Cheat Engine

[supercharger]

i saw in a game (developed after year2k) a lot of codes like these
fld dword ptr [xxxxxxxx] - which have the same output as fld1 or fldz or fldpi.
i wonder why they don't use fld1 (or fldz or fldpi) , which are shorter and don't require reading memory.

between fld dword ptr [xxxxxxx] and fld1, which as higher efficiency?
i have been writing codes and i prefer fld1 which is easier to read, unless fld1 costs more CPU time.

another question:
as CPU and FPU can work simultaneously, should i write codes this way:
mov eax,[xxxxxxx]
fadd dword ptr [ecx]
fsub dword ptr [eax]

or this way:
fadd dword ptr [ecx]
mov eax,[xxxxxxx]
fsub dword ptr [eax]

this is just an example of arranging the order of instructions, both ways are not very different in this case. but you can get the idea what i mean.

[justa_dude]

Yes, fld1/pi/0 are faster than loading a real from memory. But loading from memory allows you to grab arbitrary values - not just constants.

With regard to your second question, I don't think that either code does what you're expecting. Values have to be loaded and copied/popped onto and off of the FPU stack, and I don't see you moving your values off in either snippet.

Cheers,
adude

[Flyte]

It's also a pointless optimization for a game, as the bottleneck is usually the graphics card.

[Slugsnack]

to your second question, it probably doesn't make any difference since your cpu pipelines and takes advantage of out of order execution

[supercharger]

[Flyte]