It would be swell to test in a multi-drive scenario, though we only have one drive on the test bench for the moment. Regardless, we have quite a few tests that’ll focus on both sequential and random performance. Our large transfer test involves moving a massive 33.8GB Steam game folder that has a variety of compressed and uncompressed files. It’s a particularly difficult test due to the largely random nature of the data being written. There’s all sorts of sizes and types of files that it’s a valid challenge that should show the real-world transfer speeds to a drive being used as your storage drive.
For these performance tests we hook it up to the Intel 6G SATA controller for consistencies sake.
CrystalDiskMark shows it doing fairly well in sequential tasks though it seems to suffer during random 4K size writes compared to the only HDD we have on hand to compare against. Overall the peak performance is quite good considering the amount of platters and 7200RPM spindle speed. It ends up being quite a surprise. During normal operations, transfering photos, video assets and generally using it as a working drive there’s very little response time delay and transfer speeds tend to be around 40-210MB/s depending on the task. This is meant to be a multi-user drive, so even larger QD’s illicit good performance, just slightly slower than seen above. The firmware is capable of carrying on separate transfers at near the peak that the SATA 6G interface can handle. It’s good news, really.
For this test we chose to use IOMeter due to the fine grain control you can have over the entire test. It ends up being a most impressive and very complex test. We set this to the most difficult QD that you might encounter to show a worst case scenario. And here the IOPS, though much lower than a PCIe-based SSD, are perfectly in line with expectations. Higher QD’s have slightly better IOPS performance, showing how strong of a drive this really is. The head noise is there, and persistent, though not completely annoying.
Next we’ll measure response time, or the time it takes for a request to do something on the disk is acknowledged. For this we use HD Tune Pro again as the tool of choice to see just how long it takes to actually begin a task. What we see is that on average, it’s very responsive. Though there are times where there is a barely perceptible delay. It’s important to note that this shouldn’t necessarily be felt by the user, though it does effect the user experience negatively. In real use it wasn’t noticeable compared to anything except for an SSD of any type. It was, however, persistently loud.
Massive Transfer Test
The next test is a homemade test where a massive mixed-file folder that’s 33.8GB in size is transferred from one to the other. I’ve used HD Tune Pro to measure the results, including the time it takes to transfer the folder, the average IOPS during the transfer and, of course, the transfer speed itself. This is a write-only test. It’s a static Steam game folder that’s been placed elsewhere precisely for this test. It isn’t updated, so it can be used as a point of comparison in the future.