OSDev.org

hi guys, i would like to know where can I find some docs "intel/amd" about operand size and how much clock cicle that operand takes.

as my english is so poor I´m trying to have a help with this forum

my question is something like that:

wich is better?

mov eax, _xxx_handler
call eax

----- or -----

call _xxx_handler

?

I saw, don´t know where that call addr takes about 20 cicles os clock

and call reg, takes only 2-4 bytes

is that right ?

The second, in about any case, since it's a fixed jump. The first might be seen as an unpredictable jump and in worst-case could cause a pipeline flush, making for about 30 cycles of delay on a NetBurst-cpu.

Instructions don't have a cycle count anymore on current machines, they have, at best, a cycle latency. They're too unpredictable and unreliable to use for any sort of instruction ordering. Why do you care about sub-nanosecond offsets by the way?

You may want to look at "Intel Architectecture Optimization Reference Manual". I don't know if AMD has simular manual.

For AMD: the top two hits at google: http://www.google.nl/search?client=fire ... gle+zoeken

tnx so much