Replace FPU-instructions to SSE-instructions

[Mikl__]

Hi, All !

• What instruction replaced the FPU-instruction FCHS? I use

Code Select Expand

xorps xmm0, xmm0
subss xmm0, xmm1; in the register xmm1 the value for which you want to change the sign
movss xmm1, xmm0or so

Code Select Expand

xorps xmm0, const80000000hBut is that right?

• The FPU was many trigonometric instructions (FSIN, FCOS, FSINCOS, FPTAN, FPATAN) What it replace in SSE?

Thank you

[hutch--]

I have seen formulas around the internet for replacing trig functions with sse2. The attached PDF is what I have found so far.

[TouEnMasm]

The answer must be in the CRT using /Fa to get an asm translate.
Perhaps also is there some options to switch on in VS project.

[Siekmanski]

Floating point has 1 sign bit.
So we can flip the sign bit using boolean algebra ( Exclusive OR )

pxor / xorps / xorpd

.const
flip_bit dd 080000000H,080000000H,080000000H,080000000H

.code

xorps xmm0,dword ptr [flip_bit]
4 at once:
xorps xmm0,oword ptr [flip_bit]

square root functions in SSE:

SQRTSD Compute Square Root of Scalar Double-Precision Floating-Point Value
SQRTSS Compute Square Root of Scalar Single-Precision Value
SQRTPD Square Root of Packed Double-Precision Floating-Point Values
SQRTPS Square Root of Packed Single-Precision Floating-Point Values

RSQRTPS Compute Reciprocals of Square Roots of Packed Single-Precision Floating-Point Values
RSQRTSS Compute Reciprocal of Square Root of Scalar Single-Precision Floating-Point Value

Other trigonometric functions you have to create yourself.

for example: http://masm32.com/board/index.php?topic=4118.msg49276#msg49276

[Mikl__]

Hi, hutch--, TouEnMasm, Siekmanski
Thank you very much

[daydreamer]

well sometimes you can eliminate change sign if you have mulps later in calculation
constants or variables for example taylor series use 1/3!,-1/5!,1/7!,-1/9!
use mulps -100.0, instead change sign and mulps 100.0

[Mikl__]

Hi, daydreamer!
I will use your advice. Thank!