All GPUs have different optimization points. Many optimizations are common but do not assume an application optimized for one platform automatically performs well on another.
For example, ARM recommends you sort objects or triangles into front-to-back order in your application. This enables early culling of fragments, reduces the load on the fragment processor, and reduces overdraw.
This optimization is not unique to Mali GPUs, it also works on some other mobile GPUs and desktop GPUs.