The SandForce Roundup: Corsair, Kingston, Patriot, OCZ, OWC & MemoRight SSDs Compared
by Anand Lal Shimpi on August 11, 2011 12:01 AM ESTAnandTech Storage Bench 2011 - Light Workload
Our new light workload actually has more write operations than read operations. The split is as follows: 372,630 reads and 459,709 writes. The relatively close read/write ratio does better mimic a typical light workload (although even lighter workloads would be far more read centric).
The I/O breakdown is similar to the heavy workload at small IOs, however you'll notice that there are far fewer large IO transfers:
| AnandTech Storage Bench 2011 - Light Workload IO Breakdown | ||||
| IO Size | % of Total | |||
| 4KB | 27% | |||
| 16KB | 8% | |||
| 32KB | 6% | |||
| 64KB | 5% | |||
Despite the reduction in large IOs, over 60% of all operations are perfectly sequential. Average queue depth is a lighter 2.2029 IOs.
As the only 240GB drive running 3.20 the Kingston HyperX is the fastest drive in our test here (only the Vertex 3 and possibly the Corsair Force GT even have 3.20 available for update). The advantage is still pretty small though, at 6% above its closest competitor in an IO bound test you'll never see that advantage in the real world.
Facebook
Twitter
RSS
90 Comments
View All Comments
imaheadcase - Thursday, August 11, 2011 - link
I was wondering the same thing...this seems to happen a lot lately with roundups.Anand Lal Shimpi - Thursday, August 11, 2011 - link
My apologies! An older version of the graphs made its way live, I've updated all of the charts :)Take care,
Anand
Nickel020 - Thursday, August 11, 2011 - link
I always thought the difference in price between a 25nm SF1200 drive and a synchronous SF2200 was mainly due to the cost of the controller, but since you put the controller at $25, it's the NAND in the SF1200 that must be cheaper.A Corsair F115 with synchronous 25nm (G08CAMDB)* costs $170, a Force 3 with asynchronous NAND costs $185 and a Force GT with synchronous NAND costs $245. The synchronous NAND in the F115 must be way cheaper than the synchronous in the Force GT thus.
I'm guessing the SF2200 is more expensive than the SF1200, so that basically means that following your cost breakdown, the asynchronous NAND in drives such as the Force 3 or Agility 3 must be similarly priced as the synchronous NAND in the 25nm SF1200 drives.
Why is the synchronous in the SF1200 drives so much cheaper than the one in the SF1200 drives? Could you decipher the the whole part number?
*I'm assuming the F115 uses the same NAND as the first Vertex 2s with 25nm:
http://www.tomshardware.de/ocz-vertex-2-25nm-ssd,t...
Coup27 - Thursday, August 11, 2011 - link
If the current state of affairs are due to the reasons you have outlined in the first couple of paragraphs then this has been brought on by the manufacturers themselves.All the manufacturers have tried to bring costs down as much as possible for obvious reasons, but they should not have brought them down so low that they sacrifice validation and testing to get there.
The benefits SSD's have over HDD's are enormous and I am sure I am not alone when I say that I would quite happily pay an additional 15-25% than the current prices for my drive knowing that it works, full stop.
QChronoD - Thursday, August 11, 2011 - link
I understand sync and async, but not really sure what toggle means. Is it safe to assume that means that it can switch between the two modes? Or is there something else that is special about it?Nickel020 - Thursday, August 11, 2011 - link
It's a different NAND standard. Intel/Micron NAND follows the ONFI standard (which they developed afaik), Toggle is another standard that's developed by Samsung and others, the Toggle NAND in SF2281 SSDs is 34nm from Toshiba.If I understand it correctly, the difference is mainly the interface, with which the MLC cells are connected to the controller. Both are MLC though, the basic principle on which they are based is the same.
The Toggle NAND SSDs are generally faster, because 34nm means less density, more NAND dies, and thus more interleaving. Same thing causes bigger SSDs to be faster than smaller ones (read Anands other recent articles if you want to know more).
Conscript - Thursday, August 11, 2011 - link
is there a reason the same products aren't in every graph? Corsair GT seems to be missing from quite a few?Anand Lal Shimpi - Thursday, August 11, 2011 - link
Fixed :)Take care,
Anand
Shadowmaster625 - Thursday, August 11, 2011 - link
Is there a way you can force the drive to run at SATA2 speeds to see if that eliminates the lockups?irev210 - Thursday, August 11, 2011 - link
You open this SandForce article on Intel 320 SSDs firmware bug.I love how the BSOD is a page two reference.
Anand, your OCZ/sandforce bias bleeds through pretty hard. I hope you can be a bit more objective with your reports moving forward.
The speed difference between SSDs at this point is pretty trivial. As you continue to hammer about reliability, you never even reviewed the Samsung 470, rarely talk about the Crucial C300/M4, and Toshiba seems to be an afterthought.
At least tomshardware made an attempt to look at SSD reliability.
Bottom line, it seems like sandforce-driven ssds have the biggest number of issues, yet you still recommend them. You say "well I never really experience the issues" but just because you don't doesn't mean that it is the most reliable drive.
I think you should work a little harder at focusing on reliability studies instead of performance metrics. For most users, it taking 1.53 seconds or 1.54 seconds to open an application is pretty irrelevant if SSD A is 10x more likely to fail over SSD B.