Jon Berndt writes:
>
> Are there any reasons to choose one method over the other in the declaration
> of local variables that are used each frame?
>
> Here are some examples. Note that this hypothetical routine would be called
> each pass through the EOM (for example):
>
< following use local variables for local temporaries >
> ============= Method 1)
> ============= Method 2)
These should be equivalant
< following uses class variables for local temporaries >
> ============= Method 3)
This will use a memory store to the class variable which may or may not
happen when using local variables.
i.e an optimizing compiler will just use register 'space' for the temporaries
if fpu stack or register 'space' permits
Of course the only way to really tell is to profile
HTH
Norman
==== cut jon.cpp ====
#include <math.h>
extern float abc;
float testa(float a, float b)
{
abc = sin(a);
return abc + b;
}
float testb(float a, float b)
{
float abc = sin(a);
return abc + b;
}
=== cut resultant assembly for gcc -O2 jon.cpp ===
.file "jon.cpp"
.text
.align 2
.p2align 4,,15
.globl __Z5testaff
.def __Z5testaff; .scl 2; .type 32; .endef
__Z5testaff:
pushl %ebp
movl %esp, %ebp
subl $24, %esp
flds 8(%ebp)
fstpl (%esp)
call _sin
fstps -4(%ebp)
flds -4(%ebp)
fsts _abc
fadds 12(%ebp)
movl %ebp, %esp
popl %ebp
ret
.align 2
.p2align 4,,15
.globl __Z5testbff
.def __Z5testbff; .scl 2; .type 32; .endef
__Z5testbff:
pushl %ebp
movl %esp, %ebp
subl $24, %esp
flds 8(%ebp)
fstpl (%esp)
call _sin
fstps -4(%ebp)
flds -4(%ebp)
fadds 12(%ebp)
movl %ebp, %esp
popl %ebp
ret
.def _sin; .scl 2; .type 32; .endef
_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
