New SSDs crippled?

With SSD prices decreasing, I’m looking to buy a 60GB model. (At the moment, the Intel 25-M 80GB is too expensive for me.)

I was considering the OCZ Vertex 2, but recent models suffer from lower program-erase (P/E) cycles, lower usable capacity, and write speeds limited to 37MB/s. These problems apparently steam from OCZ’s quiet transition to 25nm memory. I don’t know if these problems are correctable via a firmware fix or not. There is no known way to determine (prior to purchase) which units suffer from these issues. (Source: Support Question 60GB Vertex 2 (Extended) - Poor Write Performance)

I then considered the Corsair Force Series F60. However, a recent review shows slower writes speeds than review benchmarks conducted shortly after launch. (Source: Corsair Force Series F60 - DigitalVersus)

Are manufacturers degrading performance and coasting on previous reviews? I don’t want to roll the dice with a Vertex 2, but I’m not 100% confident with the F60 either. If anyone recently bought a Corsair Force Series SSD, I’d appreciate seeing benchmarks that contradict this review.

 

get yourself a x25m and call it a day.

The hassle of dealing with these things way outweight any performance gain(if there is any) or price difference(if there is any).

60GB is also a bit too small for modern day computer. Even my 80GB is a bit tight(and I have zero video on my computer with flash forever disabled).

 

Awaiting for Intel G3 SSDs. 160 GB here I come

 

BTW, if they are really using 25nm NAND for the same size, there is no surprise that it is slower, no matter what brands we are talking about.

SSD get its speed from mutiple channel + mutiple block per channel(basically 2 evel of RAID 0). With the new 25nm NAND, for the same drive size, they would be using less chips(may be only 1/2 of the original number) so either there is less channel and/or block.

This is also the reason why the 40GB Intel is only about half the speed of the 80GB.

If it is using the 25nm NAND, it needs to be double the drive size to get the old generation speed(roughly, there may be some controller improvement).

 

Okay if the smaller the NAND gets why bother making it smaller? I thought it makes them last alot shorter, read/write speeds are worse? Sounds like a step backwards?

 

Thanks for the info, chimp.
Tsunade, from what I've read, 25nm is cheaper to manufacture. This would lead to higher profit margins.

 

size, you get more gates for the same area. the read/write speed for the NAND itself may have improved(but only kind of incremental) but the read/write speed for the SSD is worse, for the same size drive.

my 80G is using 8x10 (10 channels as far as I know). The 120 and 160 is using 2x8x10 or (4+8)x10 which is why they are faster. If Intel use 25nm for the same 80G, it will most likely be 16x5, and I would be getting the 40V speed.

edit:
there is a way to know it. just open the drive and count the chips. the less there are, the slower its speed.

 

That, and 25nm will allow higher densities. So the chips could be 4GiB each instead of 2GiB (pulling numbers out of the dark). So they could run the same number of chips and have more capacity, or fewer chips and the same capacity but be slower. But you'll still get more storage and likely more performance for about the same price. It's how these things work. A Pentium Pro had about a ~300mm^2 die size, and about 20 million transistors at .6 microns (600nm). A current Sandy Bridge chip is only 150mm^2, yet packs in 624 million transistors at a 32nm scale. The cost is roughly the same for something "smaller", but overall everything has progressed. Same thing will happen with SSDs.

 

Right I know you get more space...but if the life of the SSD goes down, what's the point?

 

What does it matter to the manufacturer if the life goes down when they need to compete with other companies whom are ALSO rebranding their SSD.
The pubic wants cheaper SSD, only way to do that is to decrease the lithography, the lower the lithography the more capacity the nand can hold@ nearly the same manufacturing cost but nearly doubling capacity.
Add 40gbs to an 80gb, making it 120gb and sell it for previous 80gb prices while your manufacturing cost has nearly gone down by half.

Math:
Previous generation SSD:
80gb SSD=$160 or $2 a GB
40gb SSD= $110~ or $2.75 a GB

New wave of SSD:
80gb SSD+40GB=120GB for $160~ or $1.33 per GB

So you say, but the price has gone done, indeed it has. However manufacturing costs have stayed the same, the difference is they are still pulling the same profit as the 80gb and the 120gb.
The fact they can Pull the same profit, and sometimes even a little more considering the manufacturing costs go down, they can buy even more produce and get even larger savings, SPECIALLY when talking about such massively produced NAND, were supply never ends.

