Initial Performance Testing
This is the perfect opportunity to detail my SSD testing systems. At the heart of each system is an Asrock Z270 Taichi motherboard fitted with an Intel 7700K running with each core locked at 4.5GHz. Each system uses two Crucial Ballistix DDR4 memory modules that offer exceptional compatibility with all of my test systems regardless of the chipset.
If you take our advice and run your new high capacity SSD as a secondary volume for cold file storage the data likely changes. Secondary files are usually based on large block sizes and sequential in nature. This includes your pictures, music, and movies. It is possible to load your games and other applications to secondary storage.
You simply install the software to drive D: or whatever the new drive letter is. I go as far as making the Program Files folder (D:Program Files (x86) and D:Program Files) to make it a little easier to keep the data organized. It's very easy to let the data on a 2TB drive to get out of control. Organization and a little planning is the key to success here, or else you will just start throwing data on the drive, and it can take hours to get it all sorted back out. Don't even ask me about the condition of my NAS right now.
We combined our five top picks for high capacity SSD storage in the same charts. It's rare to see legacy SATA in the same charts with NVMe, and you will see why soon. There are no losers here today in this article. Every drive we chose has an advantage somewhere, be it random performance, large block-size write performance, an exceptional value or enough endurance to last a lifetime.
Sequential Read Performance
We need to preface the sequential read test results with a disclaimer. The MyDigitalSSD BPX Pro and all E12-based drives we've tested have an issue with our script that shows reduced sequential reads at queue depth 2 and 4. We think the issue is just a simple case of timing, when these two tests come in our script, the amount of delay between the tests, or in simple terms, bad luck. Phison is looking into the issue.
The two NVMe SSDs deliver towering sequential read performance compared to the three SATA SSDs. They also cost significantly more but as a secondary volume for large block size data, it's impossible to beat a proper 2TB NVMe SSD.
Sequential Write Performance
Again, we see the NVMe SSDs outperform the SATA SSDs. Actually using the increased performance is another issue entirely. Since this is sequential data it likely comes from another drive or from across the network. If the data comes from a SATA SSD in the same system then you will likely see data writes between 300 and 500 MB/s even with a NVMe SSD as the target drive. Over a network, you will need a 10-gigabit connection just to write roughly 1000 MB/s to the NVMe SSD, and that's assuming your NAS or other device sending the data can sustain bursts that high. To fully utilize the NVMe SSDs you need another NVMe SSD as the source of the data transfer.
Sustained Sequential Write Performance
We have three-bit per cell flash media to thank for low-cost large-capacity SSDs available today. This media does have a drawback compared to older two-bit per cell (TLC) media when it comes to sustained writes. The drives use a buffer to write at high speeds but in many cases, the buffer ends, and the data transfer rate slows. Some drives continue to pass the data through the buffer, but that data has to slow so it can flush to the TLC area. Other drives will hold the data in the buffer and then start to write directly to the TLC area. Regardless of the method, there is usually a limited space for maximum performance sequential data writes to the drives.
Random Read Performance
If you plan to use your new high-capacity SSD for primary storage with your operating system and cold storage files on the same drive, then the random read performance charts are more important than the sequential results.
You will quickly notice the fastest drive at queue depth 1 random reads is actually the cheapest SSD on the charts today. The Crucial MX500 uses Micron 64-layer memory that is exceptionally quick. We often hear from readers that NVMe SSDs don't feel any faster than newer SATA SSDs. The random read performance is why. This is the area that you notice the latency between clicking an icon and seeing the requested action start or complete. We will look at this more on the next page where we start ripping through applications.
Random Write Performance
Random write performance is still importable but the SLC buffer speeds make these actions complete so quickly that we've reached a point where the other components in the system, and the underlying operating system / file system, don't allow normal applications to take advantage of the increased performance.
70% Read Sequential Performance
NVMe has a great feature that we often overlook in our reviews. The PCIe bus and NVMe protocol allow the SSD to read and write data at the same time. The SATA bus doesn't allow the other drives to do this. The advantage comes when multitasking with several programs running at the same time.
70% Read Random Performance
The real difference comes with random data reading and writing at the same time. NVMe SSDs are faster but you have to put a little work into your system to benefit from the technology. Operating system buffers do a good job of masking SATA's limited abilities to the point where the two technologies are very similar in daily use activities.
Last updated: Sep 24, 2019 at 12:27 am CDT
PRICING: You can find products similar to this one for sale below.
United States: Find other tech and computer products like this over at Amazon.com
United Kingdom: Find other tech and computer products like this over at Amazon.co.uk
Australia: Find other tech and computer products like this over at Amazon.com.au
Canada: Find other tech and computer products like this over at Amazon.ca
Deutschland: Finde andere Technik- und Computerprodukte wie dieses auf Amazon.de
- Page 1 [Introduction, Specifications, and Pricing]
- Page 2 [Initial Performance Testing]
- Page 3 [Class Real-World Performance Testing]
- Page 4 [Final Thoughts]