Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Revert "neuron: Disable rdma eager messages by default" #650

Merged
merged 1 commit into from
Oct 7, 2024
Merged
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 1 addition & 13 deletions include/nccl_ofi_param.h
Original file line number Diff line number Diff line change
Expand Up @@ -261,20 +261,8 @@ OFI_NCCL_PARAM_INT(net_latency, "NET_LATENCY", -1);
/*
* Eager message size limit when using RDMA protocol. Message sizes greater than
* this limit will always be sent using RDMA write instead of eagerly.
*
* Neuron perf is better without the eager protocol, so we set the
* default to 0 on Neuron platforms. We really need to have a way to
* tweak defaults from the platform file, but this fits our needs for
* now.
*/
OFI_NCCL_PARAM_UINT(eager_max_size,
"EAGER_MAX_SIZE",
#if HAVE_NEURON
0
#else
8192
#endif
);
OFI_NCCL_PARAM_UINT(eager_max_size, "EAGER_MAX_SIZE", 8192);

/*
* Decide whether or not mutexes should default to errorcheck mode.
Expand Down
Loading