Brian Barrett a4fabb7e26 openib: Disable CUDA async by default ...

Disable async receive for CUDA under OpenIB.  While a performance
optimization, it also causes incorrect results for transfers
larger than the GPUDirect RDMA limit.  This change has been validated
and approved by Akshay.

References #3972

Signed-off-by: Brian Barrett <bbarrett@amazon.com>
(cherry picked from commit 9344afd485d23c41a18ab57d6df8550826c24b9e)
Signed-off-by: Brian Barrett <bbarrett@amazon.com>

2018-09-20 14:20:54 -07:00

..

base

opal/btl: add support for flushing RDMA/atomic operations

2018-02-13 12:49:51 -07:00

ofi

btl/ofi: Added FI_CONTEXT as requirement.

2018-07-17 12:18:43 -07:00

openib

openib: Disable CUDA async by default

2018-09-20 14:20:54 -07:00

portals4

btl-portals4: don't free module resources when proc count goes to zero

2018-04-10 14:20:22 -05:00

self

mca: Dynamic components link against project lib

2017-08-24 11:56:16 -04:00

sm

btl/sm: fix CID 1415105

2018-03-26 14:21:21 -07:00

smcuda

Fix typo

2018-04-11 10:29:43 +08:00

tcp

tcp/btl: fix a cast

2018-06-29 07:25:46 -07:00

template

mca: Dynamic components link against project lib

2017-08-24 11:56:16 -04:00

uct

btl/uct: add missing opal_mem_hooks_unregister_release call

2018-09-05 14:02:58 -06:00

ugni

btl/ugni: fix up some warnings

2018-07-03 16:30:44 -06:00

usnic

btl/usnic: missed a preprocessor check in d36648b

2018-02-19 07:03:05 -08:00

vader

btl/vader: ensure that the send tag is always written last

2018-09-14 12:34:26 -06:00

btl.h

btl/vader: add support for atomics and emulated rdma

2018-07-02 13:57:11 -06:00

Makefile.am

Purge whitespace from the repo

2015-06-23 20:59:57 -07:00