2013 SSD Roadmap

epitaphic

n00b
Joined
Feb 24, 2012
Messages
20
We can readily find out when new stuff is coming out/what to expect for CPUs and GPUs, but when it comes to SSDs things get a bit more nebulous.

Anyone want to take an informed stab when and what we can expect for the next gen of SSDs?

I'm hoping for a Samsung 850 Pro in Q2-3 on a sub 20nm 128Gbit MLC NAND (I'm assuming the M500 is the blueprint to the next gen). Am I far off?
 
You are certainly off on the time frame. I cannot see Samsung replacing the 840 Pro only 6 to 9 months after it came out, when it is selling very well and there is no competitor that is a major challenger (and there is the TLC 840 that is cost-competitive with the M500).

Maybe Samsung will replace the 840 Pro late in the year (that would be about 12 months of life), but even that seems doubtful to me. They may let it run longer. If I were Samsung, I would not replace the 840 Pro until it is time to come out with the next-gen controller (SATA/PCI Express)
 
After some googling seems like the release dates were:

470 ~Sept 27, 2010
830 ~Oct 25, 2011
840pro ~Sept 21,2012

Late Q3/early Q4 it seems, Q2 is definitely too optimistic. Though I'm assuming the barrier to next gen introduction would be the NAND process itself (for which I can't find any dates) and controller optimizations.

Not too sure about them waiting it out till Sata Express comes out, that would be mid-2014 at best with Broadwell [1]. That's an awful long time to stay put. I bet we'll see a SATA-Express 860pro by then.

I'd imagine there's a lot of optimizations left on random read/writes to mitigate the fact that we're basically at the limit of SATA3 on sequential. Besides, that's just Samsung. Surely Intel must have something up its sleeve...
 
I'd imagine there's a lot of optimizations left on random read/writes

There is a lot of room for that since random low queue depth 4K reads are in no way close to SATA III speeds. Although it is not an easy problem to hide the latency of the nand cheaply without risking data loss..
 
Although it is not an easy problem to hide the latency of the nand cheaply without risking data loss..

That is an understatement. The latency of flash is basically staying the same or increasing with shrinks (larger page sizes). The only way to "hide the latency" is a large DRAM cache, a technique that no major SSDs use for data (mostly they use them for metadata) since it does not make a large difference in most cases (cache misses are frequent), and, as you say, it is dangerous as a write cache.

I would be very surprised if there is any significant improvement in 4K performance before SATA/PCI Express. Even then, the improvement probably won't be huge, since the latency in SATA is not that bad.
 
Last edited:
After some googling seems like the release dates were:

470 ~Sept 27, 2010
830 ~Oct 25, 2011
840pro ~Sept 21,2012

The 840 Pro general availability was later than that. More like October.

But just looking at the dates, you miss the fact that, performance wise, the 830 was a large improvement on the 470, and the 840 pro was a significant improvement over the 830. It will not be possible to make any significant performance improvement over the 840 Pro until SATA/PCI Express. 4K speeds are basically limited by the latency of the flash at this point. And Samsung has the TLC 840 for the value market. It just does not make sense for Samsung to release a new model very soon. Crucial waited about 2 years before the M500 (and that is not very impressive, except for the price on the 1TB model, and notice that is the only model that is out of stock most places -- Crucial is probably taking a loss on that, keeping supply low hoping that flash prices will come down), and Intel has not had any groundbreaking consumer SSD releases for more than a year.
 
@ OP, if you haven't noticed, SSD companies are spending more on market saturation and cutting costs than they are improving the products we use. While it's true we've seen some great improvements to consumer-level SSDs in the last 3 years, the speed difference between today's SSDs and Sandforce-era is not huge except in benchmarks. As far as real world performance goes, the performance gains have been minimal while reliability and warranties have improved vastly.

Prices are going down and companies are getting more competitive, so expect more of the same for at least a couple years.
 
I cannot see Samsung replacing the 840 Pro only 6 to 9 months after it came out, when it is selling very well and there is no competitor that is a major challenger.

How is Plextor not a major challenger to the Samsung ssds?
 
How is Plextor not a major challenger to the Samsung ssds?

The Plextor M5ProX is not quite as fast as the Samsung 840 Pro, but more importantly, the M5ProX is not selling as well as the 840 Pro. Why would Samsung consider them a major challenger when the performance is no better and Samsung is outselling them?
 
Why would Samsung consider them a major challenger when the performance is no better and Samsung is outselling them?

Because Plextor often offers better prices on their drives with just as good of a warranty as Samsung does. A value oriented consumer is likely to choose the Plextor if it's cheaper and it usually is. Any how many consumers are really going to be able to tell the Plextor is faster in regular non-benchmark use? I think here price difference will often be a bigger factor than the performance difference which is not much at all.
 
Last edited:
Well, outside of the geek-spectrum, Plextor is as good as a no-name brand. Of course, most non-geeks wouldn't really consider buying an SSD in the first place.
 
Because Plextor often offers better prices on their drives with just as good of a warranty as Samsung does. A value oriented consumer is likely to choose the Plextor if it's cheaper and it usually is.

This has not been my experience -- the prices are close. For example, at amazon at the moment the M5Ps are at $132, $251, and $493, while the 840 Pros are at $139, $238, and $484 (128, 256, 512GB). The price can vary on both models, as they both tend to go on sale occasionally, but overall the prices have been quite close in my experience.

But all of that is beside the point, which I made earlier. Even if the Plextor were cheaper, since the Samsung is outselling the Plextor, the Plextor is not a serious threat to Samsung.
 
This has not been my experience -- the prices are close. For example, at amazon at the moment the M5Ps are at $132, $251, and $493, while the 840 Pros are at $139, $238, and $484 (128, 256, 512GB). The price can vary on both models, as they both tend to go on sale occasionally, but overall the prices have been quite close in my experience.

But all of that is beside the point, which I made earlier. Even if the Plextor were cheaper, since the Samsung is outselling the Plextor, the Plextor is not a serious threat to Samsung.

serious threat or major challenger make up your mind they mean different things
 
serious threat or major challenger make up your mind they mean different things

They mean the same thing to me in this context -- whether it is something that Samsung needs to bring out a new model to compete with.
 
Fast forward today and here we sit in December of 2013 with no Samsung 850 series in sight!

I guess the 840 EVO elongated the product cycle. Perhaps around Q2 of 2014 they'll release another refresh product.

We're all excited about sata express to be released, but not sure if we'll see it until the latter end of 2014, if that.
 
Heh, I just read my prediction again from 7 months ago. It looks to be holding up so far.

Maybe Samsung will replace the 840 Pro late in the year (that would be about 12 months of life), but even that seems doubtful to me. They may let it run longer. If I were Samsung, I would not replace the 840 Pro until it is time to come out with the next-gen controller (SATA/PCI Express)
 
I decided that what I want is to buy an ssd with 750gb-1tb of space. And for now I am waiting for the 850.
 
Back
Top