Tool/software:
Dear TI Support Team,
I am currently working with the TDA4AL chip running QNX OS, and I need to measure the resource utilization of the C7x and C7+MMA deep learning accelerator, checking this NPU's 8 TOPs are enough or not.
We are already using perf analyzer, monitoring only CPU(A72) resources. Given this setup, I would like to confirm the following:
- Are there any specific libraries or binaries that must be installed to accurately measure the resource usage of the C7+MMA accelerator?
- Is it possible to obtain resource utilization data via sysfs or procfs interfaces on QNX for the TDA4AL platform? If so, could you please provide the relevant file paths or usage instructions?
- Alternatively, are there APIs or sample codes available within the TI SDK that allow for direct measurement of C7+MMA usage? If so, I would appreciate detailed guidance or references to documentation.
I would be grateful if you could confirm whether the approaches I described above are correct, and advise on the recommended method for resource monitoring in this environment.
Thank you for your support.
Best regards,