Age | Commit message (Collapse) | Author | |
---|---|---|---|
2019-04-05 | NCCL 2.4.6-1 | David Addison | |
Added detection of IBM/Power NVLink bridge device. Add NUMA support to PCI distance calculations. Added NCCL_IGNORE_CPU_AFFINITY env var. Fix memory leaks; GithubIssue#180 Compiler warning fix; GithubIssue#178 Replace non-standard variable length arrays. GithubIssue#171 Fix Tree+Shared Memory crash. GithubPR#185 Fix LL cleanup hang during long running DL jobs. Fix NCCL_RINGS environment variable handling. Added extra checks to catch repeat calls to ncclCommDestroy() GithubIssue#191 Improve bootstrap socket connection reliability at scale. Fix hostname hashing issue. GithubIssue#187 Code cleanup to rename all non device files from *.cu to *.cc | |||
2018-11-05 | Add install target | Sylvain Jeaugey | |
Fix issue #145 | |||
2018-09-26 | 2.3.5-5v2.3.5-5 | Sylvain Jeaugey | |
Add support for inter-node communication using sockets and InfiniBand/RoCE. Improve latency. Add support for aggregation. Improve LL/regular tuning. Remove tests as those are now at github.com/nvidia/nccl-tests . | |||
2017-06-14 | Add support for CUDA9 half semantics | Sylvain Jeaugey | |
2017-03-24 | Added Pascal nvcc flags, bumped versionv1.3.4-1 | Boris Fomitchev | |
2017-03-02 | Bumping version to 1.3.3 | Sylvain Jeaugey | |
2016-12-02 | 1.3.2 release | Sylvain Jeaugey | |
Broadcast tuning Better checking of inputs Copy/reduce code simplification | |||
2016-11-24 | Add a static library target "staticlib" to the Makefile. | Peter Jin | |
Rename the static library "libnccl_static.a" to disambiguate from the dynamic libraries. | |||
2016-11-18 | Add Fortran bindings | Kyle Fernandes, ne Jacobs | |
2016-10-13 | Bump to 1.3.1 | Sylvain Jeaugey | |
2016-09-22 | Add scan testsv1.3.0-1 | Sylvain Jeaugey | |
2016-09-22 | Heavy code refactoring to remove a lot of code in collectives (~1000 lines). | Sylvain Jeaugey | |
Have all collectives use the same args, the same ring, and the same primitives for synchronization between threads with the same pattern. | |||
2016-09-15 | Merge pull request #41 from jia-kai/master | Sylvain Jeaugey | |
Some minor fixes for compile/usage | |||
2016-08-27 | Updated LICENCE.txt | Sylvain Jeaugey | |
2016-08-19 | link library with -lrt; otherwise there is undefined reference to shm_open | jiakai | |
2016-07-07 | Improved Deb generation | Sylvain Jeaugey | |
2016-06-17 | Fix version number | Sylvain Jeaugey | |
2016-06-17 | Add a debug level to NCCL and CUDA versions at init | Sylvain Jeaugey | |
2016-06-16 | Increased version to 1.2.3 | Sylvain Jeaugey | |
2016-06-16 | Packaging : Generate shlibs.local | Sylvain Jeaugey | |
2016-06-16 | Move deb to build directory | Sylvain Jeaugey | |
2016-06-16 | Fix make install to use BUILDDIR | Sylvain Jeaugey | |
2016-06-16 | Rework debian packaging | Sylvain Jeaugey | |
2016-06-06 | Gencodes changed to NV recommendedv1.2.2-1+cuda7.5v1.2.1-2+cuda7.5 | Boris Fomitchev | |
2016-06-04 | Bump to 1.2.2 | Sylvain Jeaugey | |
2016-06-03 | Better name for GENCODE | Sylvain Jeaugey | |
2016-06-03 | Makefile improvements | Sylvain Jeaugey | |
- Use standard CXX env var - Permit redefinition of more env - Separate lib from tests | |||
2016-04-22 | Merge pull request #21 from borisfom/master | Boris Fomitchev | |
Fixed install location, new .deb version | |||
2016-04-22 | Fixed install location, new .deb version | Boris Fomitchev | |
2016-04-21 | Merge pull request #17 from Hopobcn/master | Nathan Luehr | |
Enable compilation with specific g++ | |||
2016-04-20 | Moved no-as-needed flag to link rule. | Nathan Luehr | |
Avoids link errors for tests linked with nvcc. | |||
2016-04-13 | Add --no-as-needed to make sure that cudart library gets liked | Adam Paszke | |
2016-04-12 | Enable compilation with old g++ when the default g++ is not supported (+5.0) | Pau Farré | |
2016-03-18 | Removing Tegra | Boris Fomitchev | |
2016-03-17 | Use arch=5.3 as well | Boris Fomitchev | |
2016-03-17 | Added compute 5.3 | Boris Fomitchev | |
2016-03-17 | Preparing for pbuild | Boris Fomitchev | |
2016-02-29 | Updated package version, added manpage | Boris Fomitchev | |
2016-01-29 | Added NCCL error checking to tests. | Nathan Luehr | |
Also cleaned up makefile so that tests and lib are not built unnecessarily. Change-Id: Ia0c596cc2213628de2f066be97615c09bb1bb262 Reviewed-on: http://git-master/r/999627 Reviewed-by: Przemek Tredak <ptredak@nvidia.com> Tested-by: Przemek Tredak <ptredak@nvidia.com> | |||
2016-01-28 | Moved tests to separate dir and improved MPI test | Sylvain Jeaugey | |
test sources moved to test/ directory. MPI test displays PASS/FAIL and returns code accordingly. Change-Id: I058ebd1bd5202d8f38cc9787898b2480100c102b Reviewed-on: http://git-master/r/936086 Reviewed-by: Przemek Tredak <ptredak@nvidia.com> Tested-by: Przemek Tredak <ptredak@nvidia.com> | |||
2016-01-08 | Merge pull request #6 from lukeyeager/deb | Nathan Luehr | |
Deb packaging | |||
2015-12-19 | Update deb packaging scripts | Luke Yeager | |
2015-12-19 | Merge pull request #5 from lukeyeager/tests-nvml | Nathan Luehr | |
Don't link tests with NVML | |||
2015-12-19 | Don't link tests with NVML | Boris Fomitchev | |
2015-12-19 | Merge pull request #4 from lukeyeager/build-sm50 | Nathan Luehr | |
Build SM 5.0 code | |||
2015-12-19 | Build SM 5.0 code | Luke Yeager | |
2015-12-18 | Use semantic versioning | Luke Yeager | |
2015-11-17 | Initial release. | Nathan Luehr | |