Didn’t we see some benchmarks where it seemed that the integrated GPU of some Apple chip was very fast, probably because of a lack of memory transfers, having access to a lot of memory, and therefore no tiling?