[size=15][size=13]Follow
Senior Software Engineer (2019–present)Updated Dec 14
Why is there no 128-bit processor, as there are 64- and 32-bit?
[ltr]A great deal of this comes down to a question of what you really mean by a “64-bit processor”.[/ltr]
[ltr]Back when the world was young, an “8-bit processor” had 8-bit registers, but most had 16-bit addressing (and the ALU was sometimes 4 bits, so when you added two 8-bit registers, the CPU often did that in two steps, adding the lower 4 bits, then separately adding the upper 4 bits plus carry from the lower 4).[/ltr]
[ltr]Nowadays, we’ve sort of reversed that: a “64-bit” processor (sort of) has 64-bit addressing, but has 128-bit, 256-bit, and in some cases even 512-bit registers. But since uses for numbers bigger than 64 bits are pretty rare, those bigger registers are normally used to carry out operations on a number of smaller operands with a single instruction. So, instead of an ALU that’s half the register size carrying out a single operation in multiple clock cycles, we have an ALU that’s 2–4 times the size of a normal register, carrying out multiple operations in a single clock cycle.[/ltr]
[ltr]Depending on how you prefer to view things, you could make a reasonable case that a current “64-bit” CPU is really a 128-bit, 256-bit or even 512-bit CPU.[/ltr]
[/size]
404.8K views
View 552 upvotes
View 2 shares
1 of 20 answers
Upvote
552
46
2
[/size]
[size=13]46 comments from
Ken Netherland
and more[/size]
_________________
Senior Software Engineer (2019–present)Updated Dec 14
Why is there no 128-bit processor, as there are 64- and 32-bit?
[ltr]A great deal of this comes down to a question of what you really mean by a “64-bit processor”.[/ltr]
[ltr]Back when the world was young, an “8-bit processor” had 8-bit registers, but most had 16-bit addressing (and the ALU was sometimes 4 bits, so when you added two 8-bit registers, the CPU often did that in two steps, adding the lower 4 bits, then separately adding the upper 4 bits plus carry from the lower 4).[/ltr]
[ltr]Nowadays, we’ve sort of reversed that: a “64-bit” processor (sort of) has 64-bit addressing, but has 128-bit, 256-bit, and in some cases even 512-bit registers. But since uses for numbers bigger than 64 bits are pretty rare, those bigger registers are normally used to carry out operations on a number of smaller operands with a single instruction. So, instead of an ALU that’s half the register size carrying out a single operation in multiple clock cycles, we have an ALU that’s 2–4 times the size of a normal register, carrying out multiple operations in a single clock cycle.[/ltr]
[ltr]Depending on how you prefer to view things, you could make a reasonable case that a current “64-bit” CPU is really a 128-bit, 256-bit or even 512-bit CPU.[/ltr]
[/size]
404.8K views
View 552 upvotes
View 2 shares
1 of 20 answers
Upvote
552
46
2
[/size]
[size=13]46 comments from
Ken Netherland
and more[/size]
_________________