Will Windows 8 require a 128-bit processor?
- Thread starter punchy71
- Start date
salvage-this
Active Member
no.
^This.
wolfeking
banned
no. 128 bit is not yet possible. It will take a whole new instruction set redesign to get 128 bit. maybe windows 10 or 11. Even then it will be 64 or 128, just like 32 or 64 now.
TrainTrackHack
VIP Member
There won't be a 128-bit CPU for PC in the near future. There's simply no need - 64-bit is currently fine for addressing all the memory and for high-performance computing there's specialised instruction sets that already operate on 128- and 256- bit data (like SSEn and AVX).
Not for awhile. I know Microsoft has been working on it for a long time. Plus you know AMD and Intel has been. If like with 64 bit, Intel will probably be using what AMD will end up calling AMD128 x64-128
DMGrier
VIP Member
I could not even imagine what the daily PC user would even need a 128 Bit system for, I mean if you take a desktop and max it out to what 64 Bit can see and then handed it to a gamer he could not even bring it close to it't breaking point. On servers it might be needed.
I wonder if M$ will make 128 Bit before a Linux Distro unlike what they did with 64 Bit.
I wonder if M$ will make 128 Bit before a Linux Distro unlike what they did with 64 Bit.
DMGrier
VIP Member
Oh I know this, Linux beat M$ to 64 Bit by a year and know that Linux ownes over 50% of the server market it would be to profitable for them to make it first not to mention as a community we do have some of the most gifted programmers.
wolfeking
banned
it requires 32 or 64 bit to run. Most, if not all of the ARM processors that run smart phones these days are x86-64.
All current game systems are x86-64.
the bit rate has little to do with power. It is more related to the amount of memory that can be addressed. x86 (32 bit) can read to a maximum theoritical level of 4GB of RAM (actual amount will vary due to hardware and BIOS support). x86-64 will read to a theoritical maximum of 192GB currently. I do not think that is the current max for x86-64 though, as the max memory has went up with XP64 to Vista64, and up again to 7-64.
scratch that, 2008R2 x64 and IA64 supports 2TB of ram.
All current game systems are x86-64.
the bit rate has little to do with power. It is more related to the amount of memory that can be addressed. x86 (32 bit) can read to a maximum theoritical level of 4GB of RAM (actual amount will vary due to hardware and BIOS support). x86-64 will read to a theoritical maximum of 192GB currently. I do not think that is the current max for x86-64 though, as the max memory has went up with XP64 to Vista64, and up again to 7-64.
scratch that, 2008R2 x64 and IA64 supports 2TB of ram.
Last edited:
wolfeking
banned
a 20 bit (8086) could only address 1 MB of RAM. I would assume that the 8bit would be in the KB or Byte range.
16 Bit, probably somewhere in the area of 500-750 KB.
Memory addressing is a exponential operation. Exery bit added to it adds a lot more memory addressing capibility.
just think of writting each Bit out.
(8) ABCDEFGH (40312 possibilities if i remember the math right)
(16) ABCDEFGHIJKLMNO (2092278989000 possibilities)
(20) ABCDEFGHIJKLMNOPQRT (243290200800000000 possibilities).
each possibility is the address to some data stored in RAM. the higher the Bit rate, the more memory can be addressed, and the faster it can be addressed.
16 Bit, probably somewhere in the area of 500-750 KB.
Memory addressing is a exponential operation. Exery bit added to it adds a lot more memory addressing capibility.
just think of writting each Bit out.
(8) ABCDEFGH (40312 possibilities if i remember the math right)
(16) ABCDEFGHIJKLMNO (2092278989000 possibilities)
(20) ABCDEFGHIJKLMNOPQRT (243290200800000000 possibilities).
each possibility is the address to some data stored in RAM. the higher the Bit rate, the more memory can be addressed, and the faster it can be addressed.
TrainTrackHack
VIP Member
Th 8086 was 16-bit but it had a 20-bit address bus. With 16 bits you could address only 64KB RAM but I'm not aware of any CPUs that could address only 64KB of RAM (that of course doesn't mean they don't exist, I just never heard of any). With 8 bits you cousd only address 256 bytes (not MB or KB, just plain bytes). The memory you can address with a given number of bits is given by 2^n where n is the number of bits (and ^ means exponent).
I'm pretty sure there's no 64-bit ARM CPUs out there, ARMv8 is the newest version and has support for 64-bit but I don't think any CPUs based on that core have been shipped yet (or if they are they are far from being majority). In any event, they certainly won't run x86-64 code.
Theoretical maximum for most x86 CPUs is 64GB with PAE. Most current operating systems use PAE (Windows has since XP but most editions limit the physical address space to 4GB anyway).
What do you mean by faster?
Last edited:
TrainTrackHack
VIP Member
You can't address any more of it at once - you can only address one memory location at once, and it's done in a single cycle regardless of whether it's x86 or x64 (well, since the 386 anyway).
Well, to move 64 bits with a 32bit-command, you need to call it twice. In 64bit you need to call it once.
I would call it a practical maximum. It is the way PAE is implemeneted that limits it to 64 GB. Modern 64bit CPUs have a much higher limitation.
Last edited:
TrainTrackHack
VIP Member
But that's irrelevant as far as addressing goes - on x86-64 the registers are 64-bit but so are the pointers, so it's still going to take one instruction address a location in memory. Likewise, in x86 the registers are 32-bit and so are the pointers, so it's still going to take a single instruction.
Should've probably clarified, I meant x86 CPUs in 32-bit mode. But yea, practical does indeed sound a bit nicer than theoretical in the context...