The Western Digital VelociRaptor has ruled the roost when it comes to high performance hard drives, being the only relatively modern consumer option for 10,000 rpm drives. Descendants of the earlier Western Digital Raptor drives which I covered earlier, I had not been able to witness the performance of the VelociRaptor myself. Coming across one in a junk pile in Hong Kong, an HP special WD1600HLFS dated 18th September 2008 complete with its cooling heatsink, that unit proved to be non-functional.
It was my luck that in the recent trip to the CCARC Wyong Field Day, I managed to pick one up for AU$5 to try my luck for a second time. Since then, it has been sitting waiting for me to take the time to give it a spin.
The Drive
Image may be NSFW.
Clik here to view.
This particular drive is yet another HP special, this time, missing its IcePack cooling frame. The label design has changed from the old design, now featuring a “golden” stripe at the top and the label of “Enterprise Storage”. Accordingly, these fourth generation drives apparently have firmware bugs, so I don’t know how true that Enterprise Storage banner is, especially as this is a plain old ordinary SATA drive rather than a SAS drive.
This drive is a little newer than the last one I got my hands on – 19th September 2010 making it two years and one day younger than the first example I put my hands on. This one claims to be Made in Malaysia.
Image may be NSFW.
Clik here to view.
The underside of the drive is as expected, in true WD style, all components are placed on the other side of the PCB leaving only test pads on this side.
Image may be NSFW.
Clik here to view.
Image may be NSFW.
Clik here to view.
The drive appears to be 12.7mm height rather than the more common 9.5mm found in standard sized laptop bays. However, this drive won’t work even in those bays due to the requirement for a 12V rail.
Image may be NSFW.
Clik here to view.
Image may be NSFW.
Clik here to view.
The sides have nothing special to mention either.
Image may be NSFW.
Clik here to view.
Removing the PCB from the drive reveals a bit of foam and mylar between the board’s electronics and the chassis.
Image may be NSFW.
Clik here to view.
Not much to say about the electronics on the board – the main controller is labelled WDC WD70SC56-G and is a native SATA II controller. There is a 16MiB Hynix DRAM chip used as a cache. A small EEPROM is located at U12, and the motor control is performed by the ST Microelectronics “SMOOTH” chip at the bottom. There also appear to be three rotational velocity/vibration piezoelectric sensors – one at each opposing corner and one near the middle, likely to optimise head tracking performance in high vibration rack environments.
Performance
Does the drive work? In a word, yes. It does! The drive was tested on my old desktop platform, using a Phenom II x6 1090T overclocked to 3.9GHz on a Gigabyte 890FXA-UD7 with 16GB of RAM and Windows 7.
Image may be NSFW.
Clik here to view.
The drive even appears to be relatively young, clocking 11,146 power on hours and seemingly having no reallocated sectors. Unfortunately, the drive does seem to “buzz” and “click” a lot while idle, at least initially, suggesting there is perhaps some surface damage on the drive that was being “covered up” by background recovery routines.
Image may be NSFW.
Clik here to view.
The read throughput graph is shown above, however, this is the second time I had to run the benchmark. The first time, the drive had a dip right in the middle along with a lot more buzzing and clicking. No reallocations occurred, and I waited until the drive went quiet before trying again. I suspect it was perhaps some pending sector that didn’t show up but does suggest the drive may be on its way out.
The graph shows a maximum sequential read of 118.9MB/s, average of 102MB/s and minimum of 82.9MB/s. This is considerably better than the second generation Raptor WD1500ADFD which I tested that managed an average of 74.2MB/s. That is the result of increased recording density and improved electronics. Access times were also a blistering 6.96ms compared to the 8.14ms recorded on the ADFD.
Image may be NSFW.
Clik here to view.
The write curve showed similar performance, although with a number of dips which probably suggests that the media is not doing too well.
Image may be NSFW.
Clik here to view.
Image may be NSFW.
Clik here to view.
Extra tests were performed, which show the read cache having a decent effect, but the write cache being a little inconsistent. Compared to SSD performance, the drive still lags far behind – a prisoner of its mechanical nature.
Image may be NSFW.
Clik here to view.
Testing of random access IOPS was performed for read and write, but for some reason the read graph did not save correctly. The write graph seems to illustrate that even at 512 bytes, the peak IOPS recorded was just 326, far short of what SSDs can offer today.
Image may be NSFW.
Clik here to view.
With threading, the IOPS does improve, in read it posted 566, which is quite amazing for a hard drive. In doing so, it definitely sounded rather angry, with loud buzzing seeks.
Image may be NSFW.
Clik here to view.
The drive was also tested with CrystalDiskMark 7, showing the above results.
Image may be NSFW.
Clik here to view.
Testing with ATTO seemed to show that full performance was reached by 16kB accesses, with 8kB being quite close to full performance. This was definitely a monster of a mechanical drive, although it is outclassed by SSDs.
Conclusion
At last, I have been able to witness the performance of a Western Digital VelociRaptor drive for myself and the figures are quite amazing for a hard drive. Seeks are loud, but fast and IOPS figures are impressive for a hard drive. Sequential throughput does lag behind the denser, larger 3.5″ cousins of today, but that is not surprising due to the smaller platters in use. Even then, such mechanical technology is constrained by its nature, making the transition to SSDs for boot drives rather unsurprising if performance is concerned.
Aside – A Bit of Quiet for Now
It’s been quiet around the blog this week, but that’s not because I’ve been relaxing. Not at all. In fact, I’ve been very busy pulling together all the content and writing up the text for another major review which I have to deliver in about a week and a half. Other than that, I’ve also had the idea to give a few things a try, some of which didn’t work out and others which are work in progress, perhaps to be reported on once it is complete. Other things are waiting for bits and pieces, which are taking their time to arrive thanks to the new normal caused by COVID-19.
But for now, and the upcoming few weeks, it will probably be a little quiet around here as I put in the final bit of effort and deliver another major review. Hope you’ll stick around for it when it is published.