What I learned from HD Tune

I ran HD Test with the following drives:

WD5002AALX Black 500 GB 32 MB cache (comparatively heavy with four platters)
min: 1.3 MB/s; max 129.7 MB/s; ave: 78.9 MB/s; access time: 17.1 ms; burst rate: 230.4 MB/s

WD5002AALX Black 500 GB 32 MB cache (lighter than above with two platters)
min: 52.5 MB/s; max 111.4 MB/s; ave: 88.5 MB/s; access time: 13.2 ms; burst rate: 166.5 MB/s

WD5003AZEX Black 500 GB 64 MB cache (single platter, I think)
min: 90.3 MB/s; max 195.2 MB/s; ave: 154.3 MB/s; access time: 16.1 ms; burst rate: 171.5 MB/s

WD2500HHTZ Velociraptor 250 GB 64 MB cache (single platter, I think)
min: 129.1 MB/s; max 180.3 MB/s; ave: 158.8 MB/s; access time: 6.87 ms; burst rate: 188.0 MB/s

Crucial M4 64 GB SSD
min: 166.4 MB/s; max 207.9 MB/s; ave: 189.4 MB/s; access time: 0.078 ms; burst rate: 136.6 MB/s

Intel 530 120 GB SSD
min: 159.8 MB/s; max 200.1 MB/s; ave: 188.9 MB/s; access time: 0.053 ms; burst rate: 186.1 MB/s

The first drive was used for two years and showed its wear by frequent severe drops in transfer rate. I have another one and it shows the same pattern. Both of them become hot during use. The WD5003AZEX and the Velociraptor are fairly new. The Crucial SSD is a few years old and the Intel is less than one year old.

All of these drives pass SMART. And just to show how inconsistent this testing is, the third drive was tested again and its burst rate was only 145.5 MB/s. I think HD Tune was confused when it awarded the first drive such a high burst rate. And is the good access time for the second drive an artifact of HD Tune or real?

I would have predicted that the Velociraptor would best the WD5003AZEX in terms of transfer rate. I was expecting around a 33% improvement in performance overall. The Velociraptor is by far the best hard drive in terms of access time, but I thought that would be manifested in transfer rates as well. The advertised data transfer rate is 200 MB/s, but mine fell short.

I had to chuckle at Internet comments regarding how loud the Velociraptor allegedly is. I had to put my ear almost to the surface to hear it.

I was about to replace the second WD5002AALX with the WD5003AZEX or the Velociraptor as a system drive, but after seeing the access times I’m decidedly undecided. I probably will because the transfer rates on the second WD5002AALX are not comforting.

Feel free to make any comments. This is purely an educational exercise.

I reran drive #2 because the low access time bothered me. This time the figures were:

min: 64.6 MB/s; max 129.7 MB/s; ave: 102.7 MB/s; access time: 16.5 ms; burst rate: 180.3 MB/s

This time the access time is as expected. I have no idea why HD Tune gave its previous low figure.

And just because I had everything connected, I reran the heavy WD5002AALX drive, with its figures being:

min: 2.9 MB/s; max 111.2 MB/s; ave: 85.3 MB/s; access time: 13.0 ms; burst rate: 157.3 MB/s

The spurious burst rate is gone, but the two WD5002AALX drives conspired behind my back to trade the low access time!

And I ran a new drive, a WD3200AAKS, one which has only been lightly used:

min: 3.3 MB/s; max 111.5 MB/s; ave: 90.5 MB/s; access time: 16.2 ms; burst rate: 139.3 MB/s

It’s safe to say that there is an enormous amount of variability in these tests. I would have to run each drive at least five times and average the data to trust it. The access time does not appear to be influenced by the amount of cache a drive has. And there seems to be an improvement in quality since the original WD5002AALX drives were released.

Remember what you paid for all this.


Thank you for sharing this. I recommend you take a look a this link. It contains some useful information.

Benchmarking programs indicate that my hard drive is performing slower than expected

The last piece of advice in the referenced article – Update the BIOS of the computer or motherboard – is terrible advice. Intel recommends that BIOS only be modified if a specific problem is being fixed. There are plenty of cases where moving to a new BIOS actually causes a problem. Intel’s motherboards for Sandy Bridge and Ivy Bridge cannot be reverted back to Sandy-only, with there being outstanding bugs which have not been fixed (one preventing overclocking). One of these boards is sitting about three feet from me.