3d Printing a CPU

[bit-n...@hotmail.com]

OK, I'm probably talking out of my ASS here, but.....is such a thing possible, or WILL it be possible one day in the future? I thought people here might know.... :) (cause I already heard about some MIT students 3D printing a circuit board!!)


Thanks.

[Rick C. Hodgin]

On Sunday, April 1, 2018 at 3:00:49 AM UTC-4, bit-n...@hotmail.com wrote:
> OK, I'm probably talking out of my .. here, but.....is such a thing possible, or WILL it be possible one day in the future? I thought people here might know.... :) (cause I already heard about some MIT students 3D printing a circuit board!!)

They run around 1 MHz max at present:

https://www.youtube.com/watch?v=01y6bR6ETpA

--
Rick C. Hodgin

[Chris M. Thomasson]

Nice. I wonder if they can stack a bunch of them together. There would
be a lot of processors if this thing was, say a couple of inches thick.
The length of the z-axis would determine the number of layers. Just
thinking out loud here. Sorry.

[paul wallich]

On 4/1/18 3:00 AM, bit-n...@hotmail.com wrote:
> OK, I'm probably talking out of my ASS here, but.....is such a thing possible, or WILL it be possible one day in the future? I thought people here might know.... :) (cause I already heard about some MIT students 3D printing a circuit board!!)

So much depends on what you consider "3D printing". Current chip fab
lines are not entirely unlike the various layer-by-layer 3D printing
processes, if you make allowances for materials and dimensions.

People have also used layered ink-jet printing to make active devices
(transistors) and conductors, but the minimum feature dimensions are
rather large for a modern CPU.

paul

[Rick C. Hodgin]

On Sunday, April 1, 2018 at 3:50:37 PM UTC-4, paul wallich wrote:
> People have also used layered ink-jet printing to make active devices
> (transistors) and conductors, but the minimum feature dimensions are
> rather large for a modern CPU.

Size should not be a factor for ink-jet printed devices. For one,
they are getting smaller and more capable performance-wise. Two,
they are effectively one-off devices able to be printed as needed,
making the pipeline from mind to design to layout to product a much
faster turnaround of 10s of hours rather than weeks.

3D ink-jet CPUs are the most exciting thing in computers. They will
allow compute everywhere, in every product, for every application
and design. And, they're only getting better.

--
Rick C. Hodgin

[Chris M. Thomasson]

Agreed that they are only getting better.

[John Levine]

In article <00576c24-59e0-4e67...@googlegroups.com>,

That's not 3d printed -- it's conventional lithography on flexible
plastic. It's pretty clever but it's not the same thing.

--
Regards,
John Levine, jo...@iecc.com, Primary Perpetrator of "The Internet for Dummies",
Please consider the environment before reading this e-mail. https://jl.ly

[Rick C. Hodgin]

On Sunday, April 1, 2018 at 7:01:09 PM UTC-4, John Levine wrote:
> In article <00576c24-59e0-4e67...@googlegroups.com>,
> Rick C. Hodgin <rick.c...@gmail.com> wrote:
> >On Sunday, April 1, 2018 at 3:00:49 AM UTC-4, bit-n...@hotmail.com wrote:
> >> OK, I'm probably talking out of my .. here, but.....is such a thing possible, or WILL it be possible one day in the
> >future? I thought people here might know.... :) (cause I already heard about some MIT students 3D printing a circuit
> >board!!)
> >
> >They run around 1 MHz max at present:
> >
> > https://www.youtube.com/watch?v=01y6bR6ETpA
>
> That's not 3d printed -- it's conventional lithography on flexible
> plastic. It's pretty clever but it's not the same thing.

Where do you get that it is conventional lithography? I read a metal
oxide layer on a polymer, but it does not say how the polymer is
deposited, or that it's etched. He refers to patterning on layers
and interconnect registration, but that doesn't mean conventional
lithography-like techniques.

I always assumed it was some type of ink-based deposition for each
layer.

Production facility cost is 100x to 1000x lower than a conventional
fab. That's a modern ~$2 billion fab scaled down to $2 to $20 million
for a PragmatIC fab.

--
Rick C. Hodgin

[Chris M. Thomasson]

On 4/1/2018 4:47 PM, Rick C. Hodgin wrote:
> On Sunday, April 1, 2018 at 7:01:09 PM UTC-4, John Levine wrote:
>> In article <00576c24-59e0-4e67...@googlegroups.com>,
>> Rick C. Hodgin <rick.c...@gmail.com> wrote:
>>> On Sunday, April 1, 2018 at 3:00:49 AM UTC-4, bit-n...@hotmail.com wrote:
>>>> OK, I'm probably talking out of my .. here, but.....is such a thing possible, or WILL it be possible one day in the
>>> future? I thought people here might know.... :) (cause I already heard about some MIT students 3D printing a circuit
>>> board!!)
>>>
>>> They run around 1 MHz max at present:
>>>
>>> https://www.youtube.com/watch?v=01y6bR6ETpA
>>
>> That's not 3d printed -- it's conventional lithography on flexible
>> plastic. It's pretty clever but it's not the same thing.
>
> Where do you get that it is conventional lithography? I read a metal
> oxide layer on a polymer, but it does not say how the polymer is
> deposited, or that it's etched. He refers to patterning on layers
> and interconnect registration, but that doesn't mean conventional
> lithography-like techniques.
>
> I always assumed it was some type of ink-based deposition for each
> layer.

Imagine a printer with access to many colors... Now, think of each color
being represented by a different material that can be used as a form of
"ink". The ink cartridge would consist of various exotic metals and
polymers, like a spectrum of color. Perhaps the print head(s) can
possibly be fractal like where multiple inks can be used at once to
create different "reactions" between different materials used in the
same location.

Sorry if this sounds too off the wall. ;^o

[Quadibloc]

Why not 3-D print something easy, like a Stradivarius violin, and use
an FPGA to make a CPU?

Actually, though, while the technology isn't here yet, even to
3-D print a CPU at an old-fashioned process point, I'd love to be able to
do it. But it's not an option yet.

[Quadibloc]

Remembering IBM's solid logic technology, all you would need to do
is squirt transistors, a conductor, and an insulator for trace crossings,
on an insulating substrate.

Actually printing a transistor that works is too much to expect.

Trouble is orienting the transistors. I suppose they could be round with
circular contact rings, maybe even on both sides.

[Chris M. Thomasson]

Fwiw, for some reason this makes we want to think of the tangential
points of all the circles within an Apollonian gasket as actual
"connections", or possible points of networking. Now, since the border
of a gasket is a circle, well, why not tile those little fractals as if
they were representing a "far away" "zoomed out" view of a "proprietary"
_zoomed_in_ location... Again, sorry for thinking out loud here, try to
think of a little sci-fi here... ;^)

[Robert Wessel]

On Sun, 1 Apr 2018 22:24:20 -0700 (PDT), Quadibloc <jsa...@ecn.ab.ca>
wrote:

Or you could just buy an FPGA.