Upload & Sell: On
The problem seems to be that they have chosen a rendering pipeline process that is inherently very CPU intensive, and somewhere in the transition from V3 to V4 one or more of the algorithms (probably NR) have changed and become a bottleneck. Perhaps the NR algorithm needs some tuning, perhaps slowness is inevitable for the NR method they have chosen to implement and is the price we have to pay for the excellent NR that LR does. We don't know.
To me, this is pertinent to the OP question @ whether or not a change (of any kind) will enact performance improvement. Assuming the operation to be considered to be CPU intensive, then the issue of performance improvement hoped for would be relegated more to the discussion of a change from XEON processors to the processor in the proposed change (Mac, PC or otherwise) ... rather than brand X vs. brand Y.
It is possible to think that a Honda is more efficient than a semi, and maybe it is. That doesn't mean that changing from a semi to a Honda will get you more power. The move away from XEON processors may not realize any improvement in LR due to the bottleneck. But, I can't help but wonder if the move away from XEON processors (we don't know which ones) was a step backward and the bottleneck is masking that. Kinda depends on which XEON processors (multiple ) are in the workstation.
I genuinely aspire for the OP (Mac or otherwise) to be pleased with his setup, but I can't help but wonder if the discussion @ PC vs. Mac bypassed the issue of how LR works and what components he would be changing from & to. Until we know which (& how many) XEON processors it's tough to know if the change was a step forward or backward for the setup, if LR is masking performance with an inherent bottleneck.