Sequential Performance Versus Transfer Size
All of our sequential transfer tests in Iometer are run at a queue depth of one in order to represent the most typical desktop workloads. It's not hard to imagine a situation where an enthusiast might subject an SSD to higher queue depths, though. And indeed, that's where an SSD really stands apart from a conventional hard drive.
Here we're using ATTO to test the sequential reads/writes over 2 GB, starting with a queue depth of two. Why only a queue depth of two? Even when you're pushing more taxing workloads, I/O commands are handled so much faster on an SSD that queue depths higher than two or three are far less common on an average desktop.
The other reason to use ATTO is its ability to easily test different transfer sizes. While 128 KB is fairly standard for measuring sequential performance, larger or smaller transfers are of course still relevant.
Using a queue depth of two, we finally see the 256 GB 830 edge ahead of the 250 GB SSD 510, which is known for its sequential alacrity. There is, however, a margin of only 10 MB/s in Samsung's favor.
When we turn to sequential writes, we see the 830 struggle a bit compared to some of the other drives. Second-gen SandForce drives, in particular, pick up speed versus our tests with a queue depth of one. The Samsung 830 follows closely behind though, and it finished first amongst the SSDs not based on SandForce's technology.
The 256 GB Samsung 830 continues to dominate read performance even after we turn up the queue depth to four. But again, Samsung doesn't have a big advantage. Overall, we're still looking at performance similar to Intel's SSD 510. The difference is that the 830 runs 10-20 MB/s faster for transfer sizes between 32 KB and 256 KB.
Samsung's 830 is the odd ball in this test. Most drives tend to have better performance when we turn up queue depths. However, the 256 GB 830 actually drops as much as 100 MB/s when we push larger transfer sizes.
