Introduction and Technical Specifications
When Solid State technology first appeared on the marketplace it took a little while before the full benefits of it were felt. Not in terms of marketplace saturation, or pricing, but in pure unadulterated performance.
OCZ Technology were one of the first companies to really bring us a drive that was a leap forward over the current mechanical drives when they introduced the Vertex series. It quickly became the weapon of choice for the early adopters and found homes in many high-end systems.
Since then controllers and chips have become much more homogeneous and differences are far less apparent between read speeds of even the base models in a companies line. Write speeds have become the defining factor that separates an "average" SSD from the high performance variants. Of course average in SSD terms is still a quantum leap forward from mechanical drives.
Today we take a look at the OCZ Vertex 2, the follow up to the extraordinarily successful Vertex. Will it be yet more of the same, or has it got enough to keep it on everyones short list?
As always here at OC3D we want to make absolutely sure that not only do you have the most accurate information, but also that manufacturers aren't making unsubstantiated claims. Therefore we always ensure we take the technical specifications from the official website. With SSDs this naturally, much like the testing, ends up being fairly simplistic. How quick is it, and that's about it. Note that although performance figures for the 3 lower capacity drives are beneath those quoted here, for brevity we are only using the ones that we should obtain with the model on test today.
- Available in 40GB to 400GB capacities
- Available in 60GB to 480GB extended capacities
- Native TRIM support
- Max IOPS Firmware
- Seek Time: .1ms
- Slim 2.5" Design
- 99.8 x 69.63 x 9.3mm
- Lightweight: 77g
- Operating Temp: 0°C ~ 70°C
- Storage Temp: -45°C ~ +85°C
- Low Power Consumption: 2W in operation,
.5W in standby
- Shock Resistant up to 1500G
- RAID Support
- Included 3.5" Desktop adapter bracket
- Compatible with Windows XP, Vista, 7, Mac OS X and Linux
- MTBF: 2 million hours
- 3-Year Warranty
- Read: Up to 285 MB/s
- Write: Up to 275 MB/s
- Sustained Write: Up to 250 MB/s
All fairly standard stuff. I'd love to have one of these long enough to find out the accuracy of that meantime before failure, but even if I did I don't think I'm going to live another 171 years, so it's safe to say that it will outlive all of us. Time for a look at what we have.
The Vertex 2 in Pictures
SSD packaging is always a thorny issue. Anyone who has purchased a HDD will tell you that nearly all of them come in barely enough plastic to stop damage to the drives. Solid States on the other hand are a memory device and so require the same anti-static shielding we find on motherboards and the like. Therefore they also require some form of packaging to protect it as much as possible.
OCZ have got a very cool looking box. I'm not sure which part of our brains black and dark greys speak to, but it always makes us think of speed, performance, power and other things that are a benefit when trying to sell a high end product. In the same way Kharma don't sell baby blue speakers, so the world of PC Hardware is dominated by black.
One thing that is thankfully becoming more common is the inclusion of a caddy to hold the 2.5" drive in the standard 3.5" bays of 99% of cases. OCZs solution is identical, barring logo, to the one we saw with the Corsair Nova and certainly does the job.
The drive itself is finished in a matte textured black although curiously the reverse isn't. More often than not people don't see this side, but it's still strange to see the bottom of the case different to the top, especially as this is about the only area of the drive that a manufacturer can use to differentiate themselves from the pack.
Once the board is open we can see that this is a Sandforce 1200 controlled drive. The main differences this means to us as end users are two-fold.
Firstly Sandforce utilise a specialist architecture that uses the NAND MLC's themselves to act as the cache, theoretically providing faster transfer rates, whilst also helping keep costs slightly reduced because we no longer need an extra chip as cache to eliminate the stuttering that occurred in very early Solid State Drives.
The second difference is most obvious in the capacity. Unlike most drives this isn't actually 128GB but 120GB. This is because of part of the capacity is dedicated to performing the cache function.
If there is one thing we've learnt over our time testing drives here at OC3D it's that the larger the chunk being read or written, the better the performance. Although SSDs suffer far less than mechanical drives there is still a performance drop when dealing with very small data sizes.
HD Tune allows us to test using a few different data sizes and this should help highlight if the Vertex 2 with its SandForce controller manages to smooth this out with its alternative cache methods.
Sure enough the gap between the small 128KB chunk which normally would see a very low score, and the data chunks that are in the megabyte range, is far smaller than we'd expect.
However by far the most noticeable aspect is that the write speeds are consistently faster than the read speeds. Given that reading is exponentially more common than writing, even on an OS drive, this is surprising to say the least. Let's see if this is maintained throughout our testing.
