• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

32/64 bit

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

fiji

Member
Joined
Jul 14, 2002
Location
berlin
im wondering with the new 64bit amd's comming out


whats the differencE? i dont even know what the 32 bit or 64 is for, is it like the 32 instructions (i can't really find the word for it) it can do per cycle?
 
Uh. 32 bit processors can do an operation on a binary number 32 digits long, i.e. add two 32-bit integers. By the way, no one can really do 32 instructions per clock - athlon/pentiums only get around 2 instructions per cycle, whereas the powerpc cpu for mac g4's get around 3 per cycle (that's due to having a different instruction set though). Ace's hardware and Ars technica have pretty good reviews of what 64-bit processing actually entails - it's a bit hard to sum up, so you're encouraged to go there and read for yourself :D
 
well the primary difference I would say is its greater memory allowance...but as tourmal said, get yourself to arstechnica - they will tell you all you need to know about these architectures
 
There was an article on the front page some time ago that explained it very well

let's take the calculation 2+2 in a 32-bit processor it looks like this

0000000000000000000000000000010
0000000000000000000000000000010=
0000000000000000000000000000100

but in 64-bit it looks like this

0000000000000000000000000000000000000000000000000000000000000010
0000000000000000000000000000000000000000000000000000000000000010=
0000000000000000000000000000000000000000000000000000000000000100

64-bit can do much longer calculations but the way AMD is implementing it they can make it calculate both 32-bit and 64-bit calculations
 
Thankfully though, it won't take any longer to mess with all those zeroes (if everything else remains constant) since it was designed to take a bigger 'mouthfull' of data at once.

Also, AMD is adding additional registers to their chip. This may or may not have a significant impact on speed. If you were to think of registers as variables, then the A64 can keep more variables directly in it's calculatation lines. Though wether this spiffy addition does much remains to be seen when 64 bit compiled apps come out.

JigPu
 
It should be noted that while the actual execution time doesn't take any longer (compared to the Athlon), it *does* take up more cache to store a 64-bit data type vs a 32-bit data type and also takes more memory bandwidth to transfer it. So all those extra zero's will eat up cache space and memory bandwidth. Luckily, AMD forsaw this and have set the default integer size on x86-64 to 32-bit. *However*, the memory pointer size (memory address) is 64-bit in all 64-bit modes (including compatibility) unless specifically told otherwise, so memory addresses will take up extra cache space and memory bandwidth.
 
Back