AMD Typewriter x86 사용자 설명서
64
Replace Branches with Computation in 3DNow!™ Code
AMD Athlon™ Processor x86 Code Optimization
22007E/0—November 1999
Example 5:
C code:
#define PI 3.14159265358979323
float x,y,xa,ya,r,res;
int xs,df;
xs = x < 0 ? 1 : 0;
xa = fabs(x);
ya = fabs(y);
df = (xa < ya);
if (xs && df) {
res = PI/2 + r;
}
else if (xs) {
res = PI - r;
}
else if (df) {
res = PI/2 - r;
}
else {
res = r;
}
float x,y,xa,ya,r,res;
int xs,df;
xs = x < 0 ? 1 : 0;
xa = fabs(x);
ya = fabs(y);
df = (xa < ya);
if (xs && df) {
res = PI/2 + r;
}
else if (xs) {
res = PI - r;
}
else if (df) {
res = PI/2 - r;
}
else {
res = r;
}
3DNow! code:
;in:
MM0 = r
;
MM1 = y
;
MM2 = x
;out: MM0 = res
MOVQ
MOVQ
MM7, sgn
;mask to extract sign bit
MOVQ
MM6, sgn
;mask to extract sign bit
MOVQ
MM5, mabs
;mask to clear sign bit
PAND
MM7, MM2
;xs = sign(x)
PAND
MM1, MM5
;ya = abs(y)
PAND
MM2, MM5
;xa = abs(x)
MOVQ
MM6, MM1
;y
PCMPGTD MM6, MM2
;df = (xa < ya) ? 0xffffffff : 0
PSLLD
MM6, 31
;df = bit<31>
MOVQ
MM5, MM7
;xs
PXOR
MM7, MM6
;xs^df ? 0x80000000 : 0
MOVQ
MM3, npio2
;-pi/2
PXOR
MM5, MM3
;xs ? pi/2 : -pi/2
PSRAD
MM6, 31
;df ? 0xffffffff : 0
PANDN
MM6, MM5
;xs ? (df ? 0 : pi/2) : (df ? 0 : -pi/2)
PFSUB
MM6, MM3
;pr = pi/2 + (xs ? (df ? 0 : pi/2) :
; (df ? 0 : -pi/2))
; (df ? 0 : -pi/2))
POR
MM0, MM7
;ar = xs^df ? -r : r
PFADD
MM0, MM6
;res = ar + pr