LSI SandForce Next Gen Flash Storage Processors Appear in Taipei – Computex 2013 Update

Perhaps one of the most frequent SSD questions we get these days is where has LSI SandForce been with their new generation of controllers, or as they like it termed, ‘flash storage processors’.  The SF-2000 has had a very successful run in the SATA 3 world and built on the success of the SF-1000.  Since it’s release however, we have seen SATA 3 controllers such as the Marvell 88SS9187, LAMD 87800, Samsung 3-core MDX and most recently the OCZ Indilinx Barefoot 3  BF3-M10.  Let’s face it; LSI needs to step up to the plate.

Concurrently, we are also seeing a bit of the LSI reputation where there is complete silence with respect to their next offering at a time when most might be making that impending announcement of things to come.  Having just received confirmation that ADATA’s new generation of  enterprise SSDs will be released in late Q3 or early Q4 2013, we would only expect some good news or at least a decent leak by now.  The success of countless confidentiality agreements between 50 plus companies who are most likely already sampling the new FSP is pretty impressive.

And then there is the question of product branding as even we predicted that, sooner or later, LSI might be considering rebranding away from the tried and true SandForce name and considering something more in line with…say the LSI Nytro family.  Our latest information seems to reflect just that and, in fact, we may even attribute a bit of this delay to the fact that LSI is still considering its naming convention for this new FSP, this being the first created solely under the LSI umbrella.

ADATA has this week announced three new enterprise SSDs and they do contain the new LSI SandForce flash storage processor, regardless of how little attention they wish to pay to it.  The SSDs are branded as the SX1000, SX1000L and the SX2000 which we have reported on in detail earlier today with mind boggling speeds of 1.8GB/s (yes thats gigabytes per second) and 200,000 IOPS at low 4k read disk access.  All three versions are available in 100, 200 and 400GB capacities while the SX2000 adds 800 and 1600GB versions to its portfolio.  The capacities account for 28% over provisioning to enhance performance and drive endurance which is typical of many enterprise SSDs.

The key factor in determining that these three SSDs are not SF-2000 series SSDs (and we could be wrong) is that their IOPS shows read strength over write strength, the latter being common in all previous iterations of the SF-1000 and SF-2000 families.  Whereas low 4k write IOPS were the high point prior, low 4K read IOPS are highest with these SSDs.  To add to this, none of these SSDs identify the processor specifically which is definitely not the norm with SF SSDs.  The unknowns open up a whole new thought process with respect to the new family of LSI FSP’s.  One consideration definitely has to be whether they have an entire new recipe in store or have changed ingredients, remaining with their traditional drive compression to maintain low write amplification.

The SX1000 and SX1000L versions list read and write performance of 550/500MB/s with 75,000 IOPS read and 45,000 IOPS write at low 4k random disk access.  Each speaks to TRIM and NCQ support, along with lower latency, however drive encryption is not mentioned whatsoever which might be another indicator of a change in recipe. In any case, kudos to ADATA for displaying these drives as we haven’t found any similar from the dozens of SSD companies present at Computex thus far.