Eric LaForest: Next Generation Stack Computing [stack vs register-based CPUs]

Digg.com ^ | 8/8/2006 | Eric LaForest

[TChris]

Eric LaForest delivers a crash-course on modern stack computing, the Forth programming language, and some projects of his own. Stack systems have faster procedure calls and reduced complexity (shorter pipeline, simpler compilation) relative to their conventional counterparts, as well as more consistent performance, which is very important for real-time systems. Many consider stack-based architecture's crowning feature, however, to be the unrivalled price-to-performance ratio.

[TChris]

Very interesting stuff. Seems they would be a very powerful machines, these stack thingies.

(Videos in many formats available at link)

[llevrok]

New stack computing??? I'm just getting used to RPN!

[theFIRMbss]

>Eric LaForest delivers a crash-course on modern stack computing

Very cool thread! Thanks!
My favorite thread of the week.
Have an Anna pic!

[TChris]

New stack computing??? I'm just getting used to RPN!

Ah yes... I used to have an HP 15C. Those were the days.

[2 Kool 2 Be 4-Gotten]

You're in luck - RPN is very closely related to the concept of a stack!

[TChris]

My favorite thread of the week.
Have an Anna pic!

Really? Umm... OK. Anna's cute. *shrug*

[Army Air Corps]

Don't forget Maria Sharapova...

[fedupjohn]

Ah yes... I used to have an HP 15C. Those were the days.

I had a HP 25, the memory got me thru calculus in college. God what long time ago that was.

[theFIRMbss]

>Really? Umm...

Well, now that I've read the thirty slides, there isn't too much new stuff there. In fact, there's more meat in this old '89 [!] book. Stacks are interesting.

[TChris]

In fact, there's more meat in this old '89 [!] book. Stacks are interesting.

OK, I want AMD to start working on the K10s architecture, with 64-bit stacks (at 6+ GHz) for everyone!

[HAL9000]

It's an interesting concept. I remember mainframes that had registers and no stack. But I think we'll continue to use architectures with both features for the foreseeable future. Stacks frames are great for passing arguments and containing local variables in structured programs, but good usage of registers can make a program fly.

[SunkenCiv]

I loved foolin' with the stack on the good old Apple II.

[fedupjohn]

God ..you're gonna make us old Freeper's post about this, aren't yea....

Give us a break.!!!!

[HAL9000]

Of course, you are aware that Gene Amdahl's decision to omit a stack from the IBM System/360 was a conspiracy to drive up software and maintenance costs.

[TChris]

It's an interesting concept. I remember mainframes that had registers and no stack. But I think we'll continue to use architectures with both features for the foreseeable future. Stacks frames are great for passing arguments and containing local variables in structured programs, but good usage of registers can make a program fly.

AFAIK, Curren't processors don't have a hardware stack. They just have registers. A "stack frame" is only a memory construct, not a hardware feature.

[HAL9000]

You're probably right. I only browsed the slides, and missed the full explanation of the stack hardware.

The presentation seemed to be based on the assumption that the registers would be used exclusively in read/modify/write cycles. I guess that's especially true for x86 processors with relatively small number of registers. But in processors with a large register set, a good compiler or assembly language program can optimize the code to minimize reads/writes to external memory.