128 bit processors???
-
Ars is claiming[^] that MS is planning to add support for 128bit processors in win8. What hardware is this supposed to run on? Attempting to find anything about forthcoming 128bit x86 CPUs only turned up rehashes of this rumor, clueless wishlisting, and confusion over the width of the processor architecture vs the width of SSE type instructions.
The latest nation. Procrastination.
I have heard no such thing, and it doesn't make much sense to me. A 64-bit address space is quite sufficient for my needs. Wider data paths are fine in vector units, I don't expect x86 to go and unify its integer and vector units though. :)
Luc Pattyn
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
Local announcement (Antwerp region): Lange Wapper? Neen!
-
I have heard no such thing, and it doesn't make much sense to me. A 64-bit address space is quite sufficient for my needs. Wider data paths are fine in vector units, I don't expect x86 to go and unify its integer and vector units though. :)
Luc Pattyn
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
Local announcement (Antwerp region): Lange Wapper? Neen!
Luc Pattyn wrote:
A 64-bit address space is quite sufficient for my needs.
I quite agree. Also didn't a certain W. Gates, reputedly, say 640K is more than enough for anyone?
Henry Minute Do not read medical books! You could die of a misprint. - Mark Twain Girl: (staring) "Why do you need an icy cucumber?" “I want to report a fraud. The government is lying to us all.”
-
Luc Pattyn wrote:
A 64-bit address space is quite sufficient for my needs.
I quite agree. Also didn't a certain W. Gates, reputedly, say 640K is more than enough for anyone?
Henry Minute Do not read medical books! You could die of a misprint. - Mark Twain Girl: (staring) "Why do you need an icy cucumber?" “I want to report a fraud. The government is lying to us all.”
he did, however that was before he discovered you can charge heavily for bloated bug collections. :)
Luc Pattyn
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
Local announcement (Antwerp region): Lange Wapper? Neen!
-
I have heard no such thing, and it doesn't make much sense to me. A 64-bit address space is quite sufficient for my needs. Wider data paths are fine in vector units, I don't expect x86 to go and unify its integer and vector units though. :)
Luc Pattyn
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
Local announcement (Antwerp region): Lange Wapper? Neen!
Luc Pattyn wrote:
Wider data paths are fine in vector units
And registers too. Actually, I don't see the point in simply doubling things, why not bigger jumps? :cool:
-
Luc Pattyn wrote:
Wider data paths are fine in vector units
And registers too. Actually, I don't see the point in simply doubling things, why not bigger jumps? :cool:
PIEBALDconsult wrote:
And registers too
Of course. The data path is everything that handles data, i.e. data types, registers, data caches, functional units, data buses. They always just double due to physical limitations, e.g. because they need technological improvements to overcome the ground bouncing risk when twice as many bits may change at the same time. :)
Luc Pattyn
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
Local announcement (Antwerp region): Lange Wapper? Neen!
-
PIEBALDconsult wrote:
And registers too
Of course. The data path is everything that handles data, i.e. data types, registers, data caches, functional units, data buses. They always just double due to physical limitations, e.g. because they need technological improvements to overcome the ground bouncing risk when twice as many bits may change at the same time. :)
Luc Pattyn
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
Local announcement (Antwerp region): Lange Wapper? Neen!
Oh, that data path. :-D (I had assumed only data buses. :doh: )
Luc Pattyn wrote:
ground bouncing risk
But aren't they smaller bits? Plus with more of them they should cancel each other out better.
-
Oh, that data path. :-D (I had assumed only data buses. :doh: )
Luc Pattyn wrote:
ground bouncing risk
But aren't they smaller bits? Plus with more of them they should cancel each other out better.
PIEBALDconsult wrote:
But aren't they smaller bits? Plus with more of them they should cancel each other out better
yeah, all kinds of bits get smaller, their number increases faster though; and as the transistors are getting smaller, they become more sensitive to their environment. When a number goes from 0xFFF...FFF to 0x000...000 all bits change, there is no cancellation; remember, it has to be right all the time, whatever the data (or data sequence) is, under all the specified conditions of supply voltage, clock frequency, and operating temperature. :)
Luc Pattyn
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
Local announcement (Antwerp region): Lange Wapper? Neen!
-
PIEBALDconsult wrote:
But aren't they smaller bits? Plus with more of them they should cancel each other out better
yeah, all kinds of bits get smaller, their number increases faster though; and as the transistors are getting smaller, they become more sensitive to their environment. When a number goes from 0xFFF...FFF to 0x000...000 all bits change, there is no cancellation; remember, it has to be right all the time, whatever the data (or data sequence) is, under all the specified conditions of supply voltage, clock frequency, and operating temperature. :)
Luc Pattyn
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
Local announcement (Antwerp region): Lange Wapper? Neen!
Oh, I thought you were making a joke about all those flipping bits causing the computer to vibrate. On a side note: "The 8086 had eight (more or less general) 16-bit registers" that sounds like 128 bits to me so maybe Win8 will run on those. :cool:
-
I have heard no such thing, and it doesn't make much sense to me. A 64-bit address space is quite sufficient for my needs. Wider data paths are fine in vector units, I don't expect x86 to go and unify its integer and vector units though. :)
Luc Pattyn
I only read code that is properly indented, and rendered in a non-proportional font; hint: use PRE tags in forum messages
Local announcement (Antwerp region): Lange Wapper? Neen!
I've never really seen much point in 64-bit tbh, other than for memory addresses. The 8-bit C64 must have had 16-bit addressing to get 64 KB and the older 16-bit Amigas must have had more than 16-bit addressing to get 8 MB (I think that's what the limit was anyway). All without doubling the RAM required to store the optimum sized integer.
-
GPUs?