Slashdot Mirror
Matrix 3D memory is World's Smallest
nokiator writes "Most of the headlines about cool new high density memory technology are from DRAM or Flash manufacturers these days. Matrix Semiconductor, a small Silicon Valley start-up, broke the trend today and announced that the world's smallest 1-Gbit memory chip. Matrix's chip is an antifuse-based one-time programmable ROM. The total die area of the 1Gb chip is 31 square millimeters (smaller than the blue/red pills in the Matrix movie). Matrix claims that they can achieve this density through a proprietary 3D circuit technology that combines 150nm and 130nm process geometries. When Matrix moves to 90nm process technology, it should be possible to manufacture a 8Gb memory chip on a reasonable sized (i.e. cheap) die. There are many potential applications of this kind of low cost, very high density ROM technology, mostly in content distribution area. One 8Gb ROM chip would have sufficient storage capacity to store the contents of an entire movie using H.264 encoding."
...but could this be used for CPU on-die caches, or is it too slow/consumes too much power? I couldn't imagine even having 8MB of cache let alone 8GB. (Which will come to haunt me later like the ol' 640K quote).
Try 10 * (1 byte / 8 bits), or one movie. GB is different from Gb.
tasks(723) drafts(105) languages(484) examples(29106)
I can ramdisk the internet.
Except for the fact that this is ROM. It makes use of antifuse technolgy which works a lot like fuse technology. The idea in fuse technology is that you blow the pathways you don't want, thus creating the circut. With antifuse technology, the fuses don't normally conduct electricity so you have to blow the fuse to create pathways.
Info
Javascript + Nintendo DSi = DSiCade
Secondly it is antifuse-based one-time programmable ROM. It is NOT a flash which can be re-written 100,000 times. So it is more useful for storing application code but not for data storage etc.
Antifuse base memories are diode like and can be much smaller than regular FLASH memories. But these are inherently slower and also don't have any gain element (like transistor). This requires careful design to achieve good signal-to-noise ration for memory read operation
More aggressive 3D technology was demonstrated by IBM last year where they have circuits in 3D.
A startup R-cube logic is also designing 3D microprocessor where memory is put on top of the logic core to reduce latency.
Xanoptics is more into hybrid design (mixed analog, RF, optics) on a single footprint.
8 Gigabit(Gb) == 1 Gigabyte(GB)
Check your math.
A CD/DVD's data area is around 10,000mm^2. That's enough area for 343 of these chips at their current size. 343*128MB=42.875GB
A better arrangement would be to make a 5" (127mm) square cartridge to fit in the same stackable region as a 5" circle. That gives you 520 of these chips. 520*128MB=65GB, which is better than Blu-Ray, and nowhere near as fragile.
And that's on their current process, which is apparently a blend of 130nm and 150nm. Wait until they shrink that down a bit.
Data density in the world has just gone up.
tasks(723) drafts(105) languages(484) examples(29106)
You're on the right track, given this press release, but you misspelled "Nintendo DS".
When the name originated in '99, Matrix consisted mostly of engineers. At the time, there was no budget to pay someone to come up with a name, so the engineers were assigned the task of suggesting names that would somehow reflect the technology. Sadly, none of my suggestions won, and I missed my chance to leave a legacy there. The winning suggestion came from one of the process engineers, who, as far as I can recall, received absolutely no compensation for it. So, there you have it: The name describes the actual structure, was coined by a process engineer, and cost the company precisely $0 to come up with.
Perhaps the biggest advantage would be the ability to treat the array as random access instead of rotational access with the inherent rotational delay.
Nonsense! Memory chips are ALWAYS measured in bits!
Bytes are what do not make sense. The real consumers of these chips---the engineers who design around them---do not think in terms of bytes. They think in terms of bits or oddly-sized cache lines.
So get with the program. Bytes are such an arbitrary '70s concept. What's so magic about groups of 8 anyway?
You're not trying to say that the speed of light has ANYTHING to do with the speed at which a beam of light can be swept across a medium, are you?
A point of light can be moved as fast or as slow as you want it to be. Aim a laser at the moon, now sweep it across the moon as fast as you want. Poof! The "spot" of light just moved across the moon at 18 times the speed of light... no problem.
No, the real limitation to the speed at which light can be moved along a medium has more to do with how long the spot of light must be focused on each point on the medium for enough photons to reflect back to be read. The faster you sweep your laser across a surface, the more dimly that surface is illuminated.
(At the limit, you're moving the spot of light faster than the rate photons are being emitted, although at 1 Watt that's something like 10^20 photons per second. If you need at least 10 photons per nanometer, say, you can do the math to find your maximum speed.)
INsigNIFICANT
Just wondering, were you thinking when you said that light is exponentially faster?
It is faster, but where have you pulled this exponential crap out of? You can have an exponential relationship between two variables. The speed of light vs the linear speed of a DVD has SFA to do exponents. It makes you sound like you have no clue.
Big number =! exponential.
Bytes are not always the same size; bits are.
Whence? Hence. Whither? Thither.
If it wasn't the cost, one could make a pretty high speed optical reader by creating 12cm x 12cm CCD chip capable of reading every single pit in the surface of DVD disk at once. It would be just like a flat bed scanner, just faster. Scanning the whole disk would take 1/100th a second at most. Then it would be just matter of moving bits from the CCD chip to system memory fast enough. I won't even try to guess how much a 12cm x 12cm CCD chip with enough resolution to contact read a DVD disc would cost.
_________________________
Spelling and grammar mistakes left as an exercise for the reader.