My understanding is that it's a pretty small optimization on the front end. It doesn't have a lot of entries to begin with (144) so the amount of space saved is

My understanding is that it's a pretty small optimization on the front end. It doesn't have a lot of entries to begin with (144) so the amount of space saved is probably negligible. Theoretically, the loop buffer would let you save power or improve performance in a tight loop. In practice, it doesn't seem to do either, and AMD removed it completely for Zen 5.