Added RCCL env params to control setting the SO_REUSEADDR and SO_LINGER socket options

[corey-derochie-amd]

Details

Work item: SWDEV-491354

What were the changes?
Added RCCL_SOCKET_REUSEADDR and RCCL_SOCKET_LINGER env params to control setting the SO_REUSEADDR and SO_LINGER socket options.

Why were the changes made?
This can allow control over the number of file descriptors created during bootstrapping.

How was the outcome achieved?
Conducted tests of FD count while repeatedly running AllReduce under the given test conditions, and saw a reduction with RCCL_SOCKET_LINGER=0 and =2, and a change in behaviour over time using RCCL_SOCKET_REUSEADDR=1.

Additional Documentation:
This is still under investigation.

Approval Checklist

Do not approve until these items are satisfied.

• Verify the CHANGELOG has been updated, if
  • there are any NCCL API version changes,
  • any changes impact library users, and/or
  • any changes impact any other ROCm library.