Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

That wasn't the point. The point is that most of the field abandoned stack hardware for register hardware, upon which all sorts of software paradigms are implemented.


Sounds like somebody hasn't read Koopman yet.


I've read Koopman; fascinating stuff. Really solidified my understanding of the use of multiple stacks in Forth-like environments.

But I think it's instructive that none of the commercial architectures he describes (from 1989) have a contemporary descendant (although I think you can still get the RTX 2000 as a space-rated special order at an astronomical unit price). I think that tends to support my premise.


That's reasonable. I thought you were talking about things like the B5000 and the 8087.


Well...I can.

The 8087 was basically obsoleted by register-based SSE2 and SSE3, and as I understand on x86_64 the vestigial 8087 op codes are decoded microinstructions executed by the SSEx unit. And the descendant of the B5000 (now called Unisys ClearPath) is a VM running on Xeons; they stopped making stack-based MCP processors in 2015 or so.

So I think the original point still stands.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: