3 ms·
Recollections of Early Chip Development at Intel (2001) [pdf]
- etep 11y agoFigure 3, hand drawn cell layout on Mylar. Nice. Design automation begins: By 1974, the result was being digitized on a Calma GDS I
- AnonNo15 11y agoAmazingly fascinating! Interesting how this process works in modern era.
- wkoszek 11y agoNowadays it's much more complex and very much like software. Things got very complex, so a lot of things which Intel guys did by hand is nowadays done by a design program. For example: they devote a lot of time to describe how layout was done by hand. Right now you nearly never draw new rectangles of metal layer to build gates. Big fabs like TSMC will give you the library of cells which they know they can product. You'll tie it in your EDA VLSI software and you'll basically get a netlist from a Verilog simulator. And from there you'll do more via auto-placement for highly repeatable parts, and you'll do placement by hand for things you really care about. And you're assisted by software doing design checks all the way from design to manufacturing.
- krupan 11y agoI was hoping they would explain where the term "tape out" came from. I always thought it was because they copied the digital form of the design onto magnetic tape and shipped it off to the manufacturer, but nobody has been able to authoritatively confirm or deny that for me.
- imr 11y agoThat is what I have always heard in engineering school. FTP transfers replaced the magnetic tape.