A look at Solid State Drives (SSD) v. Hard Drives (HD)

Mar 01 2010
Like rocket cars, moon vacations and watchable online video we all know that solid-state drives (SSD) are the future.  Of course it will be a long time before any of us can afford to replace that 2TB drive with one, but with prices for smaller SSD finally approaching a reasonable level it makes sense to start thinking about how to fit one into the HTPC (especially a client).

Heat is the biggest concern for any PC because if we do not dissipate it correctly with active cooling (which generates noise) the stability and longevity is negatively affected.  So we will be looking mainly at power usage and the external temperature of the drive, and then discuss drive noise and some performance observations.

To run the test, Windows 7 was installed on three drives into the same PC (one at a time) and an 8GB file was copied and pasted in the same folder to generate load. 

Drives:

  • Intel X25-M G1 SSD
  • Toshiba HDD2D35 - 5200RPM 2.5”
  • Seagate 7200.10 - 7200RPM 3.5”
ssdpowerusage.jpg

At idle we can see that the SSD uses 1W less than 2.5” and 7W less than the 3.5”.  Under load, the gap narrows to 4W less versus the 2.5”, but significantly widens to 14W against the 3.5”.  These results are not at all surprising, although I was impressed with the efficiency of the 2.5” drive.  The temperature results were quite interesting however, and not in line with my expectations.

To measure drive temperature I used an infrared thermometer while the file was copying to take readings on the surface of each drive.  Note that for the test each drive was placed outside the case (mounted in the drive cage and placed on a thin piece of foam to isolate vibration) so temperatures when installed inside the case should be slightly higher.  To ensure valid idle measurements the PC was allowed to run for 20 minutes in an idle state; and for references room temperature at the time was 68 ° Fahrenheit.  

ssddrivetemp.jpg

As we can see the SSD clearly has the lowest temperature of any device at both idle (69.3°) and load (70°); where the 2.5” drive has the highest (86°/102°).  I found this quite odd given the power numbers and the intended use cases (the 2.5” was borrowed from a laptop) for each drive.  My guess is that the 3.5” drive’s larger surface area allowed it to dissipate more heat maintain a lower temperature, it would be interesting to put each drive in an enclosed environment and rerun the test. 

Unfortunately I do not have a sound meter so I cannot measure precisely how loud each drive was during the test.  That said, it’s easy to guess which made the most noise (3.5”), and which made none (SSD).  What surprised me the most (and perhaps this was because the drives were outside the case at the time) was the amount of noise coming from the 2.5” drive.  Both mechanical drives were clearly audible throughout at a reasonable listening distance, but it was possible to not hear the 2.5” by moving far enough away in the room (about 15’). While unquantifiable, there is something truly magical about a completely silent HTPC :)

SSD performance has been covered many times, so I didn’t dig into it too much, but one surprising result from the test was the drive’s poor write performance.  Oddly the SSD was only slightly faster than the 2.5” drive over the duration of the file copy.  It started out great, but about half way through performance dropped significantly.  I’m guessing that because of the lack of TRIM support for G1 SSD and the age of the drive the writes past a certain point were dirty and throughput suffered accordingly.  This is unfortunate because one of the cases I was planning to examine was using the SSD as a recording drive (for the excellent read I/O performance) and automatically moving the files after commercial scanning to a large remote disc.  This scenario isn’t that practical given the size of the SSD (80GB), but when the slightly larger drives become more competitive it’s certainly something I’d like to look at again.

Website design by Yammm Software
Powered by Drupal