Skip to content

NR UE improve Tx thread

Sakthivel Velumani requested to merge NR_UE_mthread_rework into develop

Changes:

  1. Start TX processing of slot n+4 after receiving samples of slot n. So TX (n+4) and RX (n) runs in parallel. Existing: Only PUSCH (n+5) and PDSCH (n) runs in parallel.
  2. Call RU write function from TX process thread instead of main thread. The two threads rendezvous before receiving samples for n+1 so functionally same as before.

Documentation here.

This is an intermediate MR to the final multi-thread design changes to come. No improvement in performance is expected as we still run slots processing in series.

Note: This branch can be merged after !1797 (merged).

Edited by Sakthivel Velumani

Merge request reports