After receiving a number of requests for more real-world testing, we're moving in that direction. Our trace-based benchmarking gives you a holistic performance picture. And the more specific random/sequential read/write measurements drill down into more targeted workload profiles.
However, we still haven't answered our original question in a way that makes the decision to use your new SSD on an old platform or buy something new any easier. Does our testing indicate the need for 6 Gb/s connectivity or not?
We know that most of our real-world tests see queue depths of one, and involve a mixture of compressible and incompressible data. In this particular test, we're transferring 11 GiB worth of video clips (which, of course, you can't really smash down any more than the H.264 compression standard), along with a slew of smaller files that are compressible.
Here's where things get interesting. Because of this workload's profile, there's frankly very little improvement to realize in jumping from SATA 3Gb/s to SATA 6Gb/s. Crucial's m4, for example, performs exactly the same at both data rates.
Yes, the Samsung and OCZ drives are able to present quantifiable gains. However, the real story here is a comparison to even the slowest SSD, Intel's SSD 320, and Western Digital's Scorpio Blue, a drive you might have installed in a notebook. The reality of the situation is that this battle isn't between the fastest and slowest SSDs, it's a matter of comparing hard drives to solid-state storage.
Backing up a game using Steam incorporates a mix of incompressible and compressible sequential writes, along with a sizable number of random write operations. This task also involves a good deal of host processing as individual files are packed into archives. The result is that there's not much difference between SATA 3Gb/s and 6Gb/s, or even between the SSDs and hard drive, since storage surprisingly isn't a big bottleneck.
Of course, we know that storage performance is only one determinant of benchmark results. Remember our office productivity analysis? In nearly 30 minutes of antivirus scanning, our SSD was only busy for 281 seconds. In essence, the task wasn't taxing enough to demonstrate an SSD's benefit.
In order to push the test harder, you have to add concurrent operations. For example, we can still transfer files to and from the SSDs while the backup operation executes with minimal penalty. On a hard drive, however, the same multi-tasked workload slows both actions to a crawl.
Measuring boot time is one of the best examples of how an SSD excels. You get a mix of random and sequential reads, along with some write operations attributable to logging. Queue depths during Windows boot can easily exceed four, as the operating system accesses multiple files in quick succession or at the same time.
The differences between the SSDs are again very minor, while the hard drive drags. Don't expect SATA 6Gb/s to buy you any additional speed in this metric, though.
- Should You Feel Bad That Your Board Only Supports SATA 3Gb/s?
- Test Setup And Benchmarks
- Benchmark Results: Storage Bench v1.0 And PCMark 7
- Benchmark Results: 4 KB Random Performance
- Benchmark Results: 128 KB Sequential Performance
- SandForce: Performance With Incompressible Data
- Real-World Tests
- Buy The SSD You Can Afford, Not The Fastest One
- Storage Bench v1.0, In More Detail
- More Background On Our Benchmarks