Print 40 comment(s) - last by david99.. on Apr 16 at 11:17 AM

Samsung Semiconductor becomes the first company to announced 3D memory packaging

Samsung just announced a new method to pack more memory into small spaces -- three-dimensional chip packaging.  The technique works by adding "through silicon" interconnects on the memory module and then daisy chaining multiple modules through the interconnects.  Samsung has dubbed this new technology "wafer-level stack process" or WSP.

Existing semiconductor packaging relies on wire bonding to a printed circuit board (PCB).  The wire bonding requires space between the interconnects to eliminate interference, but ultimately becomes the limiting factor when attempting to create high density memory. "Through silicon" interconnects are essentially laser cut holes between the memory dice.  The holes are later filled-in with a conductive material creating a vertical interconnect.

Samsung researchers managed to stack eight 2Gb NAND chips onto one package.  The result is a 16Gb NAND chip that is just over half a milimeter in height.  The same technology will also be used for DRAM later this year and multimedia controllers.  Cell phone, PDA and high density server components are all the likely candidates for this new process.  Samsung's newest NAND hard drive, announced a few weeks ago, would only be eight millimeters high if the WSP package allowed for all 256 modules to stack on the same packaging. 

However, 3D packaging isn't the best route for chip assembly.  On relatively slow NAND modules, the thermal envelope is not a huge factor.  High speed DRAM, on the other hand, has much higher operating temperatures and will not likely adopt Samsung's WSP or other 3D packaging in the near future.

Comments     Threshold

This article is over a month old, voting and posting comments is disabled

By kalaap on 4/13/2006 10:34:21 AM , Rating: 1
"wafter-level stack process"

Whats a wafter?

By Lonyo on 4/13/2006 10:56:14 AM , Rating: 1
And what's a memory dice?

By ShadowD on 4/13/06, Rating: -1
By elT on 4/13/2006 11:54:43 AM , Rating: 2
Stop moaning already. People who actually work make mistakes in the process. You should be all b***slaped, Rick James style, for this constant moaning. It's a simply typo....what's the fuss?!

By Exodus220 on 4/13/06, Rating: -1
By Exodus220 on 4/13/2006 12:19:14 PM , Rating: 2
They are just lucky this is submitted online because they have the ability to edit after submission. If this were printed material then they would not have a way to correct their errors and that would demonstrate to me that they do not proof read their material before submitting it.

By elT on 4/13/2006 5:00:31 PM , Rating: 2
To constantly critcize others for minor mistakes that have no real impact on the substance is to escape own inperfections and ignore working on them. Finding the resort in others mistakes from your own is a big no no, no matter by what exponent your manhood might grow.

If you were really that bothered by it, why not write a nice, advice full e-mail to the editors and point out. But no, it won't be in public where your ever endagered manhood, by your own insecurities, can be saved.


By Spinne on 4/13/2006 6:04:35 PM , Rating: 2
I think they report it publicly because writing personal email to the editor because of a spelling typo is too big a hassle. I doubt if they have ill-intentions towards DT or they probabl y wouldn't be here.

By elT on 4/13/2006 6:18:19 PM , Rating: 2
I never said they had ill intentions. But the fuss about nothing even remotly important doesn't sound to healthy to me either.

By elT on 4/13/2006 6:20:13 PM , Rating: 2
I suppose I should be banned for my splezling miztakefs.

By Knish on 4/13/2006 8:01:13 PM , Rating: 2
"They are just lucky this is submitted online because they have the ability to edit after submission. If this were printed material then they would not have a way to correct their errors and that would demonstrate to me that they do not proof read their material before submitting it. "

If this were printed material you wouldn't have heard about it yet.

By oTAL on 4/14/2006 1:42:53 PM , Rating: 2
"dice" is one possible plural for die.

By Eskimo on 4/13/2006 12:10:04 PM , Rating: 2
Dice is plural for die, a common term used in the industry to refer to a single chip on a wafer.

