Skip to content
This repository was archived by the owner on Sep 30, 2022. It is now read-only.

pml/ob1: be more selective when using rdma capable btls #1315

Merged
merged 1 commit into from
Aug 16, 2016

Conversation

hjelmn
Copy link
Member

@hjelmn hjelmn commented Aug 10, 2016

This commit updates the btl selection logic for the RDMA and RDMA
pipeline protocols to use a btl iff: 1) the btl is also used for eager
messages (high exclusivity), or 2) no other RDMA btl is available on
an endpoint and the pml_ob1_use_all_rdma MCA variable is true. This
fixes a performance regression with shared memory when an RDMA capable
network is available.

Signed-off-by: Nathan Hjelm hjelmn@lanl.gov

(cherry picked from commit open-mpi/ompi@4079eec)

This commit updates the btl selection logic for the RDMA and RDMA
pipeline protocols to use a btl iff: 1) the btl is also used for eager
messages (high exclusivity), or 2) no other RDMA btl is available on
an endpoint and the pml_ob1_use_all_rdma MCA variable is true. This
fixes a performance regression with shared memory when an RDMA capable
network is available.

Signed-off-by: Nathan Hjelm <hjelmn@lanl.gov>

(cherry picked from commit open-mpi/ompi@4079eec)
@hjelmn
Copy link
Member Author

hjelmn commented Aug 10, 2016

:bot🏷️bug
:bot:milestone:v2.0.1
:bot:assign: @hppritcha

We have confirmation from Tetsuya that this indeed does improve the bandwidth. The patch restores the v1.10.x behavior of only using the highest exclusivity btls for send/recv. I added an MCA parameter to allow using all available RDMA btls.

@mellanox-github
Copy link

Test PASSed.
See http://bgate.mellanox.com/jenkins/job/gh-ompi-release-pr/2080/ for details.

@hppritcha
Copy link
Member

👍

@jsquyres
Copy link
Member

@hppritcha Good to go.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants