SandForce SF-2000 Series SSD Controllers Detailed

[Apoptosis]

SandForce SF-2000 Series SSD Controllers Detailed

SandForce is making some news today by announcing the availability of the SF-2000 Family of SSD controllers. This new SSD controller features a SATA III 6Gbps host interface, DuraClass Technology, 60,000 sustained random read/write IOPS (Input-output Operations Per Second) and sustained sequential read/write performance of 500 Megabytes per second. Yes, you are reading that right! This new controller will be able to max out the Serial ATA Revision 3.0 bus at 500MB/s read/write!

The key changes in the SF-2000 series controller is most obviously the updated SATA III 6Gb/s interface, which helps push performance up to 60,000 IOPS for 4K Random Read and Write data transfers. It also pushed the sequential Read and Write performance numbers up to 500MB/s and SandForce confirms they have hit those numbers in their test lab just this week with their first production run silicon that they just recently got in from overseas. SandForce will have three model numbers in the SF-2000 series; the SF-2600, SF-2500 and the SF-2300.

Article Title: SandForce SF-2000 Series SSD Controllers Detailed
Article URL: http://www.legitreviews.com/article/1429/1/

[Major_A]

I guess with the negative press concerning the firmware issue they decided to come right out and keep everyone informed. Good on you SandForce for learning from past mistakes.

I know 500 MB/s is theoretical performance, but if it gets close to those numbers, DAMN!

[Sttm]

I cant wait to see the performance of a Revodrive with those in it. 1 GB/s?

I was bummed when I couldnt build a new pc this summer, but next year is looking to be an awesome time to upgrade. Zambezi or Sandybridge, UEFI, 500mb/s Solid state drives at under $1 per gig; AMD 6000 series, or possibly Fermi Successor.

[d*edward]

Fast Forward From Initial Glowing Intro of Sandforce 2000 Series Controllers TO PRESENT DEBACLE Late May-Mid June 2011 and growing purchaser/user complaint re: Corsair and OCZ Vertex3/Vertex3 Max IOPS ssd's - BSOD's, dropping out of mobo bios and sata3 interface issues...

Corsair immediately acted to recall ssd's containing Sandforce Series 2000 controllers, while OCZ has simply stonewalled by issuing several firmware revisions, several toolbox revisions and generally blaming sata3 cables, sata3 mobo controllers and user inability to install OCZ ssd's correctly. Leading to growing user frustration with an expensive non-functional ssd investment they can't rma to OCZ.

We are experiencing a replay of the intro of the original Sandforce Series 1000 ssd controllers with the same player, OCZ, stonewalling and refusing RMA requests... When Will Sandforce recall the Series 2000 Controller Chips and provide users with the correct customer support, so lacking in the past?

When Will OCZ Act In A Responsible Manner, As Corsair Has Acted, And Recall Their Defective Vertex3 & Vertex3 Max IOPS SSD's?

This matter will not simply disappear...

[Apoptosis]

The Corsair and OCZ issues are totally different from what I have been told. Corsair and OCZ share the same firmware issue, but the Corsair recall was due to a resistor not being installed on the PCB that needed to be. Corsair and OCZ are using different companies for the PCB, so like I said only Corsair needed to do a recall. This what really forced them to recall the drives as the PCB's needed to be re-worked.

[d*edward]

the Sandforce Series 2000 "teething problems" appear abit excessive even for bleeding edge...

http://forums.anandtech.com/showthread.php?t=2172998

pls see post #13, perhaps Sandforce Business Plan relies too much on hope...

[d*edward]

Apoptosis, your kind thoughts re:

Sandforce on behalf of OCZ has incorporated "Thottling" in OCZ Firmware update FW 2.09, applications include Vertex 3 ssd's and Vertex 3 Max IOPS ssd's, in a last attempt to minimize the numbers and frequency of Stuttering and BSOD Events being experienced by users.

First Sandforce SF1200, Now recently intro'd SF2281 SSD/Nand Controller THOTTLED to reduce instability...

Different pcb's having same smt SF2281 are having same issues. Either Sandforce is competent to design LSI or SF has failed to introduce a FULLY FUNCTIONAL Next Generation SSD/Nand Controller. And Sata 2.6 (Sata3 6.0Gbps) was ratified over two years ago.

Sandforce is proving very disappointing...

[d*edward]

apoptosis,

http://techinsidr.com pls view "Is A Recall Brewing At OCZ?"

