The SSD Review The Worlds Dedicated SSD Education and Review Resource | 
When SandForce emerged from stealth mode in 2009, the SSD industry took on an entirely new look. SandForce was quick, aggressive and as much as many might think they may have introduced things just a bit too early, they played a very large part in shaping the SSD industry into what it is today. They inspired competitiveness and not many would argue that the industry might not be as far ahead as we are today without SandForce.
Our meeting with Seagate this year at Computex 2015 showed SandForce in a bit of a different light. The ‘quick nature’ demeanor was replaced by that of a more deliberate, steady and surefooted attitude. Seagate has thought out their direction for SandForce long and hard, moving the SF3700 into the enterprise stream while taking this opportunity to introduce their SF3500 as their first step into the client SSD space.
This is the Seagate SandForce SF3500 controller situated in a 80mm M.2 form factor and, in this shape, it can take on the direction of either SATA or PCIe Gen2 x2. This is a 4 channel controller, capable of supporting MLC, TLC as well as the newest 3D NAND flash memory, and as well, it has dual AES-256 Bit encryption, is PCG Opal 2.0 and IEEE-1667 qualified and also fits into Windows eDrive like a glove.
As a SATA SSD, the SF3500 is capable of 550MB/s read and 450MB/s write with up to 100K IOPS. When we use this controller as a PCIe Gen2 X2 device, however, performance jumps to 900MB/s read and 525MB/s write with up to 130K IOPS. Better yet, this controller is AHCI and NVMe compatible and supports SSD capacities up to 1TB. Perhaps even more interesting was this video demonstration of the SF3500:
This demonstration displayed a single SF3500 PCIe M.2 SSD using the latest Micron 15nm TLC NAND flash memory and playing nine movies concurrently. Each video was as smooth as you might expect watching them at the theater, or on TV.
Looking back, we would be just a bit careless to not update you on SF3700 progress. The SF3700 was on display and is now being marketed as an enterprise level NVMe controller. In this example, we see the SF-3700 situated on a typical notebook form factor PCB, but with the SFF-8639 connector as it is a PCIe Gen2 x4 NVMe solution:
As well, Seagate had the SF3700 on display and in operation, but still sporting the all too familiar orange heat sink.
With word of the SF3700 having been lingering for so long, will we see it on the streets anytime soon? Absolutely…and if I was a betting man I might be looking at the Flash Memory Summit this fall for some big Seagate moves. Would we be safe in saying that the SF3700 might have been announced just a bit too soon? Yes.
We are still very interested in observing the final placement of the NVMe SF3700 at the end of the day as, although they are now listing it as an enterprise solution, the end manufacturer will determine its best sales position. All in all though, our visit to Seagate was very encouraging and we are certain that they, and the SandForce name, will be a force to recon with once again…and maybe even by years end.
Given that most announced Skylake MBs are coming out with PCIe Gen3 x4 M2 slots, why is Sandforce coming out with a new “SATA or PCIe Gen2 x2” controller aimed at the M2 market?
its a 2013-14 tech, which took long time to market. next gen will be derivative of this, so lot was/is riding on this
The problem is that they really needed that “next gen” now, if they wanted to catch the Skylake M2 wave. Sandforce is clearly losing the technological “Red Queen’s race” (where you need to innovate merely to survive, let alone succeed).
Running that route would have been the SandForce we all knew a few years back. SandForce needed stability after being bought by LSI, Avago and now Seagate. Seagate has a very large client base as well as many other relationships that have been honed throughout the years. What they will bring to SandForce is a strong validation process and then guaranteed client base, even before we start seeing their product in retail sales. Heck, all that Seagate needs now is to develop their own NAND flash memory…or purchase a NAND manufacturer. No one doubts they are capable of such.
My suspicion is that the SATA SSD market is already largely commoditised — decent stability is assumed rather than rewarded (with any blatant lack being punished) and margins are thin (and becoming razor thin). Seagate’s contacts may get them some big (e.g. OEM or corporate) orders, but only at the cost of cutting the margins even finer. Large clients also have large rolodexes of suppliers prepared to underbid each other. Having delayed their launch this long, will they recoup their R&D costs? I somewhat doubt it.
Gen2x2 does not offer much of a real-world benefit over SATA, so what little market there is for it is likely to evaporate when Gen3x4 becomes mainstream. This process will also further fuel the commoditisation of SATA-SSD, as premium products migrate to the faster technology.
You only get to make a substantial profit by offering something that nobody else (or only a few others) have got. That’s basic economics. “A very large client base” gets your foot in the door, it does not however get you a sale unless the price is right.
Because it’s not a consumer controller.
I would be very surprised if many (any) enterprise products would use either “SATA or PCIe Gen2 x2” — older ones would use SAS, newer would use PCIe with far more lanes than 2.
“playing nine consecutive movies.” Wouldn’t that be “simultaneous”?
Thanks… I was think concurrent!!!