21, 2025

• Remove python/cuda_cooperative/setup.py

  #4221 merged Mar 21, 2025

• Remove _CCCL_NO_FOLD_EXPRESSIONS

  #4189 merged Mar 21, 2025

• Document python packages for sol plot script

  #4228 merged Mar 21, 2025

• Make compiler version comparisons safer

  #4185 merged Mar 21, 2025

• Drop tuning params for benchmarks with custom ops

  #4176 merged Mar 21, 2025

• Extend tuning documentation

  #4184 merged Mar 21, 2025

• Fix C++ version used in CONTRIBUTING.md

  #4224 merged Mar 21, 2025

• Drop _LIBCUDACXX_BEGIN_NAMESPACE_RANGES_ABI

  #4210 merged Mar 21, 2025

• [BACKPORT 2.8] Fix the cuda python setup

  #4218 merged Mar 21, 2025

• Improve host/device mdspan documentation

  #4220 merged Mar 21, 2025

• Remove _CCCL_INLINE_VAR

  #4192 merged Mar 20, 2025

• Implement extended fp traits

  #4211 merged Mar 20, 2025

• Avoid errors in get_device_address tests

  #4209 merged Mar 20, 2025

11 Pull requests opened by 9 people

• Fix dereferencing nullptr in `thrust::device_reference`

  #4226 opened Mar 21, 2025

• Fix struct types handling to eliminate ABI mismatch between C++ and Numba

  #4249 opened Mar 24, 2025

• Try and enable msvc for STF

  #4252 opened Mar 25, 2025

• Implement `ranges::single_view`

  #4255 opened Mar 25, 2025

• Implement fp overflow handlers

  #4261 opened Mar 26, 2025

• Drop `_LIBCUDACXX_DEPRECATED_IN_[11|14|17]`

  #4271 opened Mar 26, 2025

• [STF] dispatch content of stf.cuh into internal headers

  #4275 opened Mar 26, 2025

• tweak the cccl compiler version check macros to better agree with intuition

  #4279 opened Mar 26, 2025

• Update CUB dispatch layer documentation with new example

  #4281 opened Mar 26, 2025

• Fix vectorized loading and storing for warpLoad, warpStore and blockS…

  #4283 opened Mar 27, 2025

• [Backport branch/3.0.x] change version check in `type_list.h` so that *NO* clang-19.X compilers try to use pack indexing

  #4284 opened Mar 27, 2025

8 Issues closed by 6 people

• [FEA]: thrust::reduce needs no_sync version

  #4264 closed Mar 26, 2025

• Build broken after removal of _CCCL_NODISCARD in #4265

  #4273 closed Mar 26, 2025

• [BUG] cuda.parallel: unique_by_key.h header should be compilable by C compiler

  #4258 closed Mar 26, 2025

• Performance of small sums could be improved

  #921 closed Mar 25, 2025

• cuda.parallel: Incorrect use of host init array in test_reduce.py

  #4250 closed Mar 25, 2025

• Correct test_segmented_reduce.py::test_segmented_reduce

  #4197 closed Mar 24, 2025

• Add c.parallel implementation of transform

  #3877 closed Mar 24, 2025

• Thrust `execution_policy::on` should take a stream_ref in addition to a stream

  #4150 closed Mar 21, 2025

5 Issues opened by 4 people

• [DOC]: Add more details to CUB dispatch layer documentation

  #4280 opened Mar 26, 2025

• [BUG] cuda.parallel: ABI mismatch between C++ and Python for struct types

  #4248 opened Mar 24, 2025

• [BUG]: cccl.c merge_sort with custom types falis for larger inputs

  #4240 opened Mar 23, 2025

• Add support for large num_segments to `DeviceSegmentedReduce` with fixed segment size

  #4233 opened Mar 21, 2025

• [BUG]: CMake fails when using nvc++ as a c++ compiler

  #4232 opened Mar 21, 2025

25 Unresolved conversations

Sometimes conversations happen on old items that aren’t yet closed. Here is a list of all the Issues and Pull Requests with unresolved conversations.

• Start working on developer overview

  #2599 commented on Mar 24, 2025 • 6 new comments

• [Thrust] Perform asynchronous allocations by default for the `par_nosync` policy

  #4204 commented on Mar 22, 2025 • 5 new comments

• Add dynamic CUB dispatch for radix_sort

  #4135 commented on Mar 27, 2025 • 5 new comments

• Simplify thrust::cuda_cub::swap_ranges

  #4182 commented on Mar 21, 2025 • 4 new comments

• Allow default-initializing Thrust vectors

  #4183 commented on Mar 21, 2025 • 3 new comments

• [STF] stackable stf resources

  #2674 commented on Mar 26, 2025 • 2 new comments

• Improve constexpr `std::complex` interoperability

  #4153 commented on Mar 21, 2025 • 1 new comment

• [Version] Update main to v3.1.0

  #4175 commented on Mar 26, 2025 • 0 new comments

• STF + MathLibs in CI

  #4121 commented on Mar 26, 2025 • 0 new comments

• Move `TEST_HAS_NO_EXCEPTIONS` to function like macro

  #4112 commented on Mar 27, 2025 • 0 new comments

• Fix inconsistent usage of vsmem helper in c.parallel merge_sort and unique_by_key algorithms

  #4090 commented on Mar 21, 2025 • 0 new comments

• Add cuda.parallel.experimental.iterators._strided with NdArrayIterator

  #4072 commented on Mar 26, 2025 • 0 new comments

• Add multi-dimensional support to block_reduce routines.

  #4064 commented on Mar 26, 2025 • 0 new comments

• Fix autogenerating release notes.

  #4052 commented on Mar 21, 2025 • 0 new comments

• Add an option to immediately create a point release PR after finalizing PR.

  #4051 commented on Mar 21, 2025 • 0 new comments

• [Backport branch/2.8.x] Update Blackwell PTX instruction availability tables

  #3900 commented on Mar 26, 2025 • 0 new comments

• Refactoring `ThreadReduce`

  #3441 commented on Mar 25, 2025 • 0 new comments

• Make CUDA-compiler specific CUB headers do nothing in other compilers

  #3378 commented on Mar 26, 2025 • 0 new comments

• Configure `workflow-run-job-linux` to use sccache-dist build cluster

  #2672 commented on Mar 27, 2025 • 0 new comments

• [Do Not Merge] Implement `<ranges>`

  #198 commented on Mar 27, 2025 • 0 new comments

• CCCL C: Avoid passing structs by value in C APIs

  #3952 commented on Mar 26, 2025 • 0 new comments

• [EPIC] RAPIDS Should not need to patch CCCL

  #1939 commented on Mar 25, 2025 • 0 new comments

• Track future deprecations and breaking changes

  #3701 commented on Mar 22, 2025 • 0 new comments

• [EPIC] Breaking changes for CCCL 3.0

  #101 commented on Mar 22, 2025 • 0 new comments

• [EPIC]: CUB large input support

  #50 commented on Mar 21, 2025 • 0 new comments