Ideal bitmap size for optimal GPU performance?

Question

I know that when working with a graphic card, you need to provide bitmap images with a power of 2. 1,2,4,8,16,32 ... I plan to make a project based on a tile, only I needed to know if any limitations in providing a giant raster image for a video card. What are the advantages / disadvantages of using small / large bitmaps? Is there a more preferred bitmap size that I should use?

Answer 1

In general, when it comes to modern GPUs, a larger data collection system is better. You do not say which API you are using (OpenGL, CUDA, etc.), but you can think of it this way:

chunk_time = overhead_time + (num_of_elements/num_of_chunks) * per_element_time total_time = chunk_time * num_of_chunks 

You will get the overhead for both memory transfer and code execution for each piece of data that you send. You may have other restrictions depending on how huge your data is: for example, the maximum size of texture binding in OpenGL depends on the implementation. I think 1k on 1k should be a safe minimum with equipment for the last 5 years or so, but the latest cards can handle 8k on 8k.