STM32N6 : CubeAI ?? Epoch Issue and Why PReLU Runs in Software After Quantization

qiqi · ‎2025-06-23

Hello everyone,

I am using CubeAI 10.1.0 and STAGEAI 2.1 to analyze my model for STM32N6, and I encountered an issue where some epochs show ?? instead of the expected results. Here's the log:

epoch ID HW/SW/EC Operation (SW only)
epoch 1 EC
epoch 2 EC
epoch 3 -SW- (DequantizeLinear)
epoch 4 -SW- (PRelu)
epoch 5 -SW- (QuantizeLinear)
epoch 6 -SW- (MaxPool)
epoch 7 EC
epoch 8 EC
epoch 9 -SW- (DequantizeLinear)
epoch 10 -SW- (PRelu)
epoch 11 -SW- (QuantizeLinear)
epoch 12 EC
epoch 13 EC
epoch 14 -SW- (DequantizeLinear)
epoch 15 -SW- (PRelu)
epoch 16 EC
epoch 17 -SW- (Conv)
epoch 18 -SW- (Add)
epoch 19 EC
epoch 20 -SW- (Conv)
epoch 21 -SW- (Add)
epoch 22 ??
epoch 23 -SW- (Add)
epoch 24 ??
epoch 25 -SW- (Add)
epoch 26 EC

In epoch 22 and epoch 24, the result is shown as ??, and I couldn't retrieve any computation results. I have a few questions:

1. What does ?? mean?

Does the ?? represent that some operators or operations failed to execute during these epochs? Does it imply that those operators are not supported on STM32N6, or could it be due to hardware resource limitations?

2. Will this affect model results?

If epoch shows ??, will it impact the final recognition or inference accuracy of the model? Should I be concerned that this issue may lead to unreliable results from the model?

3. Why is the PReLU operator still executed in software after quantization?

The official documentation mentions that PReLU is supported on STM32N6, but after model quantization, the computation for PReLU is still executed in software rather than on the hardware. Why is that? Is it due to hardware limitations, or is STM32N6's hardware acceleration for this operator not fully optimized? Is there any other reason why PReLU still runs in software?

4. How can I optimize the model to avoid these issues?

If these issues occur, are there any recommended optimization methods or adjustment strategies to address them and ensure that the model runs smoothly and gives accurate results? Should I consider simplifying the model or replacing PReLU with another activation function to avoid the operator being executed in software?

Thank you in advance for your help and suggestions!