Tests & Data
Understand product behavior under specific configurations and operating conditions. Each article explains measurement boundaries, test conditions and statistical methods, with charts and data to support engineering decisions.
Selection & control overview: compares the ~9× latency gap from SocketCAN and write(can0) boundaries, and calculates host compute windows in 1–2 kHz multi-joint loops.
Read the technical article →USB-CAN FD Real-Time Latency Benchmark Report: Methodology, Multi-Device Statistical Analysis, and Raw DatasetsFull benchmark report: T1–T4 timing model, ISO 11898-1 per-frame bit-stuffing deduction, 9-setup quantile tables, variable isolation, 30-round stability, 11-point sweeps, and raw datasets.
Read the technical article →Reading test results
Establish the measurement scope before comparing values. Latency can refer to application, kernel or bus timing boundaries. Read the device, firmware, driver and operating conditions together.
Consider typical behavior and the tail. Mean and P50 describe typical behavior; P99 and the distribution describe slower samples. An observed maximum applies to the stated sample population.
Carry the conditions into your engineering budget. Frame length, transmission period, channel count and host scheduling affect how results apply to a real system.