Not so much as the 70 classes 192 bit bus but it’ll still be impacting performance at 4k.
AMDs problem is the RDNA 3 architecture just isn’t very good and you can see that when comparing an 84 CU 7900XT to an 80 CU 6950XT.
I mean it's not that bad, it gets like 5% more CUs and performs maybe 10-15% better. Really it just needs to be a bigger part.
The main thing is just AMD obviously didn't set a particularly ambitious target for how much silicon they wanted to spend on the GCD which has lead to it being more of an incremental improvement over Navi21 in general. The XTX is a little faster but it's far from mind blowing.
For contrast AD102 is bigger than the entire Navi31 package and a lot of Navi31 is taken up by MCDs produced on a much less dense node.
I mostly think it's a bit of a teething issue with the MCM process that led them to be a bit more conservative this gen, they will hopefully learn a lot from this generation about how to better scale it for performance.