qq20739111

After nearly one year of kernel crashes, I finally found the 
stable configuration for Orange Pi Zero 3 as a 24/7 SDR server.
 
The critical discovery: legacy kernel 6.6.75-sunxi64 is the only
stable option. The current kernel crashes within hours under SDR load.
 
Additional fixes:
- Blacklist aw859a WiFi driver (causes kernel instability)
- Blacklist DVB modules for RTL-SDR
- Lock libairspyhfSupport.so with chattr +i (SpyServer/SoapySDR conflict)
- Mask systemd-networkd-wait-online
- CPU governor: schedutil
- usbcore.autosuspend=-1 for USB SDR dongles
 
Running 3 SDRs simultaneously: RTL-SDR Blog V4, Airspy HF+, 
dedicated ADS-B. 4 antennas via HMC241 GPIO RF switch.
30+ decoders. Full Android control.
 
Complete guide: https://ozcet.github.io/sdr-server-guide

Support for DP Alt Mode is now functional for the A7S running kernel 6.6.


https://github.com/NickAlilovic/build/releases/tag/Radxa-cubie-A7S-v0.6.4

I made some patches so that the unified driver / TIM-VX for the NPU works on the A7Z. I am also building a MLIR pipeline that emit TIM-VX code, so hopefully we can get more flexibility running ML models on the SBC soon.
https://github.com/MaverickLong/Radxa-A733-NPU-Unified-Driver-Support-Package

@qq20739111 thks, official community = github? https://github.com/radxa-docs/docs/

i asked claude to make a summary of my project with npu:
 
We use the vendor's ACUITY toolkit inside a Docker container (ubuntu-npu:v2.0.10.1) to convert ONNX → quantized .nb files. - Pipeline: onnxsim → pegasus_import → pegasus_quantize → pegasus_export, uint8 quantization, deployed to ~/npu_models/ via push.sh. - Runtime: custom npu_server.c (pre-allocated input buffer, no per-call mmap) that the Python app talks to. - Hard lesson #1 — quantization: only pure Conv+BN+ReLU survives uint8. Attention, SE blocks, hard-swish, LayerNorm all collapse to constant outputs. So no MobileNetV3+, no transformers. - Hard lesson #2 — concurrency hang: NPU IRQs get lost when camera ISP DMA runs in parallel (shared memory bus). Fix: suppress GStreamer buffer copies during inference. Never STREAMOFF/ON the sunxi-vin driver — instant kernel crash. - Result: 12 models running (9 NPU + 3 CPU) at ~40ms/inference.  


 

@qq20739111  I'll add it soon. 6.18 might take awhile.