Sylvain Jeaugey c0c210d2d2 Make NCCL collectives work on communicators with only one rank ...

[ROCm/rccl commit: 7edfc57228]

2016-06-06 14:35:00 -07:00

..

all_gather.cu

Make NCCL collectives work on communicators with only one rank

2016-06-06 14:35:00 -07:00

all_reduce.cu

Make NCCL collectives work on communicators with only one rank

2016-06-06 14:35:00 -07:00

broadcast.cu

Make NCCL collectives work on communicators with only one rank

2016-06-06 14:35:00 -07:00

common_kernel.h

Fixed buffer overflow in ReduceOrCopy

2016-02-12 15:13:56 -08:00

copy_kernel.h

Initial release.

2015-11-17 11:30:40 -08:00

core.cu

Fix random deadlock during ncclCommInitRank.

2016-04-19 10:47:27 -07:00

core.h

Added support for more than 8 GPUs.

2016-01-21 13:00:21 -08:00

enqueue.h

Initial release.

2015-11-17 11:30:40 -08:00

libwrap.cu

Libwrap checks for LIB.so.1 if LIB.so not found

2016-01-29 12:36:42 -08:00

libwrap.h

Initial release.

2015-11-17 11:30:40 -08:00

nccl.h

Fixed deadlock in back-to-back reduce_scatters.

2016-01-21 10:36:03 -08:00

reduce_kernel.h

Enabled support for char type to be unsigned.

2016-01-28 13:38:18 -08:00

reduce_scatter.cu

Make NCCL collectives work on communicators with only one rank

2016-06-06 14:35:00 -07:00

reduce.cu

Make NCCL collectives work on communicators with only one rank

2016-06-06 14:35:00 -07:00