I'll never tire of any analysis you do. But if you are taking requests, I'd love two chips.

The AMD 29000 series, a RISC chip with many architectural advances that eventually morphed into the K5.

And the Inmos Transputer, a Forth like chip with built in scheduling and networking, designed to be networked together into large systems.

https://en.wikipedia.org/wiki/AMD_Am29000

https://en.wikipedia.org/wiki/Transputer

At what number of layers is it difficult to reverse engineer a processor from die photos? I would think at some point, functionality would be too obscured to able to understand the internal operation.

Do they ever put a solid metal top layer?

I'd love to see an analysis of byte ordering impact on CPU implementation. Does little vs big endian make any difference to the complexity of the algorithms and circuits?

Never, the 386 is way too important.

So Epictronics recently looked at the 386SX, the version with the 16bit external bus, which was slower than the 286 at the same clock. What changed between that and this? Was the major difference the double clock hit on fetch? Or did it have a shorter prefetch queue as well like the 8088?

Ok, now do 486.

Never!

As I understand it, you can use slower carry propagation techniques in parts of a design that aren't on the timing critical path. Speeding up logic that isn't on the critical path won't speed up your circuit; it just wastes space and power.

Clock dividers (for example, for PLLs and for generating sampling clocks) commonly use simple ripple carry because nobody is looking at multiple bits at a time.

Well, the Manchester carry chain dates back to 1959. Even the 6502 uses carry skip too increment the PC. As word sizes became larger and transistors became cheaper, implementations became more complex and optimized. And mainframes have been using these tricks forever.

So many chips, so little time :-)

Microsoft and IBM were both developing OS/2 together. There were a lot of disagreements between the two companies. IBM wanted to keep supporting the 286.

https://en.wikipedia.org/wiki/OS/2#1990:_Breakup

> OS/2 1.x targets the Intel 80286 processor and DOS fundamentally does not. IBM insisted on supporting the 80286 processor, with its 16-bit segmented memory mode, because of commitments made to customers who had purchased many 80286-based PS/2s as a result of IBM's promises surrounding OS/2.[30] Until release 2.0 in April 1992, OS/2 ran in 16-bit protected mode and therefore could not benefit from the Intel 80386's much simpler 32-bit flat memory model and virtual 8086 mode features. This was especially painful in providing support for DOS applications. While, in 1988, Windows/386 2.1 could run several cooperatively multitasked DOS applications, including expanded memory (EMS) emulation, OS/2 1.3, released in 1991, was still limited to one 640 kB "DOS box".

> Given these issues, Microsoft started to work in parallel on a version of Windows which was more future-oriented and more portable. The hiring of Dave Cutler, former VAX/VMS architect, in 1988 created an immediate competition with the OS/2 team, as Cutler did not think much of the OS/2 technology and wanted to build on his work on the MICA project at Digital rather than creating a "DOS plus". His NT OS/2 was a completely new architecture.[31]

DOS extenders had started in the 1980's but they weren't a real OS but I would barely call DOS an OS either.

https://en.wikipedia.org/wiki/DOS_extender

But Unix was ported to the 386 in 1987.

https://en.wikipedia.org/wiki/Xenix#Transfer_of_ownership_to...

> In 1987, SCO ported Xenix to the 386 processor, a 32-bit chip, after securing knowledge from Microsoft insiders that Microsoft was no longer developing Xenix.[41] Xenix System V Release 2.3.1 introduced support for i386, SCSI and TCP/IP. SCO's Xenix System V/386 was the first 32-bit operating system available on the market for the x86 CPU architecture.

https://en.wikipedia.org/wiki/Xenix

I had friends running Linux from the very beginning in 1991.

In the 90s, every time you got a new computer it would have at least twice the RAM, sometimes 4x.

There was no performance improvement clock for clock between 286 and 386 when running contemporary 16 bit code https://www.vogons.org/viewtopic.php?t=46350

This appears to be a bot reposting comments from an older article on my blog.