01
Multi-platform evaluation overview
Summarizes evaluation status across J6P, J6M, S100, and X5, covering eight compilation and on-device performance stages with traceable evidence for every key step.
AI EVALUATION
AI Evaluation（Coming soon）
Run algorithms faster, more reliably, and efficiently on target silicon.
02
ONNX model inspection
Identifies model versions, node counts, and operator types while checking dynamic shapes, control flow, external weights, and custom operators before silicon adaptation begins.
03
Operator and structure analysis
Measures the count and share of convolution, activation, pooling, and other operators to guide platform support validation and performance optimization.