nicusorcitu

VG-Lite driver already have an internal thread for executing the command buffer. Hence, I am also curious what is the performance difference with a VGLite draw thread versus without. In...

You can always have that if you set LV_USE_OS to none. https://github.com/lvgl/lvgl/blob/590363f7429cd0e128f8c030f22006188bac7981/src/draw/nxp/vglite/lv_draw_vglite.c#L113 Just to be sure, I am trying to put my comments in hope that it will be a...

@cristian-stoica

1. That would be indeed an benchmark closer to the real life application. It will be very good to evaluate the performance overall. Will be nice to have also few...

> Because we cannot control the target software system that might have a lot of application tasks, measuring the RTOS idle time will be affected by many factors out of...

> Here, we are measuring CPU load, not rendering time for each frame. They are different. > Exactly my point. When I say task I mean graphic draw task. We...

> What do you mean by stop frame time? > I mean start draw frame and stop draw frame - that we had it in [lvgl 8.x](https://github.com/lvgl/lvgl/blob/73fa50afa0f075a0ff5875ee6cd314b6404967a1/src/core/lv_refr.c#L348) I see that...

When we say CPU load we mean the CPU load of the system. All the time. If we look at the CPU load while running the Benchmark App only, we...

Sorry, I do not mean to have the final decision. I don't have to. :) I am only trying to signal some issues that I faced and the solution I...

If the purpose is to measure the CPU load ONLY by LVGL tasks this can be achieved very easy on FreeRTOS by hacking the `traceTASK_SWITCHED_IN()` and `traceTASK_SWITCHED_OUT()` and count the...