lets put a 486 on a 65nm what would you get

[IfReborn]

just wondering how small we have gotten in 20 years

 

[FelixDeCat]

486 not 485

 

[IfReborn]

typo

 

[Aluvus]

This is a ... difficult question to answer directly, but let's just look at a few comparisons to get some idea.

This page lists the Intel 486DX-25 on a 1 micron process, with 1.2 million transistors in 165 mm^2.

Dropping to a 0.8 micron, it's the DX-50 with 1.2 million transistors in just 81 mm^2.

Jumping forward to 65 nm, we have the Pentium D 9xx series (which are dual-die) packing 376 million transistors into 140 mm^2, a bit less than the DX-25 on 1 micron. A good portion of that is cache. That's over 300 times as many transistors, in less space.

I am assuming all of those numbers are accurate because I'm too lazy to look for a second source to confirm.

 

[myocardia]

Originally posted by: Aluvus
I am assuming all of those numbers are accurate because I'm too lazy to look for a second source to confirm.

Haha, I really like your train of thought there.:beer:

 

[BrownTown]

Basically it would be a spec of dust in size.

 

[CTho9305]

I took the image at http://upload.wikimedia.org/wikipedia/commons/0/02/80486dx2-large.jpg and the image at http://www.chip-architect.com/news/Northwood_130nm_die_text_1600x1200.jpg and scaled the 486 so that the cache approximately matched the L1 data cache on the Northwood, and I can fit roughly 28 486s on the 90nm P4 die (7*4).

 

[coldpower27]

I would guess that on 65nm that the 486 would be around 0.5-1mm2 in size. We have come quite a long way in the last 20 years.

Well just for the record the Pentium D 9xx Series die size is off, it's actually 2x81mm2 which works out to a total 162mm2.

 

[hans007]

the whole we can put 20 486s on one die idea is sort of like the cell spe idea.

there probably are 486 class chips on a 110nm or 90nm proces out there, being made for embedded designs and such.

 

[BrownTown]

on another random note, this highlights the sillyness of going ZOMG an 80 core processor! Since for all we know it could be 80 486 cores on one die which would only be better than a normal processor in certain situations.

 

[Roguestar]

What if they had four hundred and eighty-six 486 cores on one die...?

/spacetime continuum

 

[Cheex]

Originally posted by: Roguestar
What if they had four hundred and eighty-six 486 cores on one die...?

/spacetime continuum