OCZ SSD's - now slower and reduced (usable) capacity???

[GaryRW]

I've just done a quick search and can't see this has been highlighted by anyone - apologies if it has.

It seems OCZ has quietly moved to 25nm flash in their SSD's. (I haven't seen a massive reduction in price, have you?) Lots of people are reporting both that the performance is not as good and the formatted capacity is lower due to increased over-provisioning.

The performance issue seems to be a bit unclear at the moment. OCZ certainly don't seem to be openly holding their hands up to any performance hit. But they do seem to be openly admitting that there is a requirement for increased over-provisioning. The explanations on their forum seem a bit misleading to me - they keep saying that the amount of flash is the same (e.g. for 120 gig drive, it still has 128 gig) and keep referring to the age old problem of GiB vs GB. However, people are reporting that the new 25nm Vertex 2 120 drives are showing up as 115GB (first link below confirms OCZ accept that), not 120GB and they seem to side step that the drives are actually reduced in usable/formatted capacity.

So if all this is taken at face value, OCZ seem to think it's ok to reduce usable space without making it a new product or informing customers of the change...

a few links:

http://www.ocztechnologyforum.com/fo...ighlight=115gb

http://www.ocztechnologyforum.com/fo...OCZ-SSD-drives

http://www.guru3d.com/news/ocz-verte...ported-slower/

http://www.storagereview.com/ssds_sh..._you_need_know

[nibbler]

My vertex 2e 120gb shows as 111gb, I'm pretty sure this happens with many drives, no problem for me, I knew it before hand and it's still a great drive.

[Phage]

My vertex 2e 120gb shows as 111gb, I'm pretty sure this happens with many drives, no problem for me, I knew it before hand and it's still a great drive.

Same here. I believe that those blocks are set aside as vacant space to kep performance up.

[[GSV]Trig]

Still a good drive but for people that have an older drive and are looking to upgrade or RAID 2 drives together then theres issues, model/part numbers should of been changed...
I was about to order a new drive but not so sure now...

[badass]

My vertex 2e 120gb shows as 111gb, I'm pretty sure this happens with many drives, no problem for me, I knew it before hand and it's still a great drive.

Same here. I believe that those blocks are set aside as vacant space to kep performance up.

I think you've both missed the point:
The old 34nm drives Show an unformatted capacity of 120GB
The new 25nm drives show an unformatted capacity of 115GB
The old 34nm drives Show a windows formatted capacity of 111 GiB
The new 25nm drives show a windows formatted capacity 107GiB

They are still advertised as Vertex 2 120GB despite only having 115GB and different performance.
Both drives actually have 128GB of NAND. The problem is that they use half as many NAND chips.
RAISE needs 1 NAND chip for redundant data, regardless of capacity, like RAID-5

Regardless of how you look at this, calling a drive with 115GB unformatted capacity a 120GB drive is simply a lie.
To avoid lying, they can simple call the 25nm drive a Vertex 2 115GB. They refuse to do this.

If you bought a car that advertised a 60 liter fuel tank and only got a 55 liter fuel tank in yours and it turns out that the manufacturer continued to advertise a 60 Liter fuel tank, you'd probably be unhappy.
If you were then told it's OK because very few people actually let the needle go into the red so they are not lying you'd probably be even more annoyed after running out of fuel.

TBH, this won't put me off buying one - however I'm not paying current prices for one. £110+VAT for 115GB sounds more like it.

[GheeTsar]

I have a tiny bit of sympathy for OCZ - they claim they can only buy 25nm chips now (and that most SSD manufacturers are in the same boat) and as they are not as robust as the 32nm chips they have to increase the reserved space to ensure the drives wear at the same rate as before.

In theory the new chips should facilitate a price drop too, which would be a good thing. I've not seen this materialise though.

However it does seem below board that the new drives are not as good as the previous versions and yet there is no way to tell which you are buying short of opening the SSD and voiding the warranty in the process.

I'd personally steer clear of OCZ as a result - but then again; who's to say the other manufacturers aren't doing the same too?

[nibbler]

OK, it was slightly unclear from the OP. I see your point then, but mine is 111gb formatted so I'm on an old 34nm drive then?

[GaryRW]

OK, it was slightly unclear from the OP. I see your point then, but mine is 111gb formatted so I'm on an old 34nm drive then?

Yeah, badass explained it better than me....

(tbh, I've never been 100% sure about the whole Gib vs GB thing when it comes to SSD's - until this came up, I thought it wasn't an issue as flash storage was binary in nature, but obviously I was wrong)

Anyway, from what you've said - yes, you've probably got a 34nm drive. What struck me as I was reading through the OCZ forum threads though was that even the OCZ staff couldn't tell people what drive they had easily. I did kind of feel sorry for the staff as they clearly had a lot of unhappy people, and obviously they can't just just say "yes, our company screwed up". Having said that, most of the people complaining were being fairly reasonable in what they were saying, and making it clear that whether or not this is a huge issue, the fact is the product changed but OCZ didn't tell anyone and some of the replies from staff were a bit arsey or condescending.

Gheetsar - OCZ say other drive makers are doing the same thing, and (so OCZ said...), other manufacturer forums are also full of complaints. Can't say I've looked though. Just thought I'd mention it here as it's something I'd have wanted to read if I was buying at the moment, and I hadn't seen it on Hexus, Anandtech or the other sites I usually read. It was actually just a small post on SPCR that I saw it on.

my own 2p for what it's worth is that they clearly should have just called them Vertex 2.5's or something and (ideally) passed on the cost savings. I've not been keeping that close an eye on prices, but I can't see that they've dropped much

[badass]

OK, it was slightly unclear from the OP. I see your point then, but mine is 111gb formatted so I'm on an old 34nm drive then?

