Future SSDs will get a serious speed boost thanks to SATA Express, boasting 2TB/s peak transfer rates.
Memory market watcher DRAMeXchange has published a list of predictions which will be welcomed by those planning on upgrading to solid-state storage: things are set to cheaper and faster in the none-too-distant future.
According to the group's research, mass production of latest-process NAND flash in the second half of this year holds the promise of seeing unit costs drop below $1. As a result, the cost of producing high-capacity solid-state storage devices will drop drastically.
This ties in nicely with another of the group's predictions: growing numbers of ultra-portable laptops built to Intel's Ultrabook specifications will mean an explosion in demand. According to DRAMeXchange, the 'mainstream' capacity standard will increase from the current 64GB to 128GB, while increasing numbers of manufacturers will make the jump from spinning-disk or hybrid storage systems to SSDs.
That, coupled with the lower price points, is predicted to drive tripled shipments of solid-state drives in 2012 compared to the previous year. More shipments mean more volume; more volume means that margins can be trimmed to drive down the cost of SSDs still further.
DRAMeXchange also explains that the current move to the SATA 6Gb/s interface standard is a temporary stop-gap system while the SATA-IO working group refines its SATA Express standard. Designed to combine the software infrastructure of SATA connectivity with the electrical transfer interface of PCI Express, the SATA Express standard promises peak transfer rates of up to 16Gb/s compared to the 6Gb/s offered by standard SATA. (Note: that's 2GB/s, not the 2TB/s originally claimed by DRAMeXchange.)
This is key to unlocking the true performance of NAND flash, the group claims. As technologies including the Open NAND Flash Interface (ONFI) 3.0 and toggle DDR2 enter mass production, the bottleneck for transfer speeds will shift from the NAND flash itself to the controller. Without SATA Express, near-future SSDs are likely to rapidly saturate the SATA bus causing an unavoidable performance barrier.
Sadly, it's likely to be a while before we're upgrading to the new standard. SATA Express has yet to be ratified, while DRAMeXchange predicts that commercial implementations won't appear until 2014 at the absolute earliest.
Previous Article
Next Article
17 Comments
Discuss in the forums Replyin raid0....
My pr0n will load in 0.00003e-12m/s
dunx
Unless we're going to start getting >10TB drives, it seems a bit pointless.
2TB/s is fast enough to use a harddrive as RAM. Not that SSDs are fast enough for that yet, but it paves the way for a system design that doesn't need RAM any more.
Read the manual? PFFT, what do you think I am, someone with TWO X chromosomes?
Why do people *always* say stuff along the lines of "what's the point of having something faster than I can use today"? Well silly, so you can create NEW use cases if you have hardware more capable than you need today. I mean who look back to 1995 or so and Myst - today we can freely move around in an environment, prettier than that, in real time. But out current hardware is laughably overkill for running Myst today!
Bring on the development I say - keep the competition high and we'll keep getting better and better computing at cheaper and cheaper prices!
Current RAM speeds are roughly 20-40GB/sec compared to SATA6Gbps, which is roughly .6GB/sec max real transfer rate.
So RAM is still close to 100x faster than the fastest SATA interface. NAND flash and other chip based storage doesn't look like they'll be fast enough to be on par with RAM anytime soon, but at the same time opening it up for years and years of future proofing is not a bad thing. PCI-e electrical interconnect might also reduce latency some, though most is related to the NAND itself, which is still massively slower than DRAM.
However, it doesn't hurt to be able to push a GB/sec or two. Nothing wrong with loading up Windows in 3-5 seconds instead of 20-30 seconds of a fastish SSD now. Speeds up enterprise stuff too. In portable stuff it'd reduce power consumption as well as the processor could drop back to idle faster instead of being at a higher power state processing data as it slowly comes in from mass storage instead of being able to hammer through it as it floods in and then go back to sleep.
A 30s load time for an app wastes a lot more power than running the CPU at a somewhat higher load and being able to pull it from main storage in 4-8s.
What do you consider loading windows?
Not including POST, my SSD loads a ready to use win 7 in less than 10 seconds now
Most test show we will need many software adjustment to take advantage of these speedy storage.
If it's Tb/sec, then 2000/6 = 333 times faster than today's standard
If it's TB/sec, then (2000*8)/6 = 2666 times faster than today
I think there must be a typo in the press release or something. They could mean 2GB/sec, as in 16Gbit, which would be plausible!
I'll stick a correction in.
forgot to correct that for you before
Sata 6 is 600MB/s not 6GB/s, around 550MB/s useable
DDR3 is more like 15GB/s ish or 15000 Ish MB/s,
You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.