http://www.snpx.com/cgi-bin/infosyssec.cgi?target=chapter13.htm
Hey guys, I've been reading the book in the link above. I've read my way up to Chapter 14, FPUs and Multimedia, where I got quite confused. I hope someone here can help me understand.
Firstly, I thought CPUs only processed numbers with two bases, a 0 and 1. I am a bit confused as to why the CPU processes numbers in Standard Form. I thought an electric impulse could either be off or on, so how are standard form numbers transmitted?
Secondly, if the CPU only processes 0s and 1s how does a Floating Point Unit Work?
Thirdly, what exactly are the SSE and MMX extensions. How do they help the ALU and FPU?
-
Then to add two numbers they add the whole 32-bits to each other even if it's 2+1 (you can specify the number of bits though)
1.56787 will be stored this way :
156787 will be stored and a sign bit will be kept.(integer form)
the exponentiation will be kept in another number I don't really remember but I think in this case they will keep -5 (in 2's compliment format)
One of the formats :
Signbit - 1 bits
Exponentiation - 7 bits
The integer form - 24 bits
total : 32-bits
SSE which has spinned 4 [or five] versions does the calculations in a vector format.For example to add 4 numbers to another 4 numbers you just have to fill all the numbers into the registers and give a single add instruction.The 4 add operations will be done. -
Thanks mujtaba
.
FPUs, ALUs, MMX and SSE Extensions
Discussion in 'Hardware Components and Aftermarket Upgrades' started by Lil Mayz, May 12, 2007.