Sylvain Jeaugey cabd6848e4 Heavy code refactoring to remove a lot of code in collectives (~1000 lines). ...

Have all collectives use the same args, the same ring, and the same primitives for synchronization between threads with the same pattern.

2016-09-22 11:57:56 -07:00

..

all_gather.cu

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

all_reduce.cu

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

broadcast.cu

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

common_kernel.h

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

copy_kernel.h

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

core.cu

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

core.h

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

enqueue.h

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

libwrap.cu

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

libwrap.h

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

nccl.h

Add profiling API

2016-09-22 11:56:51 -07:00

primitives.h

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

reduce_kernel.h

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

reduce_scatter.cu

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00

reduce.cu

Heavy code refactoring to remove a lot of code in collectives (~1000 lines).

2016-09-22 11:57:56 -07:00