IoMeter is new to the OC3D testing suite and so we haven't got any previous results to compare against yet. However it is a more accurate test of the drives ability to handle read and write requests. Naturally interpreting this data will become easier as we have more and more experience and data sets.
As you can see from the 4KB data test, much smaller than the 128KB test on the previous page, the drive still takes a hit when utilising such a small size to write.
Next we have a sequential write test using a 2MB data size and our results are much more like we'd expect, getting close to 220MB/s. Given that we have a much larger amount of data to push around the IOps is obviously much lower than we saw above. It's easier to send 4K lots of times in a second than it is to send a 2MB chunk.
The sequential read test using the same 2MB chunk actually dips below the 200MB/s mark. The IOps once again drops due to needing to hang around waiting for the Vertex 2 to finish what it's doing before sending the next packet along.
This is definitely a test we'll be returning to in the future, but for now the OCZ Vertex 2 has set the bar.
Our last "graph" before the couple of graphs. On the left we can see that utilising an 8mb block of data the Vertex 2 is very consistent indeed with the main peaks coming at regular intervals.
However once we shift to the 32mb data size we find a very inconsistent graph. There is no reason for the drive to be any faster or slower at the beginning or the end as we aren't reading from a moving platter like we are with a standard drive.
Crystal Disk Mark
Finally we bring our old friend Crystal Disk Mark to the party. We will be using it to compare the SandForce controller on the Vertex 2, to the SandForce controller we recently tested with the Mushkin Callisto. For comparison purposes we are also testing the Corsair Nova which uses an Indilinx Barefoot controller, and the Crucial C300 which uses a Micron controller and is the only drive on test which takes advantage of SATA3.
Starting off with our write test naturally the C300 walks it, but it's lower down the graph that the interest lies. The OCZ Vertex 2 and Mushkin Callisto use identical hardware, and yet the OCZ has a healthy lead in all of our read tests. Despite the SandForce controller dispensing with the cache chip to hopefully give improved performance, the Corsair Nova still edges ahead in the 4K and 512K testing, but has a big lead in the sequential read test.
The write testing was the area the C300 fell down on, and you can see here that all of the drives on test keep up. In the 4K testing the Mushkin Callisto just nips in front, although as every drive performs poorly with this size data variances in testing are more common. Once we move to a size that is more likely in daily use, the Vertex 2 once again comes out ahead although none of the drives can come close to the write speeds that the Corsair Nova provides.
When reviewing Solid State Drives it is very easy to find your head turned by gaudy numbers and snappy system response. Even a very poor performing drive is so comprehensively better than a hard drive that it's always difficult not to be impressed.
Strangely though the OCZ Vertex 2 never really gave that "wow" feeling.
For a long while SSDs used a combination of drive controllers, cache and NAND MLC that have constantly edged the performance boundaries upwards. The SandForce controller dispenses with the cache but, looking at our performance data and testing results, doesn't actually supply any benefit.
Read speeds are acclaimed at 285MB/s for this model, with write speeds around 275MB/s. In a couple of tests we certainly saw read speeds around 225MB/s, but once we moved into more thorough benchmarks that test the drive in "real-life" conditions, that figure quickly fell to under 200MB/s.
Write speeds conversely were significantly higher in some tests with speeds above the read results. HD Tune saw an average write of over 240MB/s in write tests. Even the harsh IoMeter test managed 220MB/s.
Considering we were using a i7 930 on the latest Gigabyte X58A-UD9 motherboard we clearly get nowhere near the reported 50000 IOps, getting just 14000 in our testing, and the read and write speeds are nowhere near OCZs claims either. Although the write speed is far greater than they claim and bettering proclaimed performance is never a bad thing.
When we take into account that SSDs are still are the very expensive end of the price/capacity bracket, it becomes clear that the SandForce controller actually lessens the performance over cache endowed models. It becomes pretty clear that the extra 8GB used up by the SandForce could be much better served as data storage.
If the loss of a chip did result in a much cheaper product then we would make allowances for some performance foibles in the name of good value. Unfortunately the OCZ Vertex 2 is available online for around £10 more than the Corsair Nova. As we have to acknowledge the Nova has the edge in price and capacity (another 8GB) and comprehensively bests the Vertex 2 in performance testing it becomes almost impossible to recommend.
It's too expensive, under performs and has reduced capacity compared to similar drives. Technology has moved on, and the Vertex 2 is clearly a step behind.
Thanks to OCZ Technology for providing the Vertex 2 for testing. Discuss in our forums.