Slashdot Mirror
Toshiba To build Tiny DRAM
Charles Bronson writes "Toshiba Corp. in Japan announced today that they've developed the world smallest DRAM chip, with help from Big Blue. '[Toshiba] had achieved a chip-zide reduction of 40 percent with 0.175 micron technology.' Mass production won't start until late '99. This actually sounds promising, because the article reports that the smaller DRAM chips will help to lower costs. The bottom half explains that Toshiba and Fujitsu have teamed up to make even smaller DRAM chips (0.13 micron) availed by March 2002. 'Sound like a lofty goal to me."
so don't get their software...i very rarely find the need to "upgrade" my software
Texas Instruments have stated their intent to drop to .07 microns by 2002. Just talk, sure, until you remember that TI was the first to drop to .20 microns (after they said that they would), and that Sun has already inked fab agreements with TI for the 1 GHz UltraSparc III.
Fun fun.
Most DRAM and microprocessor manufacturers were already at or transitioning to 0.18 microns anyway. 0.175 Is just a tweek for publicity's sake, or else they were able to fit a few more dies on the wafer. It's about as revolutionary as an amp that goes to 11. Now 0.15 Microns, with copper and low-K, NOW you're cooking with gas...
- Lawrence Person
TI's DRAM operations including fabs and engineers were purchased by Micron. Micron may not make the big publicity announcements, but they figure out a way to be the last to loose money in a downturn and the first to make money in an upswing.
---Grady Borders
Take a look at Micron's stock over the last few years. In '95, DRAM producers were making obscene amounts of money. The bottom dropped out of the market, but it looks ready to rebound. Prices have stabilized and started coming back up. Sure you can build a high margin VLSI chip that might sell, or it might tank. On the other hand, if you build DRAMs, you know they will sell. The problem is building them cheaper than the other guy.
---Grady
It seems like every new hardware announcement nowadays has some relation to Big Blue. Sounds like they're trying to become the MS of hardware (again). At least they don't seem to suffer from the "Not Invented Here" syndrome.
Silicon has defects in its crystal structure. Since the amount of defects per area is more or less constant, smaller chips cause less bad chips, a higher yield, and thus lwoer prices.
Of course it's SRAM -- fast RAM is important. I also have a lovely 64meg HD cache of slowish DRAM.
... drool...
Hey -- I write videogames. Cache performance is all. If it's not in the cache, it might as well be on the floppy drive.
Fire and Darkness rulez! 3D realtime strategy, Win95/98/NT/Linux alpha@singularity.dyndns.com
Mmmmmmmmmmmm, voodoo3
Why would this be the case? Because of the oh-so-high cost of silicon? I don't actually see any crushing need for RAM chips to be much smaller... we just need more of them.
Geoff
one thing that could posibly happen (i dont know if its possibal) but with smaller RAM sizes, if they get small enuff you could put larger and larger L2 cache with out increasing the die any more on a CPU, that and maby they could cram L3 cache on te chip also. i kmow what 2megs of L3 dose for my alpha. anyway just a thought.
-magister-
Now RAM will be even cheaper...
So lazy programmers can write even MORE memory-hungry programs, in the expectation that people will have more RAM.
What ever happened to 640k?
I was eagerly awaiting the release of quest for glory 5, and now that it's out, I find that my 3-yr old box doesn't stand a snowball's chance of ever running it.
:-)
Even if I cleared *everything* off my hd except what I absolutley need, i'd still fall short.
And of course my old 486 boots faster than my pentium even though it has old, and supposedly inferior software.
At least linux can still run on a 286 from a floppy with just a few k of ram (special stripped kernel, of course)
No, I've found IBM products to be very high quality and reliable. As long as they don't start forcing inferior products down peoples' throats (microsoft) they are ok in my book.
"640K ought to be enough for anybody"
One of the great William H. Gates quotes, along with "What's a network?" (circa 1985)
-- I reserve the right to be completely wrong --
A certain Redmond based company almost certainly used to bloat their code, due to an alliance with a certain chip manufacturer. However, now both companies have other alliances, I can't see what the benefit is to MS now. Due to the way MS dev tools work, though, they tend to generate very bloated code (don't know much about it, but I guess there's a /lot/ of static linking going on)
Still, it does mean that us Linux users either get fabulously cheap, or superbly fast, spacious machines, whilst the rest of the world have to settle for bloated grinding 'doze.
-- I reserve the right to be completely wrong --
What benefit are we going to achieve by this?
oooh 20 bucks off wow...
I say if you are going to make RAM twice as small make it so I can put in twice as many modules.
OOps that would make sense
If I were them, I'd push my expensive, profitable chips into the higher-resolution processes long before I'd push memory, which is still priced fairly low (even the 128M PC100 ECC SDRAM has come down to $200 or so).
Have the memory makers finished getting rid of their stockpiles yet?