I captured system and GPU info via Streamline 8.3 (Build 8.3.0.v20221116_1421) on ubuntu 20.04 successfully,
while i cannot analysis the output with "apc" suffix via Performance Advisor 8.3 (Build 8.3.0.v20221116_1421).
pa Capture_3588.apc/ -p video_stereo_server_V2 Importing capture... Fetching data... Preparing report type html... Problems were found preparing html report: WARNING: Build name not provided. You can add this as a command line argument (use --build-name). WARNING: Build timestamp not provided. You can add this as a command line argument (use --build-timestamp). WARNING: Device name not provided. You can add this as a command line argument (use --device-name). WARNING: Cannot find draw calls counter. Use LWI or GA interceptor when creating your capture to obtain draw calls data. ERROR: Cannot find any frames. Please refer to the user guide for methods of providing frame data to Performance Advisor
is there something wrong with these tools call, or any solution to solve this problem
Thanks a lot. I notice it after reading the instruction again.
Instead of building an android app package and running lwi_me.py, I compiled a single GPU test program by aarch64-linux-android-gcc and tested it manually through "gatord --allow-command".
In this situation, I will get an error like "ERROR: No debuggable packages found" if I run lwi_me.py, because the program to be tested is not a real Android package.
So is there a way for TCP users to use Performance Advisor
Thanks Arthur, understood.Are you able to modify the source of the GPU test program? You can manually add frame information using our annotation library, the full details are described here developer.arm.com/.../Send-annotations-from-your-application-code and here developer.arm.com/.../Include-Streamline-annotations-in-native-applications. If you send an annotation from your program at the beginning of each frame, then Performance Advisor can pick those up to perform analysis.Please let me know how you get on and if you have any more questions.
It worked! Thanks a lot!!