max_proc_delay = 1 not working.

[arman-maghsoudnia]

Hi,
There is an available configuration named max_proc_delay.

expert_phy: 
max_proc_delay: 5                         # Optional INT (5). Sets the maximum allowed DL processing delay in slots. Supported: [1 - 30].

Previously when I used to set it to 1, it worked perfectly fine. But now after updating the code to the latest, when I set it to 1 the UE cannot connect to the network anymore. Do you have any idea what can be the cause?

[AlaiaL]

Hi Arman,

Can you share your gNB configuration?
Also, could you set the log level to debug, do a quick run and share the gNB log?

You can set the log level to debug in this section of the gNB configuration:

log:
  filename: /tmp/gnb.log            # Path of the log file.
  all_level: debug                          # Logging level applied to all layers.

Thanks,

Alex

[arman-maghsoudnia]

Hi Alex,

I guess I can run the following script provided by the SRS team to set the CPU to performance mode, right? https://github.com/srsran/srsRAN_Project/blob/main/scripts/srsran_performance
If yes, I already did it and tried to connect again, but again it does not connect.
Setting the variable to max_proc_delay: 2 will fix the problem but it will increase the latency which is not favorable for some applications. As I said it used to work when I had executed the gNB code with this commit ID: b5621f12d4cc97b816b5147e9705acad7a965fe8 . But after updating to later releases, it stopped working with max_proc_delay: 2. This is the configuration I used to use back then when it was working:

gnb_id: 107

amf:
  addr: 10.43.186.29
  bind_addr: 10.42.2.83

ru_sdr:
  device_driver: uhd
  device_args: 
  srate: 23.04
  otw_format: sc16
  tx_gain: 65
  rx_gain: 15

cell_cfg:
  dl_arfcn: 632628
  band: 78
  channel_bandwidth_MHz: 20
  common_scs: 30
  plmn: "00101"
  tac: 1
  pci: 1
  tdd_ul_dl_cfg:
    dl_ul_tx_period: 4
    nof_dl_slots: 2
    nof_dl_symbols: 6
    nof_ul_slots: 1
    nof_ul_symbols: 7
  nof_antennas_ul: 1
  nof_antennas_dl: 1
  pdsch:
    mcs_table: qam256
  pusch:
    mcs_table: qam64
    min_k2: 1
  pucch:
    sr_period_ms: 2
    min_k1: 1

expert_phy:
  max_proc_delay: 1

Shall I creat an issue from this discussion, so you can fix it in the new release?

[AlaiaL]

Hi Arman,

yes, that script sets the CPU to performance. Did you tested again after executing the script with sudo?

Shall I creat an issue from this discussion, so you can fix it in the new release?

There is no need, we already opened one when you created this thread.

Thanks,

[arman-maghsoudnia]

Did you tested again after executing the script with sudo?

Yes, still it does not connect. I even used another machine with a more powerful CPU and executed the script with sudo and still the same issue.

[AlaiaL]

Hi Arman,

Yes, still it does not connect. I even used another machine with a more powerful CPU and executed the script with sudo and still the same issue.

Sorry to hear that.

if you want to do a quick test, if you use our current main commit and go to this line:

srsRAN_Project/apps/gnb/gnb_appconfig_translators.cpp

Line 1664 in 2f90c8b

unsigned ul_pipeline_depth = dl_pipeline_depth;

you can update that line with:

unsigned ul_pipeline_depth = std::max(dl_pipeline_depth, 8U);

Then build the code and test. This change should fix the issue.

[arman-maghsoudnia]

@AlaiaL Thanks for the help.