By PrinceGaz on 4/14/2006 5:04:35 AM , Rating: 2
Wrong. Dice is plural for die, but only ever means a cube (or other polyhedron) used in gaming. Dies is also plural for die and is the correct word to use when referring to multiple chips on a wafer.

By Griswold on 4/15/2006 2:50:14 PM , Rating: 2
You are both wrong and right.

Dice is a polyhedron used in gaming. But Dice is also the term used by people in industry to describe the plural of Die.

By david99 on 4/16/2006 10:47:12 AM , Rating: 2
LOL well if people are going to be daft about this
and not discuss the artical
"DICE - The Amiga/embedded/generic 68000 C compiler"

so it seems while this is a 'good thing!" its not
going to help make more compact faster CPU's, GPU's or
PPU's any time soon ?.

By The Boston Dangler on 4/13/2006 7:29:19 PM , Rating: 2
If you had one memory die, and I gave you another memory die, you would have, take a guess, cmon, yes, two memory dice.

By PrinceGaz on 4/14/2006 5:09:45 AM , Rating: 2
No you would not have two memory dice, you would have two memory dies.

Unless you painted a number on each side of the two memory dies and rolled them in games of chance, in which case I suppose you would have two "memory" dice: dice that are made from memory chips :)

By InternetGeek on 4/14/2006 3:16:18 PM , Rating: 2
Guys, go check your dictionary. Either 'dies' or 'dice' are the plural form for 'die'.

So now that we can get over the syntax issue can we start having a thoughtful conversation about these development?

By InternetGeek on 4/14/2006 3:17:23 PM , Rating: 2

(We need an edit option)

By david99 on 4/16/2006 11:02:30 AM , Rating: 2
"By The Boston Dangler on 4/13/2006 7:29:19 PM , Rating: 2

If you had one memory die, and I gave you another memory die, you would have, take a guess, cmon, yes, two memory dice"

well if the other 3 guys all put in one due each
and i put in a 'B' and my die to the pot then would that
mean that this artical would " 'B' 2 die 4 " LOL :)

its about time people started having some fun around here

By david99 on 4/16/2006 11:03:47 AM , Rating: 2
that edit option would be so nice about now

By Googer on 4/13/2006 3:55:14 PM , Rating: 2
This would be cool if AMD did this with Athlon X2 CPU's. We would get four cores and quad-channel memory controllers, that would traslate in to not having to deal with 2T memory timings and more bandwith.

By PrinceGaz on 4/14/2006 5:11:32 AM , Rating: 2
You might want to read the last paragraph of the the article where it talks about heat.

Price Reduction?
By Mclendo06 on 4/13/2006 11:06:05 AM , Rating: 2
Am I reading correctly that this could result in much higer memory densities per silicon, sort of akin to reducing the on-die size of transisitors and such - only stacking them instead? If someone is reading this differently, speak up, please. If this is the case, it could cause the price/capacity of NAND to drop a lot faster than it has been. Hard drives indeed may be on the way out, although that will still take a good while. It all depends on how the price/capacity of NAND compares to HDDs. Right now NAND's only about a couple of orders of magnitude behind. This could certainly help it catch up, though.

RE: Price Reduction?
By Mclendo06 on 4/13/2006 11:44:27 AM , Rating: 2
I'll answer my own question based on something else I just read. It won't make that much of a cost difference because this is simply a way to stack multiple wafers closer to one another. It does reduce the overall chip size (primarily thickness), but not the amount of silicon required overall.

RE: Price Reduction?
By akugami on 4/13/2006 1:02:20 PM , Rating: 2
Not so much cost as it is to reduce physical size needed to package more RAM. For instance, at the current physical specs of a iPod Nano and with traditional methods of packaging flash RAM, it'd be pretty much impossible to make a 16GB iPod Nano.

