Flash SSD Update: More Results, Answers

Conclusion

There is one conclusion we would like to make right away: OCZ’s new SATA II 2.5“ SSD, which is based on a Samsung drive, is the best Flash SSD in our small benchmark lineup. The OCZ Flash SSD provides performance that beats even the quick Mtron drive in many sections, and its power efficiency is unmatched. This is the type of result a Flash SSD should provide, and such a drive wasn’t available at the time we analyzed the other Flash SSDs.

A look at the rest of the test bed reveals two results, one of which supports the conclusion we made in the initial article: mainly, Flash SSDs are not there yet and some may even contribute to emptying your notebook’s battery more quickly. This applies to scenarios in which the requirements are limited, e.g. DVD video playback, or when the drive idles for a long time. The other result makes clear that Flash SSDs, even some poor ones, are more efficient as they have to do work. This applies to the benchmarks testing random I/O and sequential reads. As always, the final picture is somewhere in between, and a decision for or against a Flash SSD can only be made for a particular product, and having analyzed the intended operating conditions and applications.

Flash SSDs Win Under Load

Most of the Flash SSDs tested provide better random I/O performance and better sequential throughput than mechanical hard drives. Performing random I/O, the Flash SSDs show a similar power requirement than the 2.5“ hard drives, while these require more power for sequential read operations. Since sequential read throughput oftentimes is higher on a Flash SSD, the performance-per-watt ratio is better for the Flash SSDs, as they deliver better performance while requiring less power. This also applies to random I/O, but with smaller advantages for the Flash SSDs. So much for the section where SSDs win.

Hard Drives Efficient In Idle And Low Load

The two other benchmarks reveal advantages for conventional hard drives when they remain idle, or when they have to play a constant data stream, e.g. DVD video playback from the drive. Many of you requested the video playback test, which is dominated by OCZ’s new Flash SSD. Clearly, for future SSDs, the hard drive is beaten. But the Hitachi Travelstar 7K200 drive manages to best most of the other drives by requiring less power for DVD video playback thanks to a low-power idle mode that throttles the drive, while maintaining the performance it has to deliver (no, it does not spin down). If it weren’t for the slow (but efficient) SanDisk SSD and the new winner by OCZ, this section would be dominated by a mechanical drive.

The results are almost alike for our drive idle test, for which we measured drive idle power after an idle period of 10 minutes, which allows all drives to enter their power saving states if there are any. All Windows power saving features for the hard drive were disabled to watch what the drives do if they are bored. Again, if it weren’t for the new SSD from OCZ, many Flash SSDs would have to surrender to old fashioned hard drives, as their idle power is simply too high.

Consequences

As already pointed out, the power disadvantages originate from old Flash SSD designs using inefficient SATA bridges, and even though the Flash SSD Hoax article was saddled by a mistake in the battery runtime test procedure, for which we apologize, the conclusions were not far off.

We recommend against purchasing any Flash SSD without knowing details about performance and power requirements. Flash SSDs do not inherently contribute to increasing battery life and better efficiency comes with the appropriate Flash SSD used for a specific application. "Flash SSD" is not a qualifier for efficiency or performance.

Huge differences do not only exist between various Flash SSDs, but also between mechanical hard drives. We learned that a 7,200-RPM drive can be much more efficient under specific applications, while a 5,400-RPM hard drive may very well outperform the theoretically faster 7,200-RPM drive in certain benchmarks. At the same time, power requirements and efficiency highly depend on the particular model, so again, we recommend paying close attention to relevant test results before you spend your money.

Another consequence of the varying benchmark results has to be an adjustment of the testing methodology used to evaluate hard drives and solid state drives. While reviewers have been looking at performance from various angles — mostly tracking power requirements in idle and peak power — it will be increasingly important to analyze storage device efficiency based on popular applications scenarios. Expect to find this information on all upcoming reviews of hard drives or solid state drives.