OCZ Vertex Drives

[Halk]

the problem I see is that how do people even know if their drive is functioning correctly if the benchmarks are not consistent. There needs to be a way to figure out what baseline performance should be for every drive out there-

I think you're right on the money there. We've seen that results that are worse on ATTO are actually better in real use. The traditional low level benchmarks are pretty much useless for everything bar sustained transfer speeds, and for most purposes those don't relate to the general performance of the drive in any real way.

Then, if performance is not at expected levels, then there needs to be a way to distinguish improper configuration/incompatibilities versus ssd performance degradation.

Right. And from there people can see if for example their motherboard is 15% worse than it should be, etc. And from that point we can then go on to really find out what stripe size works best, what tweaks work and what don't, what effect RAID0 and RAID5 have, if a RAID controller does scale up right... etc etc. Until then we don't really have any way.

From what I have seen, no one really knows what standard to expect when benching the Vertex drives, as the benchmarks I have seen are nowhere near consistent. So, I am very interested to see what the "professional" review sites will come up with.

Dead right there again. It was pretty much at the time of Vertex being launched that all these revelations came into the "public eye", so to speak. Up until then we'd all been focused on sustained read/write performance, and it took a fair bit of convincing to set the see-saw of opinion in entirely the other direction. I think it's pretty much widely accepted by anyone who either is interested in hardware or does a bit of research before buying that the sustained rates are not the whole picture. It's reasonable to assume that there is some link between overall performance and the sustained rates - but it's not a direct link, generally speaking a drive that's faster overall will be faster at sustained transfers too. However there's now at least 2 drives out there that have sacrificed some level of sustained performance in order to get better overall performance. We know the Intel drives have a lot of controller work in the background to smooth things out, and we know the Vertex has internal cache and an IOP favouring firmware. Perhaps the reason for the lack of reviews is simply because the hardware sites don't want to produce a review with no real benchmarks.

From what I've seen on the OCZ forums, it's a lot of telling people to upgrade their hardware to hardware that was created within the last couple months, which is asinine- If there is going to be a requirement like that, there should be a very well placed disclaimer on all places where these drives are sold. Where is the compatibility list? After the Core drive issues, I would have thought a compatibility list would have been produced for all future products- As well as a proper configuration for each-

I'm not sure I entirely agree with you there. I think it tends to fall into a couple of categories. The first is mainly the JMicron drives where people have been advised that their motherboard SATA controllers are bad. This is pretty much true, however the workarounds do seem to get round it in the vast majority of situations. It's no surprise that motherboards are designed with what labels the manufacturer can stick on the box in mind, rather than future proof for what technology might be around the corner. The whole SATA standard is at best deliberately misleading. They don't want things labelled SATA 2 or SATA 3. They want them labelled SATA 3Gb and SATA 6Gb. My opinion is that they want to be able to shout big numbers, and the fact that the hardware just is not capable is a side issue.

The other category is where the performance from the drive doesn't meet the performance from other drives with different controllers. I don't know what OCZ can or should do about this. Expecting them to test every single motherboard out there is too much, it's not done with RAM either and that's a more mature market than SSDs. Certainly there's no disclaimers on memory that state : "This memory is rated at 1066, however your motherboard could be a 2 year old one which won't run it." I think the onus is on both the motherboard manufacturers and the SATA-IO (the standards organisation). The SATA-IO are happy to brand equipment that is patently not fit for purpose, simply because they did not anticipate anyone could use it at the rated speeds, motherboard manufacturers as I've said were happy to market hardware that worked with everything currently available, but didn't really do what they advertised. It's along the same lines as advertising a wrist watch that can withstand a nuclear bomb. Nobody is ever going to test that it can withstand a nuclear bomb, but it's good enough that a car can drive over it without the face cracking. All of a sudden SSD drives appeared and the equipment sucks.

Having said all that though, most people have either resolved the problem, or found out that their laptop is crippled to use SATA1. For those expecting their motherboard to deliver RAID0 performance I think it's nobodies fault but their own for assuming that - and again probably the motherboard manufacturer for not stating that : This motherboard is rated at SATA 3.0Gb but can't actually deliver it, and you can just plain forget trying to run RAID0 to increase performance since due to the chipset constraints we used a PCI lane (or an x1 PCIe 1.0)to connect the Southbridge to the Northbridge.

I think all in all the OCZ community has done a great deal towards advancing knowledge of SSD technology, and most of the problems are not related to what OCZ actually sells, it's down to Windows being optimised for mechanical drives and as I've said hardware not actually being capable of what the manufacturer claims (or probably more exactly what they imply). Why that has happened I don't know if it can all be put down to OCZ's credit. Perhaps the problems with the JMicron controllers helped the OCZ forum, perhaps it's just a fluke it happened with OCZ forums and not one of the other companies that sold JMicron controller drives. I believe it's at least partly due to the way OCZ operate that they're the ones who have the forum everybody reads.

As far as what's in store for SSD in the future I very much hope that SSD on SATA is not the only way. The SATA standard is not set by companies making SSDs, it's set by large PC vendors and mechanical drive makers. I don't for one second believe that they're going to turn around and admit that SSD exposed their marketing spin. Nor do I suppose they're going to go hell for leather towards doing all they can to help SSD technology. SSDs don't actually need SATA, currently SSD manufacturers are taking new technology and making it into old technology. There's no reason why things like Fusion drives can't be mainstream in a year or a few years without any need for a SATA controller to connect to the SATA controller on the drive. Just whack it into a PCIe slot.

Edit : I should probably link this page of cynical spin from the SATA-IO. Notice how they are insisting the technology must be referred to as SATA 6.0Gb for marketing purposes. The same way they wanted SATA2 to be referred to as SATA3.0Gb in marketing.