And guys, this is a blog styled tech site. It's not meant to be free of errors but meant to convey news fast. Does this mean that there will be a higher level of errors? Of course. I still think that they should at least better proof read their articles but online articles are not like traditional print with a dedicated editor that proof reads material.

RE: Price Reduction?
By bhigh on 4/13/2006 1:51:17 PM , Rating: 2
And guys, this is a blog styled tech site. It's not meant to be free of errors but meant to convey news fast.

A simple spell check would take care of many of the errors.

RE: Price Reduction?
By dunno99 on 4/13/2006 2:58:00 PM , Rating: 3
"Wafter" is a word, and so is "dice." So, no, a spell check wouldn't have solved the two problems pointed out by the other posts. What you need is a context-sensitive semantics check. And when you come up with such an algorithm to do just that, I have a Turing award to give you as well as a proposal for world domination.

RE: Price Reduction?
By Knish on 4/13/2006 7:59:12 PM , Rating: 3
"dice" was actually used correctly.

RE: Price Reduction?
By Samus on 4/13/06, Rating: 0
back on subject
By vtohthree on 4/13/2006 11:48:49 PM , Rating: 2
This is exciting! Samsung is really pushing the Solid State Hard drive into a reality, I can't wait to have a massive solid state CF card. And yes, I would need that much storage on my pda, use it to carry data and mp3's. Or when they start showing up in energy efficient laptops.

RE: back on subject
By shadowzz on 4/14/2006 12:12:15 AM , Rating: 2
I theory, if heat isnt an issue you could fit like several hundred gigabytes into the size of an existing flash drive. Now they just need to make it cost $100

RE: back on subject
By highlandsun on 4/14/2006 1:11:44 AM , Rating: 2
Yep, if you can stack these without sacrificing any speed, then sign me up for a solid state 2.5" laptop drive.

RE: Price Reduction?
By PrinceGaz on 4/14/2006 5:12:10 AM , Rating: 2
No it wasn't.

RE: Price Reduction?
By Eskimo on 4/14/2006 11:26:55 AM , Rating: 2

Yes it was used correctly. Don't take such a narrow minded approach to the word dice, it's not always 6 sided cubes. Maybe you should listen to those of us who actually work in the industry.
For your reference:

RE: Price Reduction?
By david99 on 4/16/2006 11:17:01 AM , Rating: 2
hmm sounds possible i guess, but anyone know what todays
NAND speed (with this tech)is compared to highspeed ram on the better HD's today, are we talking USB HD stick speed or far better read and write than your average SATA drive today ?.

Too bad..
By stugatz on 4/14/2006 2:17:42 AM , Rating: 2
this doesnt really do anything for increasing the write cycle life expectancy so solid state hard drives might be another step away, but if this means that in a year I can get an 8GB thumbdrive for under $200 then you wont hear any complaints from me.

RE: Too bad..
By The Cheeba on 4/14/2006 5:01:00 AM , Rating: 2
I have a feeling that is already pretty close:

RE: Too bad..
By PandaBear on 4/15/2006 4:09:33 PM , Rating: 2
What it does is allow manufacture to use cheap, low density dice (yes, dice) to build expensive, high density die. When the whole world is flooded with cheap slow, last generation flash, it will make the high end cheaper and consume the obsolete supply out there. Good move.

However, I would imagine it will be relatively expensive (all those work cost money) and it will only be used to build the super high end stuff from high end stuff, instead of build mid end stuff from low end stuff.

I think it is good for non-heat generating stuff like slow D-RAM (for server) and flash. It is not suitable for CPU.

"Vista runs on Atom ... It's just no one uses it". -- Intel CEO Paul Otellini
Related Articles

Most Popular ArticlesFree Windows 10 offer ends July 29th, 2016: 10 Reasons to Upgrade Immediately
July 22, 2016, 9:19 PM
Top 5 Smart Watches
July 21, 2016, 11:48 PM

Copyright 2016 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki