In rapid mode, there are indeed other factors that contribute also.
In normal mode, where there is no such cache-ing to ram,i believe that for cpu speed raised, 4K increases too.
Not meaning that the cpu is a bottleneck and it should be oc’d to give more headroom to AS-SSD, just that the I/O’s are affected by CPU also.
23% shows only a percentage of the power capacity. 23% of 3.5GHz CPU is not the same power with 23% of 4.5GHz CPU, is it ?
If you do some further runs, you can doublecheck it yourself, e.g between 3.5 and 4.5, unless the bottleneck is the SSD ( i think this is the highest 4K result i’ve seen in all the Pro reviews, don’t know if it can go higher)
I have discussed it over at oc.net with Doorules, while he was having the M3 Pro’s in his X79 system. Non-ignorable differences at 4K read/write betweeen stock and OC’d cpu. Over the years, i’ve used from Vertex Turbo’s to 840 Pro’s ( even Renice E7 IDE 
) and same SSD’s with different CPU’s or same CPU’s different clocks, behaved differently.
That is also the way some other review sites mess up with SSD review results. C states enabled on some reviews , Disabled on others, and the end result is to have numerous SSD benchmarks non-comparable to each other over the internet.
Of course, i cannot demand from all reviewers to communicate each other and use the same test machines and same settings, but there are some things that strongly affect results.
Testing is a serious procedure and it should be conducted in way to minimize error and output a repeatable and linked-to-reality result ( that is what you have done in your review and should be respected).
I hope that my above opinion is simply taken in a friendly way, not attempting to spoil the review or play smartass, i was just asking for detail about cpu. 
