In our four-threaded random read tests, the best figure we saw was 416,861 IOPS at QD32, a long way short of the maximum 650,000 IOPS. We ran a quick test at QD32 using 8 threads and saw the performance rise to 590,013.1 IOPS, still shy of the official maximum.
4K Random Read v QD Performance Compared.
The drive's performance picks up as the queue depth deepens with the best performance in comparison with its competitors coming at QD4 at 222,526 IOPS.