>>>> Not compared to modern RAM circuits. I think transistors are getting 
>>>> down to the dozens-of-atoms size at this point.
>>>
>>> I doubt it.
>>
>> What do you doubt, that silicon atoms are as big as 0.25 nm across, or 
>> transistors are as small as 25 nm?
>
> If you made a transistor that consisted of only a few atoms, it wouldn't 
> work properly.

Probably not, but nobody suggested that.

Darren said that the *size* was comparable to some dozens of atoms.  A 25 nm 
transistor is only 8 dozen atoms across, but (if you assume it's a cube) it 
consists of almost a million atoms.

Clearly that's enough to still work ok, but you're right, the limit for 
traditional transistors can't be much further away.

Post a reply to this message

>>>>> Not compared to modern RAM circuits. I think transistors are 
>>>>> getting down to the dozens-of-atoms size at this point.
>>>>
>>>> I doubt it.
>>>
>>> What do you doubt, that silicon atoms are as big as 0.25 nm across, 
>>> or transistors are as small as 25 nm?
>>
>> If you made a transistor that consisted of only a few atoms, it 
>> wouldn't work properly.
> 
> Probably not, but nobody suggested that.
> 
> Darren said that the *size* was comparable to some dozens of atoms.  A 
> 25 nm transistor is only 8 dozen atoms across, but (if you assume it's a 
> cube) it consists of almost a million atoms.
> 
> Clearly that's enough to still work ok, but you're right, the limit for 
> traditional transistors can't be much further away.

Right. So when you say "size", I thought you meant volume, but you 
actually meant diammeter?

Anyway, regardless, the advantage of the algae is that it's packed in 3D 
instead of mere 2D, massively increasing the amount you can fit into a 
given space.

Either way, you would *think* neuron cells would be the "obvious" choice 
for making RAM circuits. They already have the ability to wire 
themselves up for you, so...

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

On Thu, 17 Apr 2008 12:25:05 +0100, Invisible <voi### [at] devnull> wrote:

>
>Either way, you would *think* neuron cells would be the "obvious" choice 
>for making RAM circuits. They already have the ability to wire 
>themselves up for you, so...

Then you would have RRAM. The stuff of Frankenstein movies :)
-- 

Regards
     Stephen

Post a reply to this message

Tim Cook wrote:
> Chambers wrote:
>> No, it means we'll finally be able to store our audio files with 
>> decent quality settings.  No more "Vinyl sounds better," guys!
> 
> Been able to store audio files with decent quality settings for a while. 
>  Ape, FLAC, et cetera.  The reason the "vinyl sounds better" guys make 
> that claim is, apparently, due to the digitizing process itself that 
> makes the *original* recording unacceptable to the audiophile's 
> oh-so-sensitive ear.
> 

Oh, the vagaries of the Internet, where sarcasm is lost :(

-- 
...Ben Chambers
www.pacificwebguy.com

Post a reply to this message

Invisible wrote:
> Either way, you would *think* neuron cells would be the "obvious" choice 
> for making RAM circuits. They already have the ability to wire 
> themselves up for you, so...

Not for storage.  They work best as pattern filters.

(Just try comparing your brain's ability to recognize pictures to its 
ability to remember minute trivia).

-- 
...Ben Chambers
www.pacificwebguy.com

Post a reply to this message

>> Either way, you would *think* neuron cells would be the "obvious" 
>> choice for making RAM circuits. They already have the ability to wire 
>> themselves up for you, so...
> 
> Not for storage.  They work best as pattern filters.
> 
> (Just try comparing your brain's ability to recognize pictures to its 
> ability to remember minute trivia).

There are people who have successfully memorised Pi to thousands of 
decimal places, routinely memorise decks of cards, etc.

The default way the *human* brain is wired up is optimised for tasks 
such as shape recognition, which are critical to survival, and poorly 
optimised for memorising digit sequences [which, for many millions of 
years, was *not* an important task].

That doesn't mean that you couldn't wire up neurons in such a way that 
they reliably store arbitrary data.

Here's a thought: When you turn RAM off, all the data gets lost. When 
you "turn off" a biological storage system... all the cells die? Ouch! 
UPS for the win! ;-)

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

Invisible wrote:
> If you were to make a device so small that it consists of only a few 
> dozen atoms, you couldn't be able to reason about it using the usual 
> laws of electricity. You'd have to use quantum dynamics or something.

And you think that people building modern semiconductor wafers aren't 
using quantum dynamics to design them?  I mean, you need QED to even 
make *any* semiconductor work.

> only, say, 20 atoms in size, then they'd *all* be ordinary silicon 
> atoms, and it wouldn't be a semiconductor.

Technically, it would still be a semiconductor. Just not a very useful 
one, because it would be 100% semiconductor.

> a sandwich of N-type and P-type semiconductive matter) with that few 
> atoms. It wouldn't work properly.

I believe the junction between base and emitter in a CMOS transistor is 
all one type of semiconductor (all P or all N) for what it's worth. :-)

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

Invisible wrote:
> I doubt it. At that scale, quantum effects would become significant, 

Well, yes. That indeed is the problem people are having. Quantum effects 
are knocking electrons out of the gate and into the substrate, or 
they're tunneling across the restriction zones even when it's switched 
off.  (Likely what I'm remembering is transistors where the active part 
that serves as a barrier is in the dozens-of-atoms size.)

> Also, significantly, transistors don't just grow themselves if you feed 
> them a little nitrogen and AMP. ;-)

There is that, yes.  :-)

I don't really need to have bigger drives. I just need faster drives. :-)

-- 
   Darren New / San Diego, CA, USA (PST)
     "That's pretty. Where's that?"
          "It's the Age of Channelwood."
     "We should go there on vacation some time."

Post a reply to this message

Darren New wrote:

> I don't really need to have bigger drives. I just need faster drives. :-)

Ah. Now *that* is a problem that biological devices are most unlikely to 
solve. Current hardware runs way faster than anything biological anyway...

Of course, the "other" way to do this is to have a really ****ing wide 
bus, which should be tractable for a biological system.

-- 
http://blog.orphi.me.uk/
http://www.zazzle.com/MathematicalOrchid*

Post a reply to this message

On Thu, 17 Apr 2008 16:55:32 +0100, Invisible <voi### [at] devnull> wrote:

>
>That doesn't mean that you couldn't wire up neurons in such a way that 
>they reliably store arbitrary data.

Good point
-- 

Regards
     Stephen

Post a reply to this message