Compilation workloads are not really bandwidth sensitive at all. The existing quad-channel 32 core 2990WX compiles the Linux kernel faster than the 8 channel 32 core EPYC 7601 thanks to its higher clock speeds in spite of having half the memory channels. So I think it is safe to say that memory bandwidth is a total non-issue for this type of workload.

Benchmarks here:

https://www.phoronix.com/scan.php?page=article&item=amd-linu...

https://www.phoronix.com/scan.php?page=article&item=amd-epyc...

The AMD Threadripper 2990WX has 32 cores with quad-channel memory, and it seems to do fine.

Would be good to see some benchmarks.