Changpeng Fang eec319038e Tuning the inline and unroll to reduce the scratch usage ...

Summary:
 1. remove the noinline attribute for AllReduceThreeKernel;
 2. change AUTPUNROLL for tree functions to 1 or 2;
 Combining 1 and 2 will reduce the scratch usage from 1256 to 952

2019-10-08 14:02:25 -07:00

..

collectives

Tuning the inline and unroll to reduce the scratch usage

2019-10-08 14:02:25 -07:00

include

Merge remote-tracking branch 'remotes/nccl/master' into HEAD

2019-08-16 16:13:34 -07:00

misc

Merge pull request #125 from wenkaidu/fix_nvml_id

2019-08-19 09:08:13 -07:00

transport

RDMA HDP flush fix

2019-09-06 16:35:55 +00:00

bootstrap.cc

2.4.8-1

2019-06-25 13:22:47 -07:00

channel.cc

RCCL 2.4 update

2019-08-14 10:42:35 -07:00

enqueue.cc

RCCL 2.4 update

2019-08-14 10:42:35 -07:00

init.cc

Merge remote-tracking branch 'remotes/nccl/master' into HEAD

2019-08-16 16:13:34 -07:00

Makefile

Allow CUDA runtime library selection (#220)

2019-05-07 17:35:14 -07:00

nccl.h.in

RCCL 2.4 update

2019-08-14 10:42:35 -07:00

nccl.pc.in

Add pkgconfig file (#190)

2019-04-08 09:16:54 -07:00

transport.cc

RCCL 2.4 update

2019-08-14 10:42:35 -07:00