That shouldn't be a problem, pv is basically only counting the number of pages being written. As long as the page size is much larger than the number of threads it should keep up just fine.
The page size can't be more than half the L2 cache (else writeback to DRAM would be triggered, dramatically lowering performance). That means you have to use a relatively small page size, and lock contention in the kernel for those pages then limits throughput if you use many cores, even if the data is never read by pv.