The 3rd gen Sandforce SF-2281 Controllers are unstable, unless user "castrates" by updating firmware to "Thottle" controller chip...

How can you call this "Progress"!

[d*edward]

apoptosis,

http://www.theregister.co.uk/2011/06/22 ... lash_bsod/ pls view...

the vertex3/vertex3max problems seem to follow ocz in all regional markets...

AND SANDFORCE CAN "ONLY" IMPOSE THOTTLING AS "Non-Solution"... First SF-100-, NOW SF-2281 !

[Apoptosis]

apoptosis,

http://www.theregister.co.uk/2011/06/22 ... lash_bsod/ pls view...

the vertex3/vertex3max problems seem to follow ocz in all regional markets...

AND SANDFORCE CAN "ONLY" IMPOSE THOTTLING AS "Non-Solution"... First SF-100-, NOW SF-2281 !

Did the new patch work that came out two days ago?

[d*edward]

Apoptosis,

The patch, Firmware 2.09, written by Sandforce on behalf of OCZ and posted on OCZ Forums page, appears to resolve 80-90% of user issues with Sata2 (3.0Gbps) reducing Stuttering/BSOD's, But Does Not Greatly Reduce user issues with Sata3 (6.0Gbps) interfaces...

The Vertex3 / Vertex3 Max IOPS SSD's Are Principally Marketed To Users Who EXPECT Sata3 (6.0Gbps) Compatibility AND Performance Levels...

Even OCZ Forum Moderator Tony-the-Tiger has v. recently expressed "Everything is Too Fast..." when discussing Sata3 (6.0Gbps) Stuttering/BSOD User Events.

This is a most unfortunate comment by OCZ spokeperson on the stability of OCZ's Premium SSD Product Line.

I continue to believe that there are inherent sf-2281 stability and sata3 (6.0Gbps) compatibility issues entirely outstanding and unresolved, beyond numerous firmware re-writes/updates/patches...

I Would Never Purchase a SSD Spec'd at Sata3 (6.0Gbps) THAT ONLY FUNCTIONS WITHOUT Stuttering/BSOD's on Sata2 (3.0Gbps) -- Something Is Extremely Wrong...

[FZ1]

What are the new speeds with the updated firmware?

Anand's article notes that he feels this is a very limited issue. I have not had a single issue like this with any of the OCZ and/or SandForce drives. I've been using the Agility 3 drive full-time for a few weeks now with no issues at all.

You seem to be railing against OCZ specifically but reference that this is a SandForce issue. A lot of companies are using SF drives. Do you own an OCZ SF-2xxx drive?

OCZ has a pretty large SSD market share so of course, if there is a problem with the SandForce controller, it will have a larger impact on OCZ than the others. From what I've seen OCZ is also a little more transparent and has more support on their forums than a lot of others so it may appear that it's a larger problem with them. Corsair did a recall but that was for a different issue as Nate mentioned. From what I see, OCZ is actively working the issue but if it's as infrequent as they say and potentially tied to specific platforms, it may take them a while to chase it down. Companies generally will not recall all units for a problem yet to be fully identified when large numbers of units have been sold and the scope of impact appears to be relatively small. Plus, they need some level of support from SandForce to work this out and who knows how much cooperation they are getting.

I'm not sticking up for OCZ per se, they have had their share of problems, but we don't know what's going on behind the scenes. Until this is sorted out, I'm reserving judgment on the issue and their actions.

[pwcmed]

OCZ puts a target on there back IMO.

From the 25nm fiasco to the financial irregularities, this company is rotten. I really do not understand why reviewers do not warn buyers with the bad reputation that this company has.

[Apoptosis]

OCZ puts a target on there back IMO.

From the 25nm fiasco to the financial irregularities, this company is rotten. I really do not understand why reviewers do not warn buyers with the bad reputation that this company has.

they used to have a really bad reputation, but cleaned it up back in 2003-2005. They are still headed in the right direction with their products and are focused on SSDs. The 25nm fiasco wasn't great, but they refunded money and swapped drives for those that complained right? Sounds like a good solution. As for this Sandforce issue... Like Joe said this is a SandForce issue not OCZ they just by parts and either assemble the PCB or buy it from another company and slap their logo on it. SandForce makes all the firmaware and controls that aspect. With dozens of companies using the same controllers I think you are just seeing who sells the most drives... OCZ is hands down the biggest seller of SandForce products and we all know those that have issues take time to post around the web and those that happy don't.