Based on what OCZ have said, yours is one of the 34nm drives.

[badass]

I have a tiny bit of sympathy for OCZ - they claim they can only buy 25nm chips now (and that most SSD manufacturers are in the same boat) and as they are not as robust as the 32nm chips they have to increase the reserved space to ensure the drives wear at the same rate as before.

The decrease in capacity is because they have no choice, rather than them choosing to use more over provisioning. It's like having a RAID-5 array of 10x1TB drives as opposed to 5x2TB drives. Without redundancy, you have 10TB in both cases. With the necessary 1 disk capacity lost for parity, in one case you have 9TB usable and the other you have 8TB usable. You can't increase this by using only 1/2 disk for parity.
Based on their marketing bumph, RAISE is like RAID4 or RAID5.
Imagine the 2 RAID arrays I mentioned before:
Both arrays have an imaginary 4k stripe size and you write your data in 32k blocks.
In array 1 (the 10x1TB array), you actually write 36k. 4k to eight of the ten drives and another 4k to the parity drive. One drive has nothing written to it. Total: 4k to 8 of the drives and 4k to the smaller parity drive.
In array 2 (the 5x2TB array), you actually write 40k. You write 4k to the first 4 drives and 4k to the parity drive, then another 4k to each drive and the parity drive. Total 8k to the 4 drives and 8k to the parity drive.
Allthough you have twice as much spare area, you have to write twice as much parity data to it.
Either OCZ are lying about drive longevity being unaffected or RAISE doesn't work as advertised.
I suspect it's somewhere between the two. RAISE is probably far more complex than it's worth them explaining to the general or even enthusiast public but it's a catchy marketing term.

One more thing:
The 100GB Vertex 2 also has 128GB NAND. The 120GB Vertex 2e is the same except it has less spare area.
Perhaps they should be calling them all Vertex 2e 100GB drives since they have the same amound of NAND.

[GheeTsar]

The decrease in capacity is because they have no choice, rather than them choosing to use more over provisioning. It's like having a RAID-5 array of 10x1TB drives as opposed to 5x2TB drives. Without redundancy, you have 10TB in both cases. With the necessary 1 disk capacity lost for parity, in one case you have 9TB usable and the other you have 8TB usable. You can't increase this by using only 1/2 disk for parity.
Based on their marketing bumph, RAISE is like RAID4 or RAID5.
Imagine the 2 RAID arrays I mentioned before:
Both arrays have an imaginary 4k stripe size and you write your data in 32k blocks.
In array 1 (the 10x1TB array), you actually write 36k. 4k to eight of the ten drives and another 4k to the parity drive. One drive has nothing written to it. Total: 4k to 8 of the drives and 4k to the smaller parity drive.
In array 2 (the 5x2TB array), you actually write 40k. You write 4k to the first 4 drives and 4k to the parity drive, then another 4k to each drive and the parity drive. Total 8k to the 4 drives and 8k to the parity drive.
Allthough you have twice as much spare area, you have to write twice as much parity data to it.
Either OCZ are lying about drive longevity being unaffected or RAISE doesn't work as advertised.
I suspect it's somewhere between the two. RAISE is probably far more complex than it's worth them explaining to the general or even enthusiast public but it's a catchy marketing term.

One more thing:
The 100GB Vertex 2 also has 128GB NAND. The 120GB Vertex 2e is the same except it has less spare area.
Perhaps they should be calling them all Vertex 2e 100GB drives since they have the same amound of NAND.

Perhaps I'm being thick - but how does the die shrinking of the chips relate to how they are organised and their arrays within the SSD?

[[GSV]Trig]

Read
http://www.storagereview.com/ssds_sh..._you_need_know
and
http://www.storagereview.com/ocz_ver...czssd22vtxe60g

As i understand it the newer drives have larger "blocks", they still use 1 "block" for whatever it is but because the blocks are bigger on the newer drives you loose more space than you would on an older drive.
They work out slower because there are less physical chips that the controller is talking to, imagine a RAID array with less physical disks, performance would tail off..

Thats how I see it, might be a bit wrong but its close enough for my head to work around lol

[GheeTsar]

Thanks for the link - this is the key bit that explains it:

The 60GB 32Gbit drives feature 16x32Gbit (4GB) Nand IC with 7% OP
The 60GB (55GB formatted) drive features 8x64Gbit Nand IC with 7% OP

[badass]

Read
http://www.storagereview.com/ssds_sh..._you_need_know
and
http://www.storagereview.com/ocz_ver...czssd22vtxe60g

As i understand it the newer drives have larger "blocks", they still use 1 "block" for whatever it is but because the blocks are bigger on the newer drives you loose more space than you would on an older drive.
They work out slower because there are less physical chips that the controller is talking to, imagine a RAID array with less physical disks, performance would tail off..

Thats how I see it, might be a bit wrong but its close enough for my head to work around lol

A far less verbose explanation than mine

[miXer]

Which SSD is better out the following 2?

OCZ Vertex 2
http://www.scan.co.uk/products/80gb-...rce-controller

Corsair Force
http://www.scan.co.uk/products/90gb-...-write-275mb-s

[[GSV]Trig]

A far less verbose explanation than mine

Thats how I read it anyways, probably wrong didnt read the articles just kinda skimmed them..