It seemed like the DDR5 wasn't really given the best chance to reach its full potential with only half of the memory channels used? I think to make it an accurate test you would need to include quad channel DDR5 since each channel has a narrower bus than DDR4?
An example of the difference this can make is comparing memory bandwidth numbers between Threadripper and Ryzen chips.
An example of the difference this can make is comparing memory bandwidth numbers between Threadripper and Ryzen chips.