Assuming that your workloads are standard non-gaming workloads, they will probably also benefit substantially when gaming benefits. My reasoning being that a common bottleneck for gaming workloads is single thread performance, and unless this is highly focused on a single core clock boost, I'd expect that single-threaded performance increases would be multiplied appropriately by core count, though perhaps not perfectly linearly.

When you see marketing reaching for "non-gaming" metrics, it's often for highly parallelizable workloads, which benefits non-gaming disproportionately, but e.g. for compiling/linking there are still tasks that have to be done serially, which is where that single-threaded performance boost is critical.

I'm definitely excited at this point to see what a Zen 3 threadripper can bring to the table.