`DLPack` to `mdspan`#7047

Open

fbusato wants to merge 60 commits intoNVIDIA:mainfrom

fbusato:dlpack-to-mdspan

Commits on Dec 18, 2025

first version
fbusato
committed

Commits on Dec 20, 2025

Commits on Dec 23, 2025

complete the implementation
fbusato
committed
add unit test
fbusato
committed
cuda.coop: Use cuda.core.experimental.Linker instead of internal numba-cuda Linker to link LTO (#7011)

shwina
authored and
fbusato
committed
Make c2h vector comparisons `constexpr` (#7009)

davebayer
authored and
fbusato
committed
improves comments on decoupled lookback example (#7015)

elstehle
authored and
fbusato
committed
Extract reduce_op_sync into a free function (#7004)

bernhardmgruber
authored and
fbusato
committed
Remove experimental namespace from cuda.core import (#7022)

NaderAlAwar
authored and
fbusato
committed
reexpress completion signature transform alias to make clangd happy (#7026)

ericniebler
authored and
fbusato
committed
Qualify call to `__launch_impl` in launch.h to avoid ambiguity errors (#7024)

authored and
fbusato
committed
Rework hierarchy levels (#6957)

authored and
fbusato
committed
Use vectorized tuning for triad benchmark for dtypes of size 2 (#7019)

NaderAlAwar
authored and
fbusato
committed
[libcu++] Fix synchronous resource adapter property passing (#6976)

pciolkosz
authored and
fbusato
committed
[libcu++] Remove _view from the shared memory getter name (#6997)

pciolkosz
authored and
fbusato
committed
[thrust] Ignore CUDA free errors in thrust memory resource (#7002)

pciolkosz
authored and
fbusato
committed
[libcu++] Correctly handle extended lambda in cuda::launch (#6987)

authored and
fbusato
committed
the `<stdexcept>` header must be included when using `_CCCL_THROW`, regardless of exception support (#7028)

fbusato
and
davebayer
committed
Error out when nvrtcc cannot parse cuda_thread_count (#7035)

bernhardmgruber
authored and
fbusato
committed
Allow all public headers to be included with host compilers only (#7012)

davebayer
authored and
fbusato
committed
[cuda.compute]: Fixes and updates to benchmarks (#6999)

shwina
authored and
fbusato
committed
Support operations with side-effects (state) in `cuda.compute` (#7008)

shwina
authored and
fbusato
committed
Fix `cuda::memcpy async` edge cases and add more tests (#6608)

bernhardmgruber
authored and
fbusato
committed
Explicitly set `CCCL_TOPLEVEL_PROJECT` to `OFF` when needed (#7016)

KyleFromNVIDIA
authored and
fbusato
committed
[libcu++] Add explicit alignment specification in buffer (#7005)

pciolkosz
authored and
fbusato
committed
Use the sccache-dist build cluster for RAPIDS CI jobs (#7014)

trxcllnt
authored and
fbusato
committed
first version
fbusato
committed
complete the implementation
fbusato
committed
add unit test
fbusato
committed
Merge branch 'dlpack-to-mdspan' of github.com:fbusato/cccl into dlpack-to-mdspan
fbusato
committed
fix unit test
fbusato
committed
formatting
fbusato
committed
minor fixes
fbusato
committed
fix compiler warnings
fbusato
committed

Commits on Jan 14, 2026