static void
vmx128_combine_atop_ca (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
int i;
__vector4 vdest, vsrc, vmask, vsrca;
__vector4 tmp1, tmp2, tmp3, tmp4, edges,
dest_mask, mask_mask, src_mask, store_mask;
COMPUTE_SHIFT_MASKC (dest, src, mask);
/* printf ("%s\n",__PRETTY_FUNCTION__); */
for (i = width / 4; i > 0; i--)
{
LOAD_VECTORSC (dest, src, mask);
vsrca = splat_alpha (vsrc);
vsrc = pix_multiply (vsrc, vmask);
vmask = pix_multiply (vmask, vsrca);
vdest = pix_add_mul (vsrc, splat_alpha (vdest),
negate (vmask), vdest);
STORE_VECTOR (dest);
src += 4;
dest += 4;
mask += 4;
}
for (i = width % 4; --i >= 0;)
{
uint32_t a = mask[i];
uint32_t s = src[i];
uint32_t d = dest[i];
uint32_t sa = ALPHA_8 (s);
uint32_t da = ALPHA_8 (d);
UN8x4_MUL_UN8x4 (s, a);
UN8x4_MUL_UN8 (a, sa);
UN8x4_MUL_UN8x4_ADD_UN8x4_MUL_UN8 (d, ~a, s, da);
dest[i] = d;
}
}
static void
vmx128_combine_out_u_no_mask (uint32_t * dest,
const uint32_t *src,
int width)
{
int i;
__vector4 vdest, vsrc;
__vector4 tmp1, tmp2, tmp3, tmp4, edges,
dest_mask, src_mask, store_mask;
COMPUTE_SHIFT_MASKS (dest, src);
/* printf ("%s\n",__PRETTY_FUNCTION__); */
for (i = width / 4; i > 0; i--)
{
LOAD_VECTORS (dest, src);
vdest = pix_multiply (vsrc, splat_alpha (negate (vdest)));
STORE_VECTOR (dest);
src += 4;
dest += 4;
}
for (i = width % 4; --i >= 0;)
{
uint32_t s = src[i];
uint32_t a = ALPHA_8 (~dest[i]);
UN8x4_MUL_UN8 (s, a);
dest[i] = s;
}
}
static void
vmx_combine_in_reverse_u_no_mask (uint32_t * dest,
const uint32_t *src,
int width)
{
int i;
vector unsigned int vdest, vsrc;
vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
dest_mask, src_mask, store_mask;
COMPUTE_SHIFT_MASKS (dest, src);
/* printf ("%s\n",__PRETTY_FUNCTION__); */
for (i = width / 4; i > 0; i--)
{
LOAD_VECTORS (dest, src);
vdest = pix_multiply (vdest, splat_alpha (vsrc));
STORE_VECTOR (dest);
src += 4;
dest += 4;
}
for (i = width % 4; --i >= 0;)
{
uint32_t d = dest[i];
uint32_t a = ALPHA_8 (src[i]);
UN8x4_MUL_UN8 (d, a);
dest[i] = d;
}
}
static void
vmx128_combine_atop_reverse_u_mask (uint32_t * dest,
const uint32_t *src,
const uint32_t *mask,
int width)
{
int i;
__vector4 vdest, vsrc, vmask;
__vector4 tmp1, tmp2, tmp3, tmp4, edges,
dest_mask, src_mask, mask_mask, store_mask;
COMPUTE_SHIFT_MASKC (dest, src, mask);
/* printf ("%s\n",__PRETTY_FUNCTION__); */
for (i = width / 4; i > 0; i--)
{
LOAD_VECTORSM (dest, src, mask);
vdest = pix_add_mul (vdest, splat_alpha (vsrc),
vsrc, splat_alpha (negate (vdest)));
STORE_VECTOR (dest);
src += 4;
dest += 4;
mask += 4;
}
for (i = width % 4; --i >= 0;)
{
uint32_t m = ALPHA_8 (mask[i]);
uint32_t s = src[i];
uint32_t d = dest[i];
uint32_t src_a;
uint32_t dest_ia = ALPHA_8 (~d);
UN8x4_MUL_UN8 (s, m);
src_a = ALPHA_8 (s);
UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8 (s, dest_ia, d, src_a);
dest[i] = s;
}
}
static void
vmx_combine_xor_u_no_mask (uint32_t * dest,
const uint32_t *src,
int width)
{
int i;
vector unsigned int vdest, vsrc;
vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
dest_mask, src_mask, store_mask;
COMPUTE_SHIFT_MASKS (dest, src);
/* printf ("%s\n",__PRETTY_FUNCTION__); */
for (i = width / 4; i > 0; i--)
{
LOAD_VECTORS (dest, src);
