nicusorcitu
nicusorcitu
BTW, I saw you reworked on master branch the need_argb8565_support() so please add the "draw_ctx" argument everywhere is called. Is a very long process in order to create PR from...
I add for reference also the patch that checks the area intersection. [0010-Add-static-array-to-hold-the-pending-GPU-draw-areas-.txt](https://github.com/lvgl/lvgl/files/9281420/0010-Add-static-array-to-hold-the-pending-GPU-draw-areas-.txt) The comparing results are below: | NO wait_for_finish() | wait_for_finish() | wait_for_finish() BUT not from blend() |...
“NO wait_for_finish()” means there is no callback implementation, cureently upstreamed version in v8.2.0. E.G. The CPU flushes a command to the GPU and waits right away for its completion. ________________________________...
> I see. So NO wait_for_finish() is kind of "blocking" mode and it's the slowest. > Yes, without wait_for_finish() implementation, the CPU and GPU can not work in parallel. >...
Off Topic: Why is there need for a new fill screen (with white color) every time the screen is refreshed? E.G. at every 30ms? Is this a kind of init...
Hi Gabor, I have improved the systemview support and now I can see all the tasks and messages correctly. The default benchmark application uses rgb100x100 and that one is not...
1. This is FPS measurements. I need to find a way to improve this, the only way would be to make CPU and GPU work more time in parallel. That...
In some cases, e.g. the fill screen, the time to wait is 4 ms, so it worth yielding. But I agree with you for cases where does not have to...
Till now I concluded that checking if the CPU area is intersecting with the GPU area actually has a bigger cost than simply calling the `wait_for_finish` useless. I think we...