bzoli22

typedef
__attribute__ ((aligned(16)))
float Vector[1024];

void add( Vector a, Vector b, Vector c)
{
  for ( int i=0; i<1024; i++ )
    a[i] = b[i] + c[i];
}

bzoli22

_add:
	pushl	%ebp
	movl	%esp, %ebp
	pushl	%esi
	pushl	%ebx
	movl	8(%ebp), %ebx
	movl	12(%ebp), %ecx
	movl	16(%ebp), %edx
	leal	16(%ebx), %esi
	testb	$15, %bl

bzoli22

L7:
	movss	(%ecx,%eax,4), %xmm0
	addss	(%edx,%eax,4), %xmm0
	movss	%xmm0, (%ebx,%eax,4)
	incl	%eax
	cmpl	$1024, %eax
	jne	L7

bzoli22

void add( float a[1024], float b[1024], float c[1024])
{
	for ( int i=0; i<1024; i++ )
		a[i] =  b[i] + c[i];
}

bzoli22

_add:
	pushl	%ebp
	movl	%esp, %ebp
	xorl	%eax, %eax
	.p2align 2,,3
L2:
	movaps	_b(%eax), %xmm0
	addps	_c(%eax), %xmm0
	movaps	%xmm0, _a(%eax)
	addl	$16, %eax

bzoli22

static float a[1024], b[1024], c[1024];

void add()
{
	for ( int i=0; i<1024; i++ )
		a[i] = b[i] + c[i];
}

bzoli22

_create_vector:
	movss	8(%esp), %xmm1
	movss	4(%esp), %xmm0
	unpcklps	%xmm1, %xmm0
	movss	16(%esp), %xmm2
	movss	12(%esp), %xmm1
	unpcklps	%xmm2, %xmm1
	movlhps	%xmm1, %xmm0
	ret

bzoli22

_create_vector:
	movss	8(%esp), %xmm0
	movss	4(%esp), %xmm2
	unpcklps	%xmm0, %xmm2
	movaps	%xmm2, %xmm1
	movss	16(%esp), %xmm0
	movss	12(%esp), %xmm2
	unpcklps	%xmm0, %xmm2
	movaps	%xmm2, %xmm0
	movaps	%xmm1, %xmm2

bzoli22

typedef float v4t __attribute__ ((vector_size(16), aligned(16)));

v4t create_vector(float a, float b, float c, float d)
{
	v4t v4 = {a, b, c, d};
	return v4;
}

bzoli22

typedef float v3t __attribute__ ((vector_size(12), aligned(16)));

v3t addv3( v3t a, v3t b)
{
	return a + b;
}