3rd Cpu

[owlofcreamcheese]

Anyone think there is any chance of a 3rd flavor of CPUs becomeing at all popular ever?

we have PCs and MACs right now. and chance of a third type ever makeing headway? (not talking OS... talking CPUs we have x86s and we have risc based macs, but nothing else owned by any amount of people)

it seems like something like a PC but started from scratch right about now could be really good, take out all the sillyness backwards compatablity brings around (take out sillyness like the accumulator and make registers totally multi-pourpous)

basicly, how long do you think we are gonna have to go before some company starts with a clean slate, a CPU that doesn't have to conform to standards made in the early 80s (probobly coupled with) an OS that doesn't have to worry about running anything written before 2000 something.

does anyone even think this is something that would be good? I feel like a start from scratch would be the best things to happen to computers but I studyed assembly back when the pentium 2 was comeing out and just a footnote in the book, so mabey since then all the crazy stuff that came from makeing it backward compatable or just lazyness of never rewriteing something after it was written in an older processor has gone bye bye, its possible that pentium 3/4s have fixed some of it.

still, anyone think it will happen? anyone hope it does happen? anyone have a prediction on when it could?

[scr4]

Apple did that when they switched from 680x0 to PowerPC chips. The Intel Itanium (64-bit server CPU) is also a new architecture which is incompatible with the Pentium, requiring new software. PDA manufacturers sometimes switch to new, incompatible CPUs as well. Palm is switching from Dragonball to StrongARM processors.

But I think the x86 architecture will be around for many more years. It's an open standard with numerous OS, software, CPU and system manufacturers involved. And apparently it's possible to build new CPUs with a wrapper layer that translates x86 instructions to native, internal instructions. The Transmeta Crusoe works that way.

[Catalyst]

owlofcreamcheese, maybe.

Anyway, what about Itanium/IA-64? That was neither x86 nor a Mac RISC layout (by the way, isn't current x86 more or less RISC too?). The program used in my A+ training course (from about 1999 :rolleyes: ) claims that Itanium is the "chip of the future." Well, look how that turned out. . . .

On preview, it looks like I've been beaten to the punch. Curses, foiled again!

[yabob]

And, of course, Sun will be perfectly happy to sell you a machine with one or more SPARC processors in it. HP and IBM also sell other CPU's. I know you might not think of the high end workstation, server and supercomputer markets as "owned by any number of people", but maybe if you considered dollars spent instead of number of customers ...

[owlofcreamcheese]

caralyst and scr4, yeah, PS4s I know do not REALLY work anything like old x86 chips, but they pretend they do, and thats silly, they make new space age chips and then leave on all the awful low tech 20+ year old nonsense from the old days.

oddly enough it probobly doesn't hurt anything at all to keep around emulated versions of old stuff, I think this is just my burning hate of some of the weird gymnastics you had to do to program a pre 386 chip, and how much they could improve by throwing it out and starting over... thinking about it, they probobly have by now then just emulate the functions with microcode

I still wonder if there will be a real alternitive, one at least as popular as macs any time soon. its easy to imagin a CPU thats desined to be a "person's computer" one thats intended to have strengths in the areas people actually use computers for instead of being totally universal then make it nearly uncrashable (less diversity means less errors) then make an OS with the same ideal. (make it a typeing/e-mail/internet/mp3 machine, CPU and OS) sell it really cheap, wouldn't people flock to that?

[Alereon]

Designing a CPU that doesn't suck costs a LOT of money and time, and if the only initial market was super-cheap appliance machines, it wouldn't make nearly enough to finance itself. If someone invents a new, non-x86 processor that can run existing x86 applications in emulation at a decent speed, then sure, it might have a chance in the market, but I sincerely doubt this will ever occur. Our only hope at this point for a revolutionary new CPU is AMD's x86-64 technology, which will extend x86 into the 64-bit world. It will offer nice speed improvements, without requiring existing apps to be recompiled.

Creating an entirely new architecture is a monumental task, Intel's new Itanium processor, despite being an entirely clean slate and costing thousands of dollars, still has unacceptably low performance when compared to other server processors, or even modern x86 desktop processors.