v0.12.2 - initial runtime api & small non-breaking additions

What's Changed

• Add safe methods set_pointer_mode and get_pointer_mode by @MathisWellmann in #291
• Adds Comm::reduce_in_place and Comm::broadcast_in_place by @coreylowman in #293
• CUDA runtime api by @psmyth94 in #262
• Add more cudnn operations by @kckeiks in #284
• Fix Conda by @ur4t in #302

New Contributors

• @MathisWellmann made their first contribution in #291
• @psmyth94 made their first contribution in #262
• @kckeiks made their first contribution in #284
• @ur4t made their first contribution in #302

Full Changelog: v0.12.1...v0.12.2

v0.12.1 - Cuda 12.6

What's Changed

• Support CUDA 12.6. by @paulstansifer in #286

New Contributors

• @paulstansifer made their first contribution in #286

Full Changelog: v0.12.0...v0.12.1

v0.12.0 - Various small breaking changes to some APIs/structs.

What's Changed

• [Breaking] Adds cooperative launch to LaunchAsync by @JRazek in #221
• [Breaking] Add support for passing options to nvrtc by @brandonros in #251
• [Breaking] Making cudnn and nccl non-default features in Cargo.toml by @coreylowman in #275
• [breaking] feat: allow NCCL comm type to operate on views as well by @dkales in #281
• Update dependencies: half 2.4.1, libloading 0.8.5 by @coreylowman in #289

New Contributors

• @JRazek made their first contribution in #221

Full Changelog: v0.11.9...v0.12.0

v0.11.9

What's Changed

• Adding options for dynamic loading that include versions after .so by @coreylowman in #276
• feat: add transmute for views by @dkales in #282
• Fixing version check in cudnn for 11.4 by @coreylowman in #288

Full Changelog: v0.11.8...v0.11.9

v0.11.7 - Cuda 11.4, CUDNN path fixes, new driver functions

What's Changed

• Add documentation on how to synchronize non-default streams by @sebhtml in #261
• Cuda1104 by @elmattic in #266
• Add more search paths/lib names for cudnn dynamic linking & loading by @coreylowman in #264
• Handling self.error_string() failing in Debug for DriverError by @coreylowman in #267
• Adding various functions to driver result api for managed memory by @coreylowman in #268

New Contributors

• @sebhtml made their first contribution in #261
• @elmattic made their first contribution in #266

Full Changelog: v0.11.6...v0.11.7

v0.11.6 - More lib_names for dynamic loading, better error messages

What's Changed

• add another windows lib_name workaround by @brandonros in #250
• Add CudaDevice::name() and result::device::get_name() by @coreylowman in #252
• Improve panic message if dynamic loading failed. by @coreylowman in #257

New Contributors

• @brandonros made their first contribution in #250

Full Changelog: v0.11.5...v0.11.6

v0.11.5 - Support cuda 11.5/11.6

What's Changed

• Add CudaSlice::split_at_mut and CudaViewMut::split_at_mut by @dkales in #235
• Add support for 11.5 and 11.6 toolkit versions. by @coreylowman in #249

New Contributors

• @dkales made their first contribution in #235

Full Changelog: v0.11.4...v0.11.5

v0.11.4 - Using nvcc to find cuda toolkit version & curand dynamic loading fix

What's Changed

• Using nvcc to get cuda toolkit version instead of cuda.h by @coreylowman in #241
• Adding fix for finding curand64_10.dll by @coreylowman in #243

Full Changelog: v0.11.3...v0.11.4

v0.11.3 - Dynamic loading improvements on windows

What's Changed

• Adding fallback lib name options for dynamic loading by @coreylowman in #240

Full Changelog: v0.11.2...v0.11.3

v0.11.2 - Cuda toolkit 12.5

What's Changed

• support CUDA Toolking 12.5 by @vartec in #238

New Contributors

• @vartec made their first contribution in #238

Full Changelog: v0.11.1...v0.11.2