Hello Prophesee Team,

I successfully compiled OPENEB 4.3.0 on arm64 Ubuntu 20.04 using the guide provided at https://docs.prophesee.ai/4.3.0/installation/linux_openeb.html

Additionally, I managed to compile and run my C++ application, that utilizes the Metavision Pipeline. However, I encountered performance issues which left me dissatisfied.

In an attempt to optimize performance, I referred to the default example for a pipeline available at: https://docs.prophesee.ai/4.3.0/guides/sdk_pipelines/pipeline_metavision_filtering.html

I streamlined the example by retaining only the camera and polarity filter stages. Despite this simplification, the performance remained subpar, with the pipeline consuming approximately 110% CPU usage.

To gauge CPU usage, I utilized both jtop and htop.

I am seeking advice on enhancing the performance of the pipeline_metavision_filtering.

1. Are there any additional flags for building OPENEB or the application that are not documented? Currently, I am using the following commands for building:

cmake .. -DCMAKE_BUILD_TYPE=Release

cmake --build . --config Release

2. While monitoring with htop, I noticed that the default metavision_filtering spawns 9 processes. The main process consumes approximately 150% CPU, while the 6 childs processes exhibit CPU usage ranging from 0% to 10% each, and 2 childs — 10%-15%.  Is there a way to understand which process represents what stage of pipeline, or whatever else OPENEB uses internally?