For maximum performance, an SSD should be run on a SATA 6Gbps controller like the PCH67 used on Sandy Bridge-based motherboards. Different controllers will produce different results and the statistics reported here should be seen as a guideline, not a guarantee.
An older nVidia graphics card was used, as opposed to 6000-series ATI card, as our partition-management utility (Partedpagic) refused to function correctly with any of the newer ATI cards.
All tests are run multiple times with a secure erase being run after each batch of tests.
No case was used during testing.
Solid State Drive testing consists mainly of running benchmark software that can determine transfer speeds and access times. These are known as “synthetic” benchmarks and attempt to reveal what the true capabilities of a device are. There are other benchmarks that can be run to try and emulate what a user can expect to achieve in their daily computing environment, a/k/a “Real World” benchmarks. There is one last thing most reviews will include, and it is not actually a benchmark per se: subjective experience – the “feel” the drive gives when used by the reviewer. This is based on experience and noticeable differences in the usage of the product and, while not completely arbitrary, it is entirely based on the reviewers’ point of view and test equipment.
PCMark 7 is finally out, but will be introduced at a later date for storage reviews.
ATTO disk benchmark is an older benchmark, but a very thorough one. At default settings, it tests a variety of transfer sizes from .5KB, which are doubled each round up to 8 MB (8 Megabytes / 8192 Kilobytes), with an overlapped IO with a queue depth of 4. Other software will run 32 or 64QD for 4K IOPS, whereas ATTO runs a 4QD on all tests. Results read very high on SandForce controller drives due to the highly compressible nature of the data and the increased threads.
Amazingly, the OWC drive is the first drive ever tested that actually meets the 4K aligned IO performance. With a rating of 240 MBps write speed at 4K file size, this results in a 60K IOPS! Amazing speed.
After all testing was completed, including bootracer runs, nearly 200GB of data was written to the drive, leaving approximately 7GB Free. ATTO was run again. Only a few quick tests were done, and no secure erase was performed between software.
It was no shock to see 4K writes take a performance hit. What was shocking to see was that, across the board (except for 4K writes), performance actually increased. DuraWrite technology is really doing a great job.
Crystal DiskMark is a favorite of many SSD reviewers because of its simple to understand interface and pleasant looking screenshots. Looks aside, the app performs as expected, but end users should note that the software now includes the OpenCandy network. This is a mild form of adware that suggests to the user other software they might be interested in. Downloading the ZIP file instead of the EXE bypasses this nonsense.
Crystal DiskMark tests 4 different types of transfer files with files sizes located next to the statistics. Seq stands for sequential testing, and it uses a 1024K transfer size.
By default, Crystal DiskMark uses a random fill pattern that is not easily compressible for testing, and as such, the performance read from this will never equal the advertised maximum speeds. What it does reflect, however, is worst-case scenario for transfer rates, so there’s definitely merit to it. In testing, both random and 0 fill tests will be run to show worst and best case scenarios.
Performance is off the charts again for a single drive being tested. The QD32 4K tests are significantly higher than ATTO 4K, posting an astounding 84,000 IOPS 4K aligned! Random DATA testing put the drive at just over 60,000 as well, which is the advertised specification.
For another quick and dirty test, Crystal was run after the drive was again filled to near-maximum capacity, as in the ATTO test. 4K QD32 write tests took a hit again in 0 fill testing, while everything else decreased in performance only slightly. In Random fill testing, however, performance degraded fairly equally among all 4 tests. A very small drop for a drive nearing its storage limit (even though the storage controller itself still offers plenty of headroom).
AIDA64 is a great program for detecting, monitoring and benchmarking a home computer. Formerly known as Everest (and before that known as AIDA32) AIDA64 is the latest update for the popular system software utility, and with the newest beta fully supports the Sandy Bridge processor.
Disk testing is performed in 4 steps. A read suite is run, which includes the following four tests: linear read, random read, buffered read, and access time. Writing tests are also performed with linear and random modes, but buffered is not selected, as it is recommended not to use that test on flash-based storage devices.
As expected, ATTO performance was linear within margin of error, for the entire drive. The numbers reported here match fairly well the CDM 1MB test. As always, random read scores a small percentage better than the linear reads. Buffered read consistently score 1MBps less than Linear reads.
Linear Write performance is a straight line, it remains even the whole way through the drive. This is a perfect example of the overlooked benefits of SSD vs HDD. With no platter getting smaller and smaller, the same amount of data is processed regardless of where it’s stored, physically, within the drive.
Random Write performance is also nearly equal performance throughout. There is a bit more amplitude change in the performance, but the rate of change is consistent throughout the duration of the test. The average performance is close to the linear performance as well.
Average Write access is a full length of drive test as well. We can see that as duration increases, so does the access time. Previous generations of SandForce drives also displayed this effect; however, in those instances, it was also followed closely with a reduction in write speeds, which this drive does not demonstrate.
Amazing results from the Extreme Pro 6G.
PCMark 2005 HDD test suite focuses on XP performance with hard drives. It covers 5 areas of storage performance.
Futuremark and HWBot both have limits on XP startup speed for validation, although they differ and the value has changed over time. As this testing is not being used for validation purposes at ORB or the bot, XP startup speed was left unhindered to provide the full real world performance one can expect when using these drives.
Scoring a tremendous 88,842 PCMarks, the Extreme Pro 6GB, represents the fastest scoring SSD (or RAID of SSD) we’ve ever seen at TechREACTION. For single drive performance, nothing previously run here comes even close. 5 OCZ drives in RAID0 (SATA2 SandForce) managed to surpass the 1GBps data transfer mark and still scored under 69,000 PCMarks. This is an incredible feat, and really exemplifies the capabilities of the new generation of SandForce controllers.
The other scores are all very impressive as well, scoring confidently above anything tested so far in every aspect of this benchmark.
PCMark Vantage HDD test suite also has a startup test, this time updated to “Vista” startup. However 7 new tests have been added based on common activities for the home PC user.
Most of Vantage’s tests focus on Digital Memories, so even the HDD suite is geared at testing the capabilities of the drives for manipulating pictures, videos and music in Windows software.
Vantage will always score lower in the Storage testing benchmark than PCMark 05, although results will usually appear fairly close on slower drives. The OWC Extreme Pro 6G loses over 14,000 synthetic points, to weigh in at 74,367 PCMarks. Still, that’s a figure that beats even the 2nd best PCMark05 score by a fair margin.
Despite getting a lower overall performance score from Vantage due to the way they evaluate data, Hard Drive technology was a lot better when Vantage came out as opposed to PCMark 05, the Vantage tests score amazingly well. Windows Defender, which is an anti-spyware program, runs a scan speed of 345MBps. This boils down to a the drive being scanned in just over 10 minutes! This assumes a full drive… most OS drives are never full, and a normal 40GB OS install with some typical apps, would theoretically scan every single bit, in just a couple of minutes.
It is important to note that Anti-Spyware/Virus programs do not scan every bit of data on a drive by default, so actual scan speeds would be GREATLY reduced.
BootRacer is a great application that times how long it takes to load Windows. It measures both logon time and total time to desktop. While various tweaks are available to really improve startup time, they were not used during these tests. A full operating system load sequence was used, and all startup items were left enabled.
Boot speed is going to be affected by CPU speed and efficiency, as well as the amount, speed, and timings of your memory.
With an amazing time of 9 seconds to logon and 10 seconds to desktop, this single drive beats anything else we’ve tested to date.