DL traffic not flowing; gNB stuck at RB=5, MCS=9 on Sionna-RK (DGX Spark)

Accelerated Computing DGX Spark / GB10 User Forum DGX Spark / GB10 Projects
1

Hi,
I’m testing DL traffic with OAI/Sionna-RK on an NVIDIA DGX Spark (rfsim).
I run:

  • docker exec -it oai-ext-dn iperf3 -s -B 192.168.72.135
  • docker exec -it oai-nr-ue iperf3 -c 192.168.72.135 -B 12.1.1.2 -t 30 -u -b 10M -R

Symptoms

  • gNB logs always show NPRB 5, MCS 9 (QPSK, low code rate), never scaling up.
  • UE receives effectively no DL traffic; AMF shows UE registered, but DL data doesn’t flow.

Questions

  • Is the fixed RB=5/MCS=9 likely why 10 Mbps DL fails?
  • How to force/allow the scheduler to allocate more RBs and higher MCS (e.g., 64QAM) in Sionna-RK/OAI? Pointers to config or C code locations are welcome.

Notes

  • Running: start_system.sh rfsim_arm64
  • Images: oai-gnb-cuda:latest, oai-nr-ue-cuda:latest
  • Tried setting: dl_min_rb/dl_max_mcs via .env, but MAC logs stay at RB=5, MCS=9

Thanks!

docker gnb log.txt (15.0 KB)