TSSDR TEST BENCH AND PROTOCOL
SSD testing at TSSDR differs slightly, depending on whether we are looking at consumer or enterprise SSDs. For consumer SSDs, our goal is to test in a system that has been optimized with our SSD Optimization Guide. To see the best performance possible the CPU C states have been disabled, C1E support has been disabled, and Enhanced Intel SpeedStep Technology (EIST) has been disabled. Benchmarks for consumer testing are also benchmarks with a fresh drive so that not only can we verify that manufacturer specifications are in line, but also so that the consumer can replicate our tests to confirm that they have an SSD that performs as expected. We even provide links to most of the benchmark software being used; simply click on the section heading with the benchmark’s name.
Today’s analysis will be conducted using our Intel DZ77GA-70K test bench. Clicking on any pictures or benchmarks will open a more easily viewable high-resolution image. Our test system is comprised as follows:
|PC CHASSIS:||Cooler Master N600|
|CPU:||Intel Core i7-2600K|
|CPU COOLER:||Cooler Master Hyper N520|
|POWER SUPPLY:||Sliverstone Strider 600W|
|GRAPHICS CARD:||EVGA GT610|
|MEMORY:||Kingston HyperX DDR3 1600MHz|
|OS STORAGE:||Edge Boost Pro 7mm 240GB|
We would like to extend a TSSDR hat-tip to Cooler Master for providing the N600.
After initializing and formatting the Silicon Power Slim S80 240GB, it shows up in our disk management screen as the E: drive. It also shows as healthy and having 223.57GB of user-available capacity.
The software in use for today’s analysis is typical of many of our reviews, and consists of Crystal Disk Info, ATTO Disk Benchmark, Crystal Disk Mark, AS SSD, Anvil Storage Utilities, Iometer, PC Mark8 and PCMark Vantage. Our selection of software allows each to build on the last, and to provide validation to results already obtained. All results are displayed through screen capture of the actual benchmark’s results for better understanding by the reader of the testing process.
Crystal Disk Info is a great tool for displaying the characteristics and health of storage devices. It displays everything from temperatures to the number of hours and power-on cycles the drive has experienced, as well as informing you of the firmware version of the device.
Crystal Disk Info verifies TRIM, NCQ, and that S.M.A.R.T. drive status and health reporting attributes are enabled. We were initially a little concerned with the fact that our review sample, although delivered in a sealed package, did not have a tamper-seal on the case. This raises the question of the drive actually being new or not, but Crystal Disk Info shows only 10 power-on counts and 0 hours, verifying that this is, in fact, a new drive.
ATTO Disk Benchmark is perhaps one of the oldest benchmarks going and is definitely the main staple for manufacturer performance specifications. ATTO uses RAW or compressible data and, for our benchmarks, we use a set length of 256MB and test both read and write performance of various transfer sizes ranging from 0.5 to 8192KB. Manufacturers prefer this method of testing as it deals with compressible data rather than random (which includes incompressible data), which although more realistic to everyday use, results in lower performance results.
ATTO testing reveals read speeds peaking at 555MB/s, and write speeds peaking at 535MB/s. Write speeds exceeded the 500MB/s threshold for all transfer sizes of 256KB and larger, while the read speeds exceeded that same threshold for all transfer sizes from even only 16KB on up. These are excellent performance results for a value-oriented product.
240GB seems nice, but other capacity points just seem too expensive compared to the competition.
I mean, you can get reactor 1TB for 300€ nowdays.
Benjamin Hojnik you seem to know a lot about ssds. I have been reading your posts for a wile now. I have a question for you. You can respond if you like. Ive asked this question on the forums but so far no ones has answered it. Maybe you know? How does a good amount of mushkin ssds have 2,000,000 hours MTBF when they use asynchronous, synchronous and toggle flash? But other manufacturers have 1,500,000 1,200,000 1,000,000 MTBF. Whats there secret?s MTBF when they use Asynchronous, synchronous and toggle flash? But other manufacturers have 1,500,000 1,200,000 1,000,000 MTBF. What’s there secret?
