Let It Begin


This may cause a revolution in the PC industry. I predicted back in 2000 that the version of Windows after XP would be the last with a monopoly, it's looking like that may well be the case although I expected it to happen in 2005ish rather than 2009. Damn Microsoft and their eternally slipping release dates :)

I'll be glad to see the back of the x86 architecture too, it's held back true progress for way too long (IMO).
 
Klepto said:
I'll be glad to see the back of the x86 architecture too, it's held back true progress for way too long (IMO).
Agreed. Vive la Révolution!
Click to expand...
 
Last edited by a moderator:
Peter R said:
How has x86 held back the computing world? It's kept up with moores law projections just fine so it's hardly holding us back...

Disclaimer: I'm no guru, there are others here who know a lot more about this stuff than me (you may well be one of them).

Most of my knowledge about CPUs comes from a software perspective, and a particularly eclectic one at that. From a programmer's perspective the x86 is ugly, arcane and even a bit crazy. They are built on RISC cores but still present a CISC interface, they have irregular instruction sizes and varying instruction execution times, there are a lot of special cases and counter-intuitave optimisations. There are a whole bunch of threads here about ARM vs. x86 that touch on this issue.

x86 has had a near monopoly in the CPU market for things that Joe Bloggs calls a 'computer' for almost quarter of a century. Monocultures such as this tend to stifle innovation. I don't mean to say that there has been negligible progress, or that current x86 chips are the same internally as the venerable 8086, but we simply don't know where we would be now if x86 had never become king of the hill. It's had so much of the mindshare and investment that it was pretty much guaranteed to improve. Some really interesting ideas have been commercial failures partly because x86 was so entrenched, DEC's Alpha was considered by many to be superior but never managed to dethrone the x86. Interestingly the dominance of x86 was not brought about directly by the merits of the chip but because it was the primary target for the dominant operating system, which due to it's proprietary nature and binary distribution is not easily portable.

For the first time in a long time the stranglehold of wintel on the PC market may be facing a serious threat because of the emergence of netbooks and other similar devices. ARM and friends are the challengers on the hardware front, and Linux and friends on the software front. If ARM succeeds then native binary backward compatibility would be broken and a major reason that people stick to Windows would disappear. If Linux succeeds then it really doesn't matter what CPU you are running on as long as it runs Linux and a major reason that people stick to x86 would disappear.

As for Moore's Law (really an observation) it seems to me it can act as a destabilising factor. As competing systems reach a level of performance that's acceptable for people's needs relative performance ceases to matter so much, all that's needed is 'good enough'. Even though the TV ads tell you that you need the latest and greatest system to surf the web, type a letter, watch a movie or play music this simply isn't true. If the choice is between a $150 computer that does everything you want and a $300 computer that does everything you want then the choice seems obvious, especially when you consider mobile use.
Click to expand...
 
Last edited by a moderator:
Moore's law just dictates how transistor counts increase. Any CPU manufacturer can keep up with Moore's law if they get processing down, regardless of what CPU they're putting on the wafers. And Intel has been a world leader in making smaller dies practically since the beginning.
 
Exophase said:
Moore's law just dictates how transistor counts increase. Any CPU manufacturer can keep up with Moore's law if they get processing down, regardless of what CPU they're putting on the wafers. And Intel has been a world leader in making smaller dies practically since the beginning.
Click to expand...
I know Moore's law strictly talks of transistor counts but I meant it as an analog for "processing power".
 
Last edited by a moderator:
Processing power can be different if you have different architectures that are more efficient. x86 just runs a ton of crap. But compare how much power an Arm chip needs and than compare that to an x86 chip and you will see what different architectures can do.

x86 is a decent arch, its problem is that Intel and AMD can't break how it works because a lot of older software depends on it to run. It's taking awhile for x64 to take off really, even now Adobe only started working on flash 11 for x64 version of Linux.

It basically comes down to CISC vs. RISC.

A better way to see it is, pretend x86 is a V-6 and all we can do is figure out how to push more fuel and air into the chamber. Whereas ARM, they can take a different approach on how to move a car because they have the benefit/con of not having much mainstream support (Handheld mostly). So you look at the Cortex and you see the ARM chip with the SGX and DSP.

Quake and Unreal ran under software rending only and than discrete cards can along to help performance. But generally, x86 could do codec stuff because of the horsepower it had. You can find dedicated hardware to encode/decode but it was always cheaper to just let the x86 engine handle it.

So x86 is like a truck which just gets stuff thrown on top of it and is carrying a BIG hitch full of back software that NEEDS to be compatible. ARM is now starting to catch the eye of everyone which is why Intel came out with the Atom.

ARM to be sure, in the long run stands to overtake x86 just from an efficiency standpoint. When that is, maybe a couple of more years, we have to see how Intel markets the Atom and even AMD with their fusion chip.
 
Perhaps of minor interest is the fact that in the corporate world of Solaris, x86 is making positive progress.

Due to cost, there is a dramatic move away from SPARC and towards x86 eg Opteron.

Its fair to say that (for better or for worse) x86's death is still a long way off.
 
I think current designs like ARM and x86 will just morph into something else as time goes by. Call it evolution.
 
You must log in or register to reply here.

Similar threads

mali
Microsoft Linux
Replies
7
Views
847
quasist
Q
H
  • Locked
Big Update From Ed
5 6 7
Replies
128
Views
16K
Phawx
Phawx
Fzero
Hama mini beads
Replies
6
Views
6K
Fzero
Fzero
Stradian
Top-THIS!
Replies
16
Views
1K
Gerix
Gerix
A
Rechargeable Batteries, Chargers, And Led Flashlig
Replies
5
Views
1K
Steve-O
S
Back
Top