Understanding Benchmark Results
The result compares the read behavior of the two test files on the selected drive.
Verdict and comparison
The throughput comparison uses the best read pass for each file. The factor indicates how much slower the fragmented case was in that run. The latency chart compares the time required for read requests, including the median and higher percentiles.
The benchmark reports actual fragments as well as the number requested. If it cannot produce a meaningful fragmented test file, it reports an inconclusive measurement. The requested fragment count alone does not prove that the intended layout was achieved.
Inspect the detail sections
| Section | Useful information |
|---|---|
| All values | Extents, throughput, IOPS, latency, CPU time, device reads and split requests. |
| Measurement setup | Drive, media type, file system, cluster size, file size, block size and duration. |
| Individual passes | Values for every read pass, useful for spotting variation. |
Higher throughput means more data read per second. Lower latency means less time per request. The 95th and 99th percentiles describe the slower end of the request distribution rather than the average request.
Some device-level values may be unavailable and appear as n/a. Treat that as missing measurement data, not zero activity.
Limits of the comparison
The result applies to these test files, parameters and system conditions. It is not a universal prediction for every workload or a drive-health test. If background activity or insufficient free space interferes, address that condition and repeat with the same setup before comparing results.