MyDigitalSSD mSATA 32gb DDR2 SuperCache Drive Review

MyDigitalSSD has come out with a new drive, dubbed the 32gb DDR2 SuperCache SSD, aimed primarily at users who wish to get SSD-like performance while maintaining the capacity of a traditional platter-based HDD. In theory, this sounds like a product that would cater to a wide variety of users (after all, the small size of SSDs compared to their price is off-putting for many buyers), but in practice is quite limited in scope for the moment due to lack of availability of one of the critical pieces to make it happen. More on this later.


SSD Caching:
Caching of any sort, be it to RAM, SSD, or any other media, follows the same basic principle - take the most frequently used data from a slower medium, place it on a faster one, and the law of averages says that most things you do will be at least marginally quicker because they'll read at least a portion of their data from the faster medium. Modern operating systems, particularly Windows 7, do this for you automatically, pre-loading and storing frequently-used parts of programs into RAM so that they can be accessed quicker. The problem is, of course, that while RAM is ridiculously fast as a storage medium, it loses its data when it's powered off, so the first load of any program after every reboot is going to be as slow as the drive you're loading it from.

Enter the SSD as a cache drive. SSDs, while not as fast as RAM, still outpace the fastest hard drives by a large margin, and do not lose their data when powered down. A small, inexpensive SSD stores the "hot" data which is accessed most often by the user, while the "cold" data is read from the hard drive. Typical examples of "cold" data are pictures, movies, and music, which are accessed infrequently and for which access speed is irrelevant, as well as programs (or parts of programs) that are infrequently or never used, and even parts of the operating system that you don't use. Don't worry - if you suddenly start using something more frequently, after the first couple loads, it will be fully cached to the SSD, perhaps overwriting something else you haven't used in months, and will be available for fast access. Think of it as an automatically-managed dual drive setup, where you're not forced to determine as an end-user what you want to be sped up and what will be slow at the time you install it. Additionally, cached data exists as a redundant copy on the SSD - nothing is ever deleted from the HDD, and even if you completely remove the SSD at a later date, the system will still function perfectly, albeit at mechanical HDD speeds.

SSD caching is capable of being achieved purely in software, and as such should be universally available regardless of platform - AMD or Intel, laptop or desktop. Indeed, at least one company has already demonstrated this - NVELO's Dataplex software, distributed with the OCZ Synapse line of SSDs, works on any hardware as long as it's running Windows 7. I have no doubts that versions for Max OS and Linux will come about, eventually. Intel's Smart Response Technology takes a different approach, integrating SSD caching as part of their Z68 and all socket 2011 chipsets. This option is currently only available on desktop systems.

Although software exists to make SSD caching a reality on pretty much any modern hardware, for some inexplicable reason it does not seem to be obtainable for consumers. The Dataplex software mentioned previously is not available for purchase from the company's website, and can only be obtained alongside a certain three-letter company's overpriced "caching solutions." A few other companies also make SSD caching software, but only distribute it to businesses for server usage. Intel's SRT is available at a consumer level but requires the purchase of a new motherboard, and most likely CPU as well, to take advantage, and again, cannot be used in laptops. If none of these companies will remove their proverbial heads from the sand and market SSD caching software directly to end-users for use with hardware they already own, I have no doubts that the open-source community will beat them to the punch at some point.


The Drive:

The 32gb DDR2 SuperCache drive comes in standard 50mm mSATA configuration, and features a SATA2 interface, Phison controller, and 32gb of Toshiba asynchronous toggle-mode DDR2 NAND. According to the product specifications, this provides improved performance and reduced energy consumption. I have no way to verify this second claim, as I performed my tests using a desktop, but as to the former, the drive is rated for 270/100 MB/s read and write, respectively. These are decent numbers, considering the drive's low capacity, which tends to hinder write speeds particularly.

The drive is advertised as useful for either a cache or OS drive, particularly in notebooks that don't support a second hard drive except via mSATA connector. As a cache drive, it's really not very useful at present due to the lack of software availability, although some Z68 boards are being released with mSATA slots to allow it to be used this way in a desktop (this is how I tested caching performance). As an OS drive, you can just plug it in, install Windows, and go, but know you'll be limited on space. With only Windows 7 Professional, all necessary drivers, and my benchmarking software installed, the drive only had 13gb of free space left. This is with my page file manually shrunk to 2gb, and with hibernation enabled, which reserved 8gb of space to match my RAM capacity. Although this doesn't leave much space for additional programs, media, or games, you are running this in tandem with a secondary mechanical drive, right? Right? Alternately, you could disable hibernate to free up some additional space, but I don't recommend doing this with a laptop for that freak occurrence when you actually do run out of battery.


Benchmarks:
I put the MyDigitalSSD through its paces, both on its own and when used as a cache for a traditional hard drive. I have also included benchmarks of a 120gb Intel 320 SSD and a 1tb Samsung HDD for comparison.

The following system was used for all tests:
Motherboard: Gigabyte GA-Z68AP-D3
CPU: Intel Core i5-2500k @ 4.0 ghz
RAM: 8gb DDR3-1333
GPU: ATI Radeon 6950
Drives: MyDigitalSSD 32gb DDR2 SuperCache SSD, Intel 320 series 120gb SSD, Samsung 1tb 5900 RPM HDD


Synthetics:
I ran CrystalDiskMark and ATTO on all three drives in freshly-formatted contition. The OS was not stored on or run from the drives while they were tested.

MyDigitalSSD:


These results are not bad for a 32gb drive. The ATTO benchmark shows unrealistically high sequential write scores, for which I have no explanation at the moment.


Intel 320:


The Intel SSD beats the mSATA drive in all relevant benchmark areas, but it's also much more expensive. This was to be expected.


