Anvil’s Storage Utility allows us to configure the size of our test file. We start off with a 1 GB file after the drive is secure-erased, returning it to a fresh-out-of-the-box state. We then refill the drive with data copied over from another repository. Specifically, we moved the Windows folder, the Program Files folders, and a number of folders composed of user data, leaving 50.4% free space.
Then, we ran a series of benchmarks from ASU, incrementally increasing the test file from 4 to 16 GB. We found that sequential write performance remained steady, until we subjected the Vertex 4 to our 16 GB test file. At that point, sequential write performance dropped to 205 MB/s. The thing was, though the file itself was 16 GB-large, overall write operations resulted in 32.1 GB of test files on the drive after the benchmark had finished. Consequently, the 16 GB test file benchmark resulted in less than 50% available free space by the time it finishes, and that appears to be the tipping point for when performance begins tailing off.
After secure-erasing the drive and filling it to 60%, we repeated the test with a 16 GB file and saw write speeds drop to 104.28 MB/s. The Vertex 4 128 GB appears to need free space for optimal performance.