 

I understand why manufacturers would try to cut costs. I don't appreciate OCZ degrading performance of their drives while trying to coast on the established reputation of the Vertex 2 series.

Has anyone with a recently purchased Corsair experience poorer than expected performance?

 

Well don't quote me on this, but I remember a post that said SSD life was dramatically reduced, and you would start seeing degrading performance within like 2-3 years?

I guess it also depends on usage. I don't use my SSDs for extreme work or file servers, mine pretty much function as a normal boot drive.

I don't fancy reinstalling everything every 2-3 years...and I'm sure large IT departments wouldn't appreciate it either.

 

Yes while life expectancy has dropped, many people argue that it's very rare to see a SSD die, yet AnAndTech managed to brick one of the first Sandforce drives within weeks of use.
The issue is that SSD do get bricked eventually, the problem is most people don't know why their SSD bricked. Nor do they run benchmarks to set up data points to ensure their SSD is working as it should, most just forget about it and never record information so when their SSD begins to deteriorate, they never notice.
Trim+ windows 7 has reduced the need for maintenance, so now a days you could run without needing to reinstall windows or see much if any performance drops.

 

that is different thing. NAND endurance doesn't brick a drive. It only makes it no longer writable.

 

If you can't write to nand that means you loose storage, eventually it will become useless and will no longer write, and if I am not mistaken there may be issues with reading.
I am not too familiar with SSD, or at least as familiar as I'd like to be.

 

true but brick, AnandTech(and many other people using SF) just lost the drive in the BIOS.

As far as I know, at least all the documents and claims I have seen, a non-writable NAND doesn't lose data, you can still read whatever is stored there.

 

Gotcha. So basically while the nand write/erase cycles will expire the SSD will not be bricked by it, although technically you can't use it.

 

I did two Corsair F80 and one Corsair F120 installs around mid December and they worked great and got the advertised speeds (tested with ATTO).

 

They are setting aside more spare area to get the life expentancy up.

On the up side, drive life should be similar, on the downside it has less space.

 

Not trying to be picky but you can get the advertised ATTO speed on any SF drive, no matter what NAND it use. ATTO speed on SF is a test of the interface(whether your AHCI/IDE driver is setup properly etc.), not the NAND itself.

 

as long as the life is so big that it doesn't matter for the typical customer, who cares? yes, on the interwebs, people moan about it all day long. but it does not matter actually.

smaller transistor size == cheaper to produce, more storage, possibility to access faster, use less power.



about the 2-3 years. well, consider then having a low price high storage size ssd that delivers for 2-3 years, and then the possibility to get a then-low-price high-storage-size one (means much cheaper, much more storage, much faster). it's not that much of an issue.

AND it's the minimum lifetime of the flash chip. 2-3years is the minimum lifetime of the ssd for ordinary usage. doesn't mean it HAS to die afterwards. it can last you for years.

 

I won't buy one until there's solid data on how long the drives ACTUALLY last in normal everyday use. The 20gb hard drive in my old Athlon XP rig booted faithfully until I decided to throw the whole thing out a year or so ago.

 

there is solid data. the stuff is measured and extrapolated. it's science. and science works. (it's really simple math.. like how long till a cell dies * how many cells / how many data you write).

and that's the beauty of the ssds. it's all computable there. for a hdd, it might last for ever. or die after the first day. without doing much more than having a very short shake of the drive (only 5mm or something, just get that needle to scratch the disk somehow.. and it's quite easy on a laptop).

an ssd has a known minimum lifetime. a hdd does not. i actually prefer the ssd thanks to it. there, i know what i get.

 

Well it's not like I am anti-SSD, I personally own 2 of them. But SSD can have logical failures just like hard drives can, possibly on day 1 (BIOS not seeing it, not booting). Corrupt partition tables unfortunately can happen on an SSD too. But the way they test MTBF really isn't a good way of testing 1 individual drive to see how long it lasts.

 

it has a MUCH SMALLER failure vector, though. that's whats important. anything can fail anytime. but an ssd does have a very small random-fail-vector, a near-zero-physical-fail-vector, and a nice controllable lifetime-fail-vector.

a hdd has much more aspects where it can unpredictably fail. much more.


btw, they don't test mtbf on an ssd physically to estimate the lifetime. they use the known facts about all flashchips they ever used, tens of years of knowledge about flash.