Neowin.net

Ask Mr. Chronopolous, he knows a lot about how small things can get.

64bit is good, but useless for the next 5 years... as far as the everyday consumer is concerned...

5 years? Are you kidding? You can get cheap computers ($900) with 4gigs right now.

In 5 years, 4gigs will be like 512mb is right now.

i knew this would happen

you see, now people are confused because people call 64-bit x64 which is not correct
x86 = 32bit and 64bit

What are you talking about? There is absolutely no mention whatsoever of Microsoft or their servers in this article.

Please don't feed the trolls.

Why would you ever want to do something like that?

But to answer your question... that would be a "no." If you have two cores with two memory banks, you effectively have two computers, not one. You could link them together and do distributed computing through a fast local bus, but it still would not just be one computer. By definition, a 32-bit processor uses 32 bits to refer to the location of each byte of memory. 2^32 = 4.2 billion, which means a memory address that's 32 bits long can only refer to 4.2 billion unique locations (i.e. 4 GB). It doesn't mater if each core has it's own memory assigned to it: A core is not a CPU, and it's the CPU, not the core, that adresses the memory bank(s). It's the whole machine that is limited to 2^32 = 4.2 billion bytes, not just the cores.
Even in multi-CPU (multiprocessing) 32-bit servers, you can't have more that 4 GB of RAM. No way, no how, no duh. It's only when you move to x64 that more than 4GB becomes possible.

Also, remember that each device on the system bus that has it's own dedicated memory address(s) assigned to them use up the available addressable system memory, even though this memory may not be RAM. This is why systems with video cards cannot get to 4GB of addressable RAM under 32-bit OS: The video card has VRAM that is using up some of the addressable memory space.

Say you wanna run a 32-bit system with 4GB of ram and a pair of 1 GB video cards. Guess what, it's not gonna happen. You'll get 2GB of RAM at best, even if you slot 4GB. Hence, x64 is really the only way forward. Eventually, video games will actually NEED 4 GB of ram and then all the 32-bit gamers will be S.O.L.

ARM is a RISC (Reduced Instruction Set Computing) processor.
x86 is not.
Don't compare apples with oranges.

There are/were some x86 RISC CPUs.

What you meant to say was that there were some x86-BASED RISC CPUs.
True x86 is CISC, not RISC.

Yes. The AMD Athlon 32 bit CPU was RISC based, if I remember correctly.

Eh? For an embedded environment, say using Linux, you can directly compare performance. The only main difference is the performance of things like the FPU operations....which most ARM CPUs have added instructions to make up for it. Why do you think NVidia chose ARM for their embedded SoC if they want to support full HD on a ~400 MHz CPU? You wont get close to that on x86.

x86 is mixed RISC and CISC (Which in my mind makes it CISC)

Many processors internally split up complex instructions into simpler ones, making the actual execution RISC-like, but the high level architecture is still CISC. Look up 'microoperations'

You mean x86 for 40 years?

The "big iron" unix types have been running on a large variety of non-x86 CPUs, so we aren't really talking software here, but the CPU architecture/hardware side of computing.