Our 4KB random performance test is conducted in the same manner as our sequential tests, but once the drive is conditioned we run our saved random test profile that runs our 4KB test for two minutes without any idle time in between the tests. The queue depth is set to 32 on four workers and the test is begun. To get the benefits from NVMe based drives you must use multiple CPU queues and this is why we are now using four workers for this IOMeter test.
IOPS is the main thing we are looking at in this test scenario and the Crucial BX200 480GB drive is rated at up to 66,000 IOPS for the 4K Random Read and 78,000 IOPS for the 4KB Random Writes. On our properly conditioned drive we hit 71,000 IOPS Read, which is higher than the drive is rated for! On the write test we found the drive scored 25,000 IOPS for our prolonged test since the cache filled up. If we ran the test for just 10 seconds (before the cache filled) we found that we were able to get 83,400 IOPS Write. This is again well over the drives rated 78,000 IOPS for Random Writes.
When it comes to MBps you are looking at 292 MB/s on the 4KB Random Reads and 104 MB/s on the 4KB Random writes.
The response times on the Crucial BX200 480GB SSD were 5.05 ms for writes and 1.80 ms for reads, which is what we were kind of expecting for this TLC drive since the 6GB SLC cache was filling up on this prolonged write test.