memristor coming soon!

[mike123abc]

DailyTech - HP to Deploy Memristor Powered SSD Replacement Within 18 Months

He comments, "We’re planning to put a replacement chip on the market to go up against flash within a year and a half and we also intend to have an SSD replacement available in a year and a half. In 2014 possibly, or certainly by 2015, we will have a competitor for DRAM and then we’ll replace SRAM. Flash is a done deal, now we’re going after DRAM, and we think we can do two orders of magnitude improvement in terms of switching energy per bit."

This is very exciting. They have previously showed early prototypes running 1/10 the speed of DRAM. They obviously must be about DRAM speed now since they announced they are going to have a DRAM repacement in a couple years.

This could kill off flash memory very quickly. Flash has physics working against it - making the components smaller makes them wear out faster.

HP has shown 100 gigabits/cm2. But, they can stack layers to have 1 petabit per cubic centimenter!

Imagine a thumb drive with 125 Terrabytes of data with access faster than DRAM.

Imagine how disruptive this will be the tablet/smartphone/etc market. They can put a terrabyte on the CPU, no external bus, no DRAM or flash needed any more...

The plan is to license this technology to anyone who wants it, and we'll teach them how to make it. But you'll have to stand in line, we have a bunch of people queued for it. We're doing this because, frankly, we didn't see a hell of a lot of innovation happening out there."

Now if they could make them clear glass, they could live up to STTNG's isoliner memory modules they are always playing with...

 

[John Kotches]

Without a cost per gb of storage estimate it's tough to say how it will do.

Hard disks go for about $0.03 - 0.05 /gb, for consumer grade drives. Flash is $1 - 2 for consumer grade and RAM about $6-7 / gb.

To effectively replace hard drives it'll have to be cheaper than flash and no more than double hard drive cost per gb.



Sent from my MB855 using Tapatalk

 

[navychop]

And sadly, HP management has not been impressing anyone lately.

 

[mike123abc]

If they manage to replace the flash market with memresistors, they will end up looking like visionaries. Since they appear to want to license the technology to anyone that wants to pay $$, perhaps they will make a fortune.

If this technology pans out it could revolutionize the computer industry. If you had 10 TB of non volotile memory that was as fast as SD RAM and cost less than an hard drive, and used so much less energy, think of the applications. The only barrier would become the internet connection.

I hope it does work as well as they advertize.

 

[John Kotches]

Mike,

In all honesty, it doesn't have to perform on par with DRAM. If it's an order of magnitude or two faster than SSD it's fast enough.

At the microsecond level it will be imperceptibly fast for users. It will have to be 5 orders of magnitude faster than SSD to hit DRAM speeds.



Sent from my MB855 using Tapatalk

 

[John Kotches]

Duplicate, sorry.



Sent from my MB855 using Tapatalk

 

[mike123abc]

Mike,

In all honesty, it doesn't have to perform on par with DRAM. If it's an order of magnitude or two faster than SSD it's fast enough.

At the microsecond level it will be imperceptibly fast for users. It will have to be 5 orders of magnitude faster than SSD to hit DRAM speeds.



Sent from my MB855 using Tapatalk

That is true, but HP is talking about not only replacing DRAM but also SRAM. So, they are looking at speeds (at least in the lab HP has it working faster than 1ns) faster than DRAM/SRAM. They initially will use it as a replacement for SSDs, but eventually want to be able to layer it directly on the CPU and have a single package have all the storage needed. This would really help the smartphone/tablet industry. A smartphone could practically be one chip and the rest battery.

It really would change the dynamic and the way of thinking about programming. Essentially one would write programs that all work in memory without file storage since memory is now the file storage. The concept of saving work to a file could change for many types of applications. The concept of having to work around the fact that storage is slow has dictated computer programming since punchcards. Essentially an entire new O/S would be needed to optimize the new environment. Running a program is essentially now just like a subroutine call, everything in memory. You have not run a program in years? No problem it still is sitting in memory... all its "files" (at least by the old concept) are open and loaded in memory, just like you left it...

I hope this technology does live up to HP's hype. If they get this on the CPU and replace DRAM (and possibly even CPU L2/L3 caches) and drive storage the entire software industry could change.

 

[John Kotches]

Mike:

Unfortunately at this point it's promising but hype.

There will still be a file system, it will be quite fast though. I can see this leading to newer file systems though. And please s don't encourage even more inefficient coding from programmers.

I would not want storage on the chip unless there was an effective method of mirroring it. Lose the chip, lose the storage.



Sent from my MB855 using Tapatalk

 

[navychop]

Somehow, Microsoft would still manage to keep Windows taking minutes to boot up.

 

[Ramy]

This is very interesting. Having memory available on the die would be great.