Having already seen the performance possible using an ARM processor drive controller when we tested the OCZ Vertex we were expecting the same from the Corsair P256 and we weren't disappointed. Theoretical performance in HD-Tach pointed to the Corsair possessing phenomenal write speeds as well as top end read and burst speeds and that translated perfectly into write, copy and read speeds in FC-Test.
The P256 is the fastest MLC drive we’ve yet to test when writing or copying files, and even manages to pip the SLC based Intel X25-E when it comes to copying large files thanks to its beefy 128MB of cache. Read speeds in FC-Test were also excellent and were on par with the OCZ Vertex when dealing with small files and just a little slower when dealing with larger files – a fantastic showing all round when it comes to sequential speed.
Random read and write performance wasn’t quite as impressive though, and while the P256 is by no means going to suffer from drive stuttering, random read and write speeds were notably slower than the OCZ Vertex, and both average and maximum latencies were a little slower too. However, our random read and write tests are purely synthetic and having used the drive extensively there’s no noticeable difference in responsiveness or everyday operating system performance between the P256 and the Vertex, and the P256 is still a great deal faster than even a 10,000rpm hard disk.
SSD performance degradation is something we’re becoming more and more aware of as we use drives for longer, and there’s a lot of buzz around the TRIM command. While we watch for degradation during testing (which the P256 was happily free from), most SSD performance degradation occurs following extended periods use, when cells of data on the drive become partially filled with deleted files.
Reusing these cells requires the drive to offload the data that’s still required to the cache before rewriting it elsewhere; resulting in a read/erase/modify/write process when rewriting previously used cells that takes time and compromises performance. Windows 7 promises to resolve the issue by implementing TRIM, a command that carries out this process when a file is first deleted, but Windows Vista does not yet (and likely will never) support the function.
Click to enlarge
To combat this problem some drive controller manufacturers have started releasing manual TRIM applications which can be used at intervals to carry out the TRIM command en-masse, and hopefully going a long way to restoring original performance. As of yet only SSDs based on Indilinx’s drive controller offer such functionality, but we’re hopeful Corsair and Samsung will follow suit. While their drive does not currently support the TRIM command, Corsair and Samung told us they are "actively working on a solution and hope to give a more detailed update to our users early this summer as to update availability."
For now though, you’re limited to low level erasing the drive with an application like HDD Erase to solve cell fragmentation.
Fail on Format
While our testing of the Corsair P256 was positive in regards to results put down, we did notice something very interesting when carrying out our tests - something Corsair, Samsung and a number of other hardware review sites have failed to pick up on; an issue which has the potential to severely cripple the drive. When repeating our HD-Tach tests in Windows Vista we decided to perform a full format (rather than a quick format) on the drive to ensure “as new” performance. However, after formatting we found the P256’s awesome write speeds had slowed to a crawl and our FC-Test write tests now took over six times as long to complete. The drive’s write performance was now terrible.
After consultation with Corsair the problem remains unexplained and while entirely avoidable so long as you only ever quick format the drive, it's a potential landmine to those not aware of the issue. Once a full format has been completed the only way to restore the drive’s true performance is to use a low level erase program such as DOS based HDD Erase, a procedure that’s a fair bit more complicated than fitting a S-ATA hard disk drive and we’re concerned that some users will simply think their SSD is faulty and try and RMA it to Corsair.
Corsair had this to say regarding the format bug:
"Corsair is in the process of putting together FAQs and other items to show users how to recover performance on the drive if a full format is done in Vista. And we are continuing to work with Samsung to get a better understanding of what causes this anomaly. In the event that customers experience the problem, our tech support team will work with them to recover their performance, rather than asking the customer to return the non-defective drive."
We'll add a news post as and when we know more, or when Corsair releases a firmware fix.
Even with the very serious caveat of never, ever full formatting the drive, the Corsair P256 remains an excellent bit of kit pitched squarely at the high end. Superb sequential performance, especially when it comes to write speeds allows the P256 to better even the OCZ Vertex, with brilliant windows boot and game load times to match.
We’ve said it before but using a decent SSD has the ability to make your whole PC experience feel more responsive and the P256 delivers that convincingly. It’s also better value per/GB in comparison to the other ARM based 256GB drives currently available and while £530 might be a tough chunk of change to swallow or even justify, you’re getting a very fast drive for your money. A number of the bit-tech
and CPC staffers have recently upgraded to 128GB SSDs and have found trying squeeze an OS and a hefty Steam directory into such a small drive can be trying; having that extra space from a 256GB drive makes things a lot more comfortable.
While the fact that it’s easy to cripple its performance doesn't exactly inspire confidence, the Corsair P256 is unquestionably fast when it comes to sequential, random and real world speeds. If you’re aware of the formatting issue and are looking to grab 256GB of high speed SSD goodness then the P256 is certainly worth considering, if only for the £100 saving in comparison to competing drives.