Speaking of die size and # of transistors packed onto a chip ....
In a batch of other interesting stuff I received an odd looking camera
reduction system. Looks a good bit like a 1920's Photographers "View"
camera. Used 4" X 5" Ferroxoplates. It [unlike the 20's version] is
built entirely of metal [heavy --about 25 pounds]. It is supposed to
have been used to reduce chip drawings down to actual production die
size. No name anywhere .. almost looks to be lab or shop built. I've
posted a couple of pix at:
http://members.home.net/ip500/chipmain.JPG
http://members.home.net/ip500/chipart.JPG
http://members.home.net/ip500/chipback.JPG
The second picture is of the only exposed "plate" that came with the
system. Third pix showing the back and partial interior is interesting
in that the plate is held to the back by vacuum. A hose runs out the
back and the vacuum acts through tiny channels milled into the back of
the camera. Any thoughts or comments appreciated. NO APPRAISELS are
solicited!
Thanks, Craig
Richard Erlacher wrote:
Real data is nice, but I find it odd that the 6502 shows up with so much more
die size when the transistor count matches the 6800 so closely. The Z80, and
6502 were both presumably built with more or less the same design rules, and, as
one might expect, the 6800 and 6502 should have approximately the same
transistor count, having approximately the same internal resources. Note how
the figure for the 8085 tracks the Z80 figure, scaled back for the fact it (the
8085) lacks the alternate register set, approximately. I don't trust the
indicated die-size, however, since, back in '76, when the 650x series was new,
it was reputed to have the smallest die size of any of the then-current 8-bit
CPU's. After all, that's how they (MOS Technology) bought their market share.
Everything seems to fit, with the exception of that die size figure in this
case.
The 8008 was a couple of generations earlier than the 8085, but the 8085, 6502,
and Z80 were all built in the same geometry. The 8080 and 6800 were both about
the same generation, hence also built in the same design rules and geometry.
Note that the 1802 die is bigger, being a CMOS device.
The die size for the 68K also seems to match what I remember of the die that was
glued to my datasheet at that sales pitch I mentioned in an earlier post. Keep
in mind, however, that the area was 1/4 that of the previous generation, as
there had been a halving of the drawn channel widths. While I believe that the
transistor count may have been more of a marketing ploy than a reliable count,
if you take into consideration the relatively large register size and count
contained in the 68K and combine that with the substantially smaller geometry in
which it was fabricated, it seems to match up with the majority of the remaining
figures.
Dick