256 MB working set and a single pass. Enough to spot a failing drive or confirm that a new install is on the right disk.
- 256 MB file
- 1 pass per test
- Cache warmup skipped
A benchmark is only useful if you can repeat it. DiskSpeedTest fixes the block size, queue depth and cache handling per profile, so two runs on the same drive are directly comparable.
256 MB working set and a single pass. Enough to spot a failing drive or confirm that a new install is on the right disk.
The default. A 1 GB working set, warm cache discarded, three passes averaged. This is the profile behind the reference figures on the front page.
A sustained write that keeps going until throughput settles. Useful for spotting drives that throttle once the SLC cache fills.
The app creates a temporary folder on the chosen drive and checks free space for the selected working set. Nothing is written outside that folder.
A short pass fills the OS cache, then the app asks the system to flush it. Both warm and cold numbers are kept, because the difference is itself a result.
Sequential read, sequential write, random 4K read and access latency, in that order, with the configured block size and queue depth.
The temporary file is deleted and the folder removed. If the app is interrupted, the next launch offers to remove anything left behind.
The result lands in local history with a timestamp, drive model and profile. Export to CSV or a text report, or open the results page for the interpretation guide.
DiskSpeedTest fixes everything it is allowed to fix. The rest is the machine, and the results page is blunt about it.
Download the app and pick the standard profile. It takes eighty seconds and leaves nothing behind.