Cheat Engine

Stylo · Posted: Tue Jul 17, 2012 1:04 pm Post subject: Floating point help

Hey guys,
I'm having some troubles understanding the floating point system.
What are those 8 registers (st0,st1. . .) i couldn't get it Confused

can any of you point me to a good explanation for floating point system? or even better can explain that?

Thanks a lot :]

Slugsnack · Posted: Fri Jul 20, 2012 7:11 pm Post subject:

exactly where does your problem lie with your understanding? in x86, there are 8 floating point registers (st0 to st7). in general with the fpu registers, you need to push them onto the fpu stack with some instruction such as fld. then you perform a floating point operation which operates on the top register/s by default or specific registers if you mandate so in the operation operands. for example, if you pushed 1.5 then 2.5 then st0 is 2.5 and st1 is 1.5. as with all stacks, there is the notion of FILO.

Stylo · Posted: Fri Jul 20, 2012 11:26 pm Post subject:

There are few things i can't get, is there another stack for the fpu? or the stack is the group of stX registers?
and when you push 1.5 and 2.5 right after, how does it calculate and knows what comes before the dot and what comes after?

Slugsnack · Posted: Sat Jul 21, 2012 3:58 am Post subject:

yes the stack is a separate one. the numbers are not actually pushed as 1.5 and 2.5. instead they are approximated by a mantissa (also known as significand) and an exponent. see this well known article:
http://docs.oracle.com/cd/E19957-01/806-3568/ncg_goldberg.html

alternatively, back in the day i learnt about the fpu from here:
http://216.92.238.133/Webster/www.artofasm.com/Windows/HTML/MoreDataRepresentation.html#1000036

the conversion to this 'representation' is done by the compiler, specifically the assembler

Stylo · Posted: Tue Jul 24, 2012 1:35 pm Post subject:

As always Slugsnack, you always have the answer Smile

Thanks. i'll be reading that and will ask you if it'd be necessary.