v0.11.1

What's Changed

• CUDA 12.4 by @bitemyapp in #227
• Moving cublasHgemm link to dynamic loading by @coreylowman in #229

New Contributors

• @bitemyapp made their first contribution in #227

Full Changelog: v0.11.0...v0.11.1

v0.11.0

What's Changed

• Added result and safe functionality to support the cuFuncSetAttribute function. by @GaryMcD in #199
• Implemented LaunchAsync for raw ptr arrays by @jafioti in #204
• Safe ConvolutionND by @bloodre in #202
• Fixed Test Example 07 by @Fiend-Star-666 in #205
• Fix pipelines by @coreylowman in #207
• Adding cuda toolkit versions 12.0/12.1/12.2 of sys & feature flags by @coreylowman in #210
• support cuda toolkit version: 11.7 by @wenhaozhao in #214
• Adding cuda 12.3 bindings by @coreylowman in #215
• [Breaking] Remove Static linking. Make dynamic loading default. Add dynamic-linking feature flag by @coreylowman in #211
• Rework cuda versioning features - Users must always specify a version feature instead of having default behavior. by @coreylowman in #216
• Replace unwrap with ? in CudaDevice::new() by @coreylowman in #223
• Update README.md by @chaserileyroberts in #218
• Add CudaDevice::new_with_stream by @coreylowman in #224
• Wrapping comm_split in cfg based on cuda version by @coreylowman in #225

New Contributors

• @GaryMcD made their first contribution in #199
• @jafioti made their first contribution in #204
• @bloodre made their first contribution in #202
• @Fiend-Star-666 made their first contribution in #205
• @wenhaozhao made their first contribution in #214
• @chaserileyroberts made their first contribution in #218

Full Changelog: v0.10.0...v0.11.0

v0.10.0

What's Changed

• feat: add support for batch matrix-matrix product in cuBLASLt by @OlivierDehaene in #186
• [Breaking] Update dtoh_sync_copy by @zjsec in #183
• [Breaking] cublaslt::result::get_matmul_algo_heuristic is now unsafe by @coreylowman in #189
• fix: fix cublaslt objects memory leak by @OlivierDehaene in #192

New Contributors

• @zjsec made their first contribution in #183

Full Changelog: v0.9.15...v0.10.0

v0.9.15

What's Changed

• Fixing 07-build-workflow. by @Narsil in #175
• Making the tests pass on single GPU platforms. by @Narsil in #176
• Fix typo in docs about for_num_elems of LaunchConfig by @mert-kurttutan in #177
• Minimum changes for occupancy API by @Jark5455 in #180
• feat: Add support for cublas Lt by @OlivierDehaene in #182
• Add set_offset so that curandSetGeneratorOffset() can be called. by @mneilly in #185

New Contributors

• @mert-kurttutan made their first contribution in #177
• @Jark5455 made their first contribution in #180
• @OlivierDehaene made their first contribution in #182
• @mneilly made their first contribution in #185

Full Changelog: v0.9.14...v0.9.15

v0.9.14 - fixes bug when async alloc/free isn't available

What's Changed

• Enabling runtime choice for mallocAsync or sync version. by @Narsil in #174

Full Changelog: v0.9.13...v0.9.14

v0.9.13 - nccl support, gemm<bf16>, other small updates

What's Changed

• Implement Gemm for CudaBlas. by @LaurentMazare in #167
• NCCL support by @Narsil in #164
• Adds handle method to CudaBlas by @coreylowman in #170
• Adds Ptx::to_src method by @coreylowman in #171

New Contributors

• @LaurentMazare made their first contribution in #167

Full Changelog: v0.9.12...v0.9.13

v0.9.12 - fix multi threading/multi device context management

What's Changed

• Adding RAII version for the profiler. by @Narsil in #162
• Calling bind_to_thread on all api calls by @coreylowman in #166

Full Changelog: v0.9.11...v0.9.12

v0.9.11 - bump half-rs version

What's Changed

• half v2.3.1 by @coreylowman in #159

Full Changelog: v0.9.10...v0.9.11

v0.9.10

What's Changed

• Use compute type f32 for f16 matmul by @coreylowman in #151
• Add functions to interact with external memory by @ViliamVadocz in #152
• Adds an attribute method for device and a .rustfmt.toml by @emchristiansen in #154
• Improve external memory interface by @ViliamVadocz in #155
• WIP: Add occupancy to driver::result by @tthebst in #150

New Contributors

• @emchristiansen made their first contribution in #154

Full Changelog: v0.9.9...v0.9.10

v0.9.9

What's Changed

• Add lib as a library candidate in build.rs by @coreylowman in #142
• Adds Conv2dDescriptor::set_group_count by @coreylowman in #145
• Adding methods to get reference to internal sys handles of CudaDevice by @coreylowman in #147

Full Changelog: v0.9.8...v0.9.9