Dynamic rendering vs RenderPass #327
Description
I see inconsistency when commenting on advantages of using dynamic rendering vs traditional render passes. Particularly, the section on Mobile Development: Vulkan Extensions is quite affected. For example, Overview section gives the impression that it is just more programmer and resource friendly API. On the other hand side, Device Extension Support Details tells us that dynamic rendering "can lead to significant performance improvements compared to traditional render passes, especially on tile-based renderers." Actually, the last statement might seem just like a belief of somebody when compared with Overview section.
A reader might be confused of this. Actually, is it harmful to use traditional render passes today? The last cited statement seems to give such message.
In ideal world, I would like to hear: dynamic rendering always provides more performance than traditional render passes, or in the worst case, it provides the equal performance. In reality, it is possible to find information on internet that render passes are more beneficial on mobile devices than dynamic rendering, such as on khronos blog and search for "multiple subpasses". Moreover, much of dynamic rendering promoting words look like an advertisement, skipping all performance discussion. So, some confusion exists here. If Vulkan-Tutorial could clarify this point, possibly based on the knowledge of Vulkan WG, this might be very helpful.