I wouldn’t focus on MTBF, the industry has no standard for measuring MTBF. Different manufacturers use different workloads to specify their MTBF. All you should really care about is what type of NAND they use. SLC, MLC, or TLC. If they use MLC or TLC NAND, make sure they use SLC caching. That’s it in a nutshell for right now. Best thing to do is read up on it on your own, there’s a lot of information out there.
Type of nand really doesn’t determine reliability of an SSD.
Not under typical workloads anyway.
What one should focus about is the controller. Thats the single most important part of the ssd and is also the #1 reason for faulires.
MTBF is really meaningless, ie more MTBF doesnt mean a more reliable drive, especially when comparing different brands.
I agree on all marks but the type of NAND for reliability. The controller is definitely where reliability is the most important, but higher density NAND flash increases bit rate error. To correct those errors ECC needs to either be programmed, or included in the controller. The type of NAND also affects endurance.
Whats the most reliable nand today?
Anything SLC based 🙂
Also, sammy’s 3D MLC comes close too. Apperently they internally tested their 128GB model and went thru 8PB of writes. Pretty mad, if its actually true.
Yeah this is true, but ECC engines in modern controllers are so powerfull and there is lots of spare area invisible to the user and controller (sandisk tlc apperently has 1/5 dedicated just for that), so it doesn’t pose that much a problem.
Crappy nand can be compensated with a proper controller. Crappy controller can’t be compensated with a proper nand.
Besides, most faulires come down to controller failing or trippin on acid, no failed nand. Some controller can actually withstand flash die faulires just fine (sandforce does that for example).
Whats the most reliable controller today?
Intel, marvell, samsung are pretty good choices with excellent track record.
Sandforce, phison, jmicron… Not so much (granted, their latter stuff got better).
I thought Intel controllers were/or made by Sandforce/Lsi?
No, Intel actually makes their own controllers too.
Intel 730 uses their inhouse controller for example.
But yes, their consumer stuff uses sandforce and is no better than other sandforce based stuff (apart from cherry picked nand).
No SLC Cache, disappointing. I’ll avoid Silicon Power SSD’s after their controller bait-and-switch tactics on the S60 and S70’s. It’s even worse than what Kingston and PNY did (and I avoid those drives, too).
Agreed. Who knows if the “chip’s close-up shows a part number of” for anything produced by these guys will be accurate down the road.
Their multiple choices of controllers used in these product lines is most definitely why their power consumption specs were generic.
I wouldn’t be too concerned about a lack of SLC cache in this drive, as it’s not using cheap TLC flash with its inherent latency and longevity concerns vs MLC. SLC hybrid drives often have issues with performance consistency due to firmware design, so sometimes simpler is better when the mfg isn’t cutting corners on flash quality.
Crucial’s MX200 is a prime example of the mixed bag that SLC hybrid design (DWA in their terminology) can be when applied to an MLC-based drive…until its firmware matures. Anandtech’s review traces are a good example of the potential compromises, mainly in latency issues when the firmware stumbles while managing the SLC-MLC combo under heavy loads. Their BX100 uses a lower end controller and is the simpler “budget” product with pure MLC but doesn’t have the glaring issues under certain loads that the MX200 does, which makes the MX200 unworthy to pay any premium for until the firmware is stabilized–especially since the BX100 is more power efficient.
The S80 is now down to $84.99 — .35/GB.
The firmware version tested in this review (7.2 as seen in the benchmark result screen shots) was outdated months before this article was published, and should have been updated to 8.0. I purchased one of these SPCC drives in late 2014 and tested the 8.0 release in January. 8.2 is the current firmware as of earlier this month.
Firmware updates are very easy using SPCC’s tool, and it automatically checks and installs the required updater when the utility is run since SPCC often swaps between Phison and Sandforce controllers in some of these product lines.
Making sure firmware is up to date is pretty important on drive reviews, especially these days as they are likely to be optimizing for PCMark 8’s more intensive testing methods (only available in the non-consumer version) vs generally prioritizing burst / light load performance in the past. I have seen negligible differences in the drive’s light load benchmark performance with the new 8.x firmwares vs the stock 7.2, but the story could be very different in the heavier PCMark 8 test traces. If you still have this drive, updated results would be appreciated.