Can somebody explain to me in detail what is the difference between 64bit and 32bit?

[neonerd]

Basically, i just want to know what makes 64 bit better than 32 bit (other than it has more bits), and a small explination of what makes a 64 bit processor "64 bit"....it's just baffled me for a while. I know it's something about integers, and how memory addresses are usually manipulated into integers in moden day processors.

 

[AyashiKaibutsu]

Your standard integer is 32 bits. This limits how large your integer can be. With integers this is no big deal they end up being plenty big. Now floating points are numbers that have precision. These need to use extra bits to keep track of where the decimal place is. Generally people use Doubles which are 64 bit floating point numbers. A 32bit processor can't deal with them all at one time basicly where a 64 bit processor can.

 

[neonerd]

Originally posted by: AyashiKaibutsu
Your standard integer is 32 bits. This limits how large your integer can be. With integers this is no big deal they end up being plenty big. Now floating points are numbers that have precision. These need to use extra bits to keep track of where the decimal place is. Generally people use Doubles which are 64 bit floating point numbers. A 32bit processor can't deal with them all at one time basicly where a 64 bit processor can.

thx...i was able to extract about a page worth of information from that

now i only need 2 more pages....are there any more highly technical things about 64 bit that you or anyone else here knows about? it's a combination that i need this for a report, as well as i am truly interested in it.

 

[Rapsven]

Double or floating point variables are actually 64 bit variables.

However, most processors out there now are 32 bit. What we do now to streamline it is to store the double as a mantissa and something else and then combine them together.

64 bit processing means that your processor can address 64 bits of information at a time. This does not boost speed by 100%, but it can be substantial in certain applications.

 

[thirdlegstump]

64-32=32

 

[ViRGE]

It's worth noting that we can already do 64bit Ints, but that they are broken up, and hence take a speed hit. By moving to a 64bit architecture, they can be handled "natively", which effectively means they can go faster, and not require any special trickery.

The other major change is at least as far as the x86 architecture goes, 64bit mode adds 8 more general purpose registers(GPRs) to the preivous 8, giving us 16 GPRs. x86 is considered a "register starved" architecture, so the additional GPRs allows for another speed boost, by allowing the computer to keep more stuff on hand at once, and reduce the amount of swaps it needs to do.

PS Can someone confirm the FP thing? I could have sworn most floats were 80 bits

 

[DaveSimmons]

> PS Can someone confirm the FP thing? I could have sworn most floats were 80 bits

The x86 math coprocessor (which was actually a separate chip when I was writing x86 assembly ) uses 80 bits internally but doubles are indeed 64 bits. 10-byte FP is called something else -- extended? I never use them anymore