Kingston Technology doesn't play around and when they release the HyperX SATA III drive with the SandForce SF-2000 series controller it will be interesting to see what happens. Kingston is built on quality and revenue margins, so if there is an issue it will be brought to light shortly. No one has pull in the Flash industry like Kingston.

[d*edward]

Joe, Firmware 2.09 Thottled the SF-2281 to reduce instability (stuttering/bsod's) experienced by range of users having good representative cross section of motherboards, various generations of intel chipsets, various 3rd party (ie: Marvell & other) raid controllers and I estimate 70% native sata2 (3.0Gbps) and 30% native sata3 (6.0Gbps) - all this represents 4-7 weeks realworld global USER experience with Vertex3 & recently vertex3 max iops as posted on over 38 forums globally... OCZ Forums represents 50% of all forum postings and frankly the speed with which OCZ, posted for download, several Sandforce written Firmware Updates strongly demonstrates a significant number of v3/v3max user complaints.

I attempted to install v3max boot drive & (4) v3max raid array to Asus Maximus IV Extreme, i7-2600K & Asus GTX580 (both cpu & gpu water cooled) - I originally purchased (11) v3max having original f.w. & updated (2) times - after several days of w7 os reinstalls, driver updates, bios default changes, I stopped and returned ALL ocz ssd's for full refund. During this time there were NO overclocking attempts - although I absolutely will o.c. this motherboard build for actual daily use...

Please consider that imposing Thottling on the SF-2281 immediately negates the highly publicized core specifications and invalidates the hype regarding sata3 (6.0Gbps) design spec; as this has still NOT resolved stutter/bsod events on motherboards in realworld having native support and default function of sata3 interfaces - the intel native sata3 experiences enough problems, the marvell sata3 doesn't work.

And these issues are not limited to OCZ brand ssd's having Sandforce SF-2281 controllers, but also others, albiet in proportionally smaller numbers.

Candidly the Sandforce SF-2281 is not supporting Sata3 interfaces UNDER REALWORLD CONDITIONS EXPERIENCED BY SSD USERS.

This is not "railing", this is reality and tolerance of occasional ssd stuttering/BSOD's is complete denial - the Sandforce SF-2281 SSD/NAND Controller is Either Stable or it is Unstable AND UNUSABLE, EXCEPT FOR BENCHMARKS. The stuttering/BSOD's are random events occuring on various motherboards.

OCZ is aggressively marketing Enterprise SSD's, so OCZ certainly understands the meaning of "Stability" and "Interface Compatibility".


Thanks for considering my comments.

[FZ1]

When you say the firmware is throttled - what are the new read/write #'s? I don't believe that they would be anywhere near SATA II levels.

[pwcmed]

The 25nm fiasco wasn't great, but they refunded money and swapped drives for those that complained right? Sounds like a good solution.

They only did this when people were noticing the performance wasn't the same and raised the BS flag. They were going to charge people the difference in price if they wanted to swap it out until people started complaining.

Mark my words, OCZ will not last and will go under.

[d*edward]

FZ1, in addition to Anand's review of Vertex3 Max IOPS & Patriot SSD's, there are now (40) comments, none of which are mine. Pls view the comments.

Raw "performance" is not heart of this problem, rather it is (1) Sata3 Genuine Compatibility & (2) Random SSD Stuttering/BSOD Events Strongly Describing Controller Chip Instability.

The Intel 510 SSD functions without issues on an Intel P67 Chipset Native Sata3 (6.0Gbps), remains stable over days & weeks but is "slower" than Vertex3/Vertex3Max. This PC is completely functional and productive.

The OCZ Vertex3Max SSD has random issues (sometimes frequently, sometimes over 18-24hrs) on an Intel P67 Chipset Native Sata3 (6.0Gbps), demonstrations instability but is measurably "faster" than Intel 510 SSD. This PC is unpredictable and difficult to use in productive manner.

Which PC will you chose for your personal use?

[Apoptosis]

d*edward - I contacted a few people at SandForce and they are working on a public statement. Like a few people here have stated this is a SandForce issue and not just OCZ.

[Kaos Kid]

d*edward - I contacted a few people at SandForce and they are working on a public statement. Like a few people here have stated this is a SandForce issue and not just OCZ.

+1