Propeller Die Photo?

darco

I've always been a little curious as to what the propeller die actually looks like. Is there any chance Parallax might release a photo of the die? I'm sure others would be curious as well.

cgracey

We do have one with the various die areas outlined and labeled. We just need to post it. I'll ask Rich (at Parallax)·to post it on this thread.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

deSilva

Oh, PLEASE! My students should love it!

Jen J.

Hello all,
Rich told me you needed this.
Propeller Die diagram attached for your enjoyment.
Cheers, JJ

deSilva

Lovely!
Can you perhaps also give some additional data?
-the real size (in mil)
- production technology
- cell sizes (in nm)
...
Thanks so much!

cgracey

The die is 53 sq mm and perfectly square. So, it's 7.28mm on the edge, or 283·mils.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔


Chip Gracey
Parallax, Inc.

Lee Marshall

Thanks! really appreciated, i was just going to ask about this myself.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Holy parallel processing, Batman!


srry if im sound a little unprofessional sometimes, im 15.

Ale

The process is 180 nm, they said it some time ago .

I'd love a higher res photo

Jen J.

Here's another layout view and some measurements from Beau Schwabe (Thanks Beau!):

Attached is another die view from the layout perspective…
In answer to deSilva’s questions…

- the real size (in mil)
7.28mm on edge would be 7280 microns on one edge …. 1 mil equals 25.4 microns, So 7280 microns = 286.6 mils

- production technology
350nm

- cell sizes (in nm)
There are really too many here to name or give this kind of figure.

Post Edited (Jen J. (Parallax)) : 10/2/2007 9:11:57 PM GMT

Beau Schwabe

Ale,

The new process is 180nm. In the view that you see, the process is 350nm.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

Gavin

Beau,

Old 350nm, new 180nm?

Does that mean you have done a die shrink and we can get MLF packages?

Or is 180nm for PropII?

Regards Gavin

deSilva

Everebody knows deSilva likes charts! Here is one relating the general drain voltage (3.3, 1.8.,....) to the used process indicating that Gavin's question is quite valid... The chart was mainly produced to show the closure of the "Drain/Threshold Voltage Gap", sometimes called "Gate Overdrive".

We are accustomed to 90nm chips of course, the "forerunners" Intel/AMD produce in 65nm for a year now and Intel will now have its 45nm chips.

Just posting historic charts: Here is another (well known) one showing the number of transistors in high end micro processors over the years
Excuse when it seems OT...

Post Edited (deSilva) : 10/3/2007 8:56:01 AM GMT

Ale

deSilva:

In the chipTransistorsYear graph there are a couple of mistakes: Is not pentium proc, it is Pentium Pro(*), and is not Prescott, is Pres-Hot

BTW, those were never high end processors, well maybe the 4004/8008.

Edit: (*) My mistake, proc is the abbreviation of processor... PPro is P6 :-(

Post Edited (Ale) : 10/3/2007 10:41:23 AM GMT

deSilva

:-( I donot remember where I copied the images from, according to creation date they must have come from the same source..... But I shall retouch them according to your suggestions

I said "High end MICRO processors", and was referring to what I personally like to call "3rd phylum of computers" - see that chart, this time generically produced by deSilva, somewhere hidden in another thread here:


It should make obvious that microCONTROLLERs are a different kettle of fish than microPROCESSORs, and follow their own rules (of market!), and you should be very cautions in comparing them against each other.

Beau Schwabe

deSilva,
"We are accustomed to 90nm chips of course, the "forerunners" Intel/AMD produce in 65nm for a year now and Intel will now have its 45nm chips."
Something to also consider is production costs.· When I worked for National Semiconductor, many of the chips that we did were in 350nm, 180nm, and 110nm.· We certainly had the ability to go further and do 90nm,·65nm, and 45nm.· I think before I left we were looking into something even smaller 30nm perhaps?· At the smaller sizes·many of the designs become cost prohibitive.· Not to mention, when you do go smaller, the physical nature of how the silicon behaves becomes different or should I say enhanced.· In other words certain silicon characteristics that are not necessarily favorable are manageable at larger process variations but often require different strategies in the smaller processes in order to keep a desired functionality.
·
·



▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Beau Schwabe

IC Layout Engineer
Parallax, Inc.

deSilva

This was well worth mentioning! There are many other constraints valid for one-chip solutions in contrast to chips that can literally do nothing before a north- and southbridge is connected to them

Such things are the need of real current from the pins, robustness in handling, funny compatibility (you cannot always invent a new "socket" whenever you make a new microcontroller )

That's why I said: "You cannot simply compare the "third phylum" with the "fourth"..

Oldbitcollector (Jeff)

Moore's Law at work... (Although I don't think it still holds true.)

en.wikipedia.org/wiki/Moore's_law

Oldbitcollector

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Buttons . . . check. Dials . . . check. Switches . . . check. Little colored lights . . . check.

— Calvin, of 'Calvin and Hobbes.

deSilva

Moore's Law - in the strict sense - had just to do with the amount of integration on a die.
After the advent of multi-processor clusters the evolution of computing performance was no longoer bound to that but to archtectural constraints of connecting processors. This (and the power disipation) are AFAIK the limiting factors at the moment. A "drop" in chip integration density can now easily be compensated by a doubling of the number of processors used...

IMHO it is not fair to call this also "Moore's Law"! But as always the wikipedia has much to say about this...

I was more interested in the different "phyla" of computers having evolved during the last 60 years...

Ym2413a

I think it's interesting to see how a full computer 20 years ago can now fit into a single chip. [noparse]:)[/noparse]
Things do get smaller.

I mean look at this old vacuum tube computer memory from the 1940s:


4096-bits tops in something the size of... a vacuum tube. *lol* [noparse]:)[/noparse]

deSilva

Well, the RCA Selectrons were vacuum tubes And it was the end of the fourties to be precise. The - I should say: more successful! - English competition was the "Williams-Kilburn Tube" which - by pure chance and a little bit of phosphor - could nicely visualize the memory state!
This lead to the funny situation that the arguable first TRUE computer (the Manchester Mark I) already had a display. One of the first programs they did was Tic-Tac-Toe



Edit: And BTW: The Manchester Mark-1 is quite compareable to a single COG!

Post Edited (deSilva) : 10/4/2007 1:29:26 AM GMT