Discussion: Is x86 at EOL or is it still alive and kicking?

[Entity_Razer]

Well I wonderd today, we are all using a computer who'ms architecture (x86) is like... 10 years old if not a lot older.

In computerterms this means we are running a fossile architecture strictly looking at the age. Now I am wondering, is x86 ending its End Of Life (EOL) or is x86 still worthy of the development. I mean would it be better to switch to a new architecture all together to allow for better computing with less heat, less power consumption no restrictions vs compatibility so just start from the ground up or do you say x 86 architecture has just been scraped at the top?

Please explain why you think so and clarify as much as possible :)

[Vapor]

Based off the fact that Mac is slowly switching to it, I think it's alive and kicking ;)

I'd like to see what other developments are a potential though and what they have to offer....of course, in order to 'sample' the developments, they'd need to already have tons of money and time poured into them making it almost not worthwhile unless the entire industry would want to switch.

[Ugly n Grey]

x86 is a concept for code compatibility. Todays chips share little enough with the chips of yesteryear . The architecture we term x86 will be around for decades. We are already at the point where most machines are not capable of executing the original x86 code without modifying the code or running an emulator.

[masterofpuppets]

x86 is dead, in my opinion. Although, as UnG has mentioned, desktop chips these days are drifting away from true x86, they are still x86 components at heart. x86 is a terrible architecture in my opinion. Over a decade ago, other architectures were already enjoying performance that surpasses many modern x86 chips at much lower clockspeed. If I controlled the industry, I would've flushed the failure that is x86 down the crapper many years ago and adopted a much more effecient architecture.

[uOpt]

It is actually 18 years old, when you count common availability, Compaq came out with the first mass-produced 386 in 1987.

By now it is just a code convention. No CPU directly implements the ia32 instruction set, they have something more fundamental and then an emulation layer of various depths on top of it.

I just wish they had settled for something more elegant. In 1992 a DEC Alpha had all you ever wanted, instruction-wise and 64 bit wise.

[JAWS]

I just wish they had settled for something more elegant. In 1992 a DEC Alpha had all you ever wanted, instruction-wise and 64 bit wise.


DEC's FX!32 ;)


x86 is here to stay. x86;s prevalence is due to the amount of software that runs on it, until that changes x86 will continue to stick around.

[skate2snow]

Well.... I almost wonder if we can still call it X86.... But it would be a quite long process to change of architecture... And because of different budgets from Intel, AMD and IBM, it would be quite hard to make them change all toghether...

[Entity_Razer]

yhea I know x86 will be around for a long time unless Microsoft starts supporting a new architecture and AMD, Intel and IBM switch over together. But what I'm getting at is the architecture and chips themselves. As said before, when x86 was in its babyshoes we had computers that where a lot faster like the Alpha chips named here that where faster, more efficient and had even 64 bit support. Now also concerning performance itsself etc what do you think would be the logical choice. I mean imagen if we could have a chip convention where all the leader of the chip industrie would meet with the big OS developers (yes *nix, and the BSD's are big OS's to) and they would need to decide on a new architecture or at least big changes to the current way chips do their work/computing. I know chips today differ from the "true" chips from way back A LOT and thats only normal but I'd like to hear what you think would be the better architecture or what would need improving that could benefit the current chips when still keeping true to the compatibility we have now. I mean I don't like a swich to a new architecture as much as anybody else because AMD, Intel, IBM, ATI, Nvidia, and all other chip producers, from TWSM who bakes them to Matrox hell even HDD fabricators would need to addapt but performance and future wise, what would be the better choice?


I mean if it'd be up to me I'd either flush the whole architecture down the drain or at least overhaul it A LOT. And this is not only concerning instructions this can even just be make all chips run more efficient, even if it costs more (I'm talking about technologies like used on the Crusoe processors etc by Transmeta).


The latest P4's are a perfect example on how an architecture can go bad. I got to admit that was a dumbass dicission by intel but still....

Now I do know DEC computers or other would have problems of their own but just look at for instance (but this is kinda software based also) on how errors can happen during storage of binary code. I know this has more to do with the precision etc inside the system itsself and I think that can be alterd in a new generation of processors but its not logical that a proc takes a single precision binary, transforms it to a double to do its calculations and them, with double the accuracy writes it back to a single precision binary.... I mean youre bound to get errors one day. There were problems with this before with certain missiles from amerika in the early days. Now I know thats an old example and at the moment military hardware is nothing compared to our desktop PC's but i think you know where I'm getting at here.

I honestly do not feel x86 is future proof.

I mean, maybe it wouldn't be such a bad thing to start from the ground up. Like a W3C but then for computer architectures/chips.

edit: what I ment with transmeta is not only that their cpu's are very efficient with their power consumption but also that they alreayd have a northbridge on board (efficeon proc) and we are already seeing the proc market move in that direction like AMD with it's on board mem controller.