Samsung 1tb:


The slowest of the bunch. Just look at those awful random read/write scores, which happens to be the characteristic most indicative of real-world use. Of course, we wouldn't be looking to boost this drive's performance if it wasn't the worst.

These benchmarks really just serve to confirm what we already know - the more expensive, larger SSD will outperform the cheaper, smaller SSD, and both will completely destroy a bargain-basement mechanical drive. The MyDigitalSSD suffers a bit in the write department, particularly sequential ones, but it doesn't really matter. Few real-world tasks are sequential to begin with, if you use it as an OS drive you won't be doing many writes, and if you use it as a cache drive it will only ever write as fast as your HDD can read.


Real-world Tests:
This is what most people care about. You can look at benchmarks until you're blue in the face, but they don't really indicate true performance unless you sit around using your computer to run benchmarks all day. And who does that?

Except for the Windows installation test, all tests were performed with the OS and drivers installed on the drive while I ran them.

Windows Installation:
I installed Windows 7 Professional x64 from a USB flash drive on all three drives. Time was measured during the "automated" portion of the installation, from when it began to copy files until it asked me for my user name. No user input was required during this time, and the time taken for all restarts is included in the results, as it should be the same across all three drives.

The number in parentheses is the amount of time taken for the "Copying/Expanding Windows Files" portion of the install to complete.

mSATA: 8:56 (5:19)
Intel: 9:00 (5:26)
HDD: 11:40 (5:34)

The SSDs are evenly matched here. The portion of the install where files were actually copied to the drive is roughly the same across all three tests, probably limited by the speed of my flash drive. The HDD really slowed down during the "Completing Installation" phase and after the final restart when Windows began to configure itself. Although it was not measured as part of this test, the "Preparing Your Desktop" screen lasted mere seconds on both SSDs but a good minute or so with the HDD.

All in all, this test really shows that installing a modern OS on modern hardware is nowhere near the pain it once was. Fifteen minutes and a few clicks is all it should take on pretty much any drive. Now, finding and installing all your drivers, and tweaking everything just the way you want it, is a different story entirely...

File Copy:
I copied a 3.8gb Blu-Ray encode from the root of each drive to itself.

Intel: 45.5 sec
mSATA: 1:11 (full), 50.4 sec (open)
HDD: 1:26

Under normal OS conditions, the Intel drive won this test handily. The MyDigitalSSD exhibited performance closer to the HDD than the other SSD while the drive was full. I re-ran the test while the drive was completely empty, and performance improved by over 20 seconds. The small size of the mSATA SSD did it in during this test, as with Windows 7, OpenOffice, Vegas Pro 11, and a 3.8gb movie on it, there was a hair over 9gb of free space, even less as I created a second copy of the file. When an SSD is mostly full, performance suffers.

7-Zip
I archived my entire My Documents folder, copied it to each drive, and then unzipped it to the same drive.

Intel: 18 sec
mSATA: 19 sec
HDD: 21 sec

There was a difference of 3 seconds for an archive 1.2gb in size when uncompressed. Larger archives might differ by a few more seconds, but performance was more or less even here.


Cache Performance:
The following tests involve the MyDigitalSSD in its role as a cache drive, mated with the Samsung 1b HDD using Intel Smart Response Technology. I ran each test a total of 3 times on the cached volume, to account for everything being recognized and loaded onto the SSD.

A quick word on Smart Response Technology, it's ridiculously easy to enable, even for someone with limited knowledge. It took me a total of 3 steps to get it to work.

1. Click the "Accelerate" tab in the Rapid Storage drivers and enable acceleration.


2. Choose your drives and settings. If you have one SSD and one HDD, the defaults will work fine.


3. Click OK and you're done!


Windows Boot:
For this test, I took a clean Windows 7 installation, including only the necessary drivers, and cloned it across all three test drives with Acronis 2011. Alignment was checked on the SSDs to ensure it was correct, and boot time was measured using Boot Racer. I continued rebooting until the times I got were consistent within a second or two across multiple boots.

Intel: 14 sec
mSATA: 15 sec
HDD: 28 sec
HDD w/cache: 37 sec, 18 sec, 15 sec


OpenOffice Writer:
I disabled the OpenOffice speed launcher and measured how long it took to load.

Intel: 2.4 sec
mSATA: 2.6 sec
HDD: 5.9 sec
HDD w/cache: 7.7 sec, 2.8 sec, 2.7 sec



Sony Vegas Pro 11:
mSATA: 4.4 sec
Intel: 4.9 sec
HDD: 14.4 sec
HDD w/cache: 14.6 sec, 5.5 sec, 5.0 sec



As long as it supports TRIM, an SSD is an SSD is an SSD. In the real-world tests, the differences between the two SSDs amount to what could be considered a rounding error, and the differences apparent in the benchmarks lose their significance. When SSD caching is enabled, the first launch of a program seems to take a bit longer than it otherwise would, probably while the drivers copy it over to the SSD, but the second is much faster, and by the third launch, speeds are on par with an SSD alone. Once stand-alone caching software becomes available, I suspect these results will be repeatable on laptops as well.

 

Thank you for the great in depth review Syberia.

By the time this product is in MP (15 days from today) I hope to have a good caching software to bundle it with.

I would also like to point out something that people might not know about SSD as the GB size goes down the speed also goes down dramatically. In fact most 32GB SSDs would not even be considered for a primary boot drive. This SSD is so very different because of the type and quality of flash that it uses that it bucks the trend and is on par with a today's faster higher GB Sized and higher priced rivals.

MyDigitalSSD

 

Man, I was tempted to get this thing and try it out that way, that or the Intel 311. Good to see it does exactly what it is supposed to.

 

Thank you for the review Syberia. I've been searching for a review of this Super Cache SSD.