This drive is a legendary one in my view, as this is my longest serving drive. That was not my expectation when I bought a pair of them back in 2010. Its “brother” drive had died back in 2018, leaving this one to carry on. It was also famous to be affected by a firmware bug which could cause data corruption if certain commands were executed during writes. Nevertheless, it was a very good drive, even if the F4 series was their “green” drives.
The Drive
This drive has the standard “boxy” look of Samsung HDxxxXX drives. This one was purchased from a local computer shop and was manufactured back in October 2010. I have applied the required firmware update to this drive early on, although it seems the firmware version number was unchanged by the manufacturer which is silly. This drive has a 32MB cache, quite a luxury for a “green” style drive.
This drive seems to follow a similar formula to other Samsung drives – Marvell controller, TI spindle control and Samsung cache RAM. This one seems to carry a code of S3M.
A label covers the side access hole for the servowriter.
Again, a datamatrix barcode is located on the opposite face to the SATA connector, but it does not seem to be the serial number.
Performance Testing
CrystalDiskInfo
The drive currently carries the record power-on-hours of my fleet at 99417 hours. With so many hours on the clock, I don’t expect perfect health although the attributes don’t suggest anything critical. The drive claims a rotation rate of 5400RPM.
HD Tune
I suspect the drive has aged, with a very lumpy read graph. Perhaps it is using an adaptive density format per-head, but I doubt this. The drive peaked at 144.1MB/s, ended at 66MB/s and averaged 112.3MB/s with an access time of 16.1ms.
Writes showed similar lumpiness, peaking at 144.5MB/s, ending at 64.5MB/s and averaging 111.5MB/s with an access time of 13.3ms.
Peak IOPS reached 63 for reads and 109 for writes. This is pretty normal.
Cache buffer effects on reads and writes are very clear, although both do not show the full 32MB being effective.
File block accesses reach peak performance by 16kB accesses.
CrystalDiskMark
CrystalDiskMark shows a sequential performance of 147.7MB/s read and 145.5MB/s write. Medium-block accesses reached 41.87MB/s and 63.40MB/s write. Queued small-blocks reached 0.968MB/s read and 1.156MB/s write. The latter seems quite average, which is not unexpected for a “green” type drive.
ATTO
Performance seems to reach its peak by 16kB, although 8kB gets quite close. Consistency in I/O is a little lumpy, but at least it doesn’t exhibit the inverted performance curve that the 500GB F1 did.