vdest = pix_add_mul (vsrc, splat_alpha (negate (vdest)),
vdest, splat_alpha (negate (vsrc)));
STORE_VECTOR (dest);
src += 4;
dest += 4;
}
for (i = width % 4; --i >= 0;)
{
uint32_t s = src[i];
uint32_t d = dest[i];
uint32_t src_ia = ALPHA_8 (~s);
uint32_t dest_ia = ALPHA_8 (~d);
UN8x4_MUL_UN8_ADD_UN8x4_MUL_UN8 (s, dest_ia, d, src_ia);
dest[i] = s;
}
}
static void
vmx_combine_out_reverse_ca (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
int i;
vector unsigned int vdest, vsrc, vmask;
vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
dest_mask, mask_mask, src_mask, store_mask;
COMPUTE_SHIFT_MASKC (dest, src, mask);
/* printf ("%s\n",__PRETTY_FUNCTION__); */
for (i = width / 4; i > 0; i--)
{
LOAD_VECTORSC (dest, src, mask);
vdest = pix_multiply (
vdest, negate (pix_multiply (vmask, splat_alpha (vsrc))));
STORE_VECTOR (dest);
src += 4;
dest += 4;
mask += 4;
}
for (i = width % 4; --i >= 0;)
{
uint32_t a = mask[i];
uint32_t s = src[i];
uint32_t d = dest[i];
uint32_t sa = ALPHA_8 (s);
UN8x4_MUL_UN8 (a, sa);
UN8x4_MUL_UN8x4 (d, ~a);
dest[i] = d;
}
}
static void
vmx128_combine_over_reverse_ca (pixman_implementation_t *imp,
pixman_op_t op,
uint32_t * dest,
const uint32_t * src,
const uint32_t * mask,
int width)
{
int i;
__vector4 vdest, vsrc, vmask;
__vector4 tmp1, tmp2, tmp3, tmp4, edges,
dest_mask, mask_mask, src_mask, store_mask;
COMPUTE_SHIFT_MASKC (dest, src, mask);
/* printf("%s\n",__PRETTY_FUNCTION__); */
for (i = width / 4; i > 0; i--)
{
LOAD_VECTORSC (dest, src, mask);
vdest = over (vdest, splat_alpha (vdest), pix_multiply (vsrc, vmask));
STORE_VECTOR (dest);
mask += 4;
src += 4;
dest += 4;
}
for (i = width % 4; --i >= 0;)
{
uint32_t a = mask[i];
uint32_t s = src[i];
uint32_t d = dest[i];
uint32_t ida = ALPHA_8 (~d);
UN8x4_MUL_UN8x4 (s, a);
UN8x4_MUL_UN8_ADD_UN8x4 (s, ida, d);
dest[i] = s;
}
}
static void
vmx_combine_over_u_mask (uint32_t * dest,
const uint32_t *src,
const uint32_t *mask,
int width)
{
int i;
vector unsigned int vdest, vsrc, vmask;
vector unsigned char tmp1, tmp2, tmp3, tmp4, edges,
dest_mask, src_mask, mask_mask, store_mask;
COMPUTE_SHIFT_MASKC (dest, src, mask);
/* printf ("%s\n",__PRETTY_FUNCTION__); */
for (i = width / 4; i > 0; i--)
{
LOAD_VECTORSM (dest, src, mask);
vdest = over (vsrc, splat_alpha (vsrc), vdest);
STORE_VECTOR (dest);
src += 4;
dest += 4;
mask += 4;
}
for (i = width % 4; --i >= 0;)
{
uint32_t m = ALPHA_8 (mask[i]);
uint32_t s = src[i];
uint32_t d = dest[i];
uint32_t ia;
UN8x4_MUL_UN8 (s, m);
ia = ALPHA_8 (~s);
UN8x4_MUL_UN8_ADD_UN8x4 (d, ia, s);
dest[i] = d;
}
}
static void
vmx128_combine_in_reverse_u_mask (uint32_t * dest,
const uint32_t *src,
const uint32_t *mask,
int width)
{
int i;
__vector4 vdest, vsrc, vmask;
__vector4 tmp1, tmp2, tmp3, tmp4, edges,
dest_mask, src_mask, mask_mask, store_mask;
COMPUTE_SHIFT_MASKC (dest, src, mask);
/* printf ("%s\n",__PRETTY_FUNCTION__); */
for (i = width / 4; i > 0; i--)
{
LOAD_VECTORSM (dest, src, mask);
vdest = pix_multiply (vdest, splat_alpha (vsrc));
STORE_VECTOR (dest);
src += 4;
dest += 4;
mask += 4;
}
for (i = width % 4; --i >= 0;)
{
uint32_t m = ALPHA_8 (mask[i]);
uint32_t d = dest[i];
uint32_t a = src[i];
UN8x4_MUL_UN8 (a, m);
a = ALPHA_8 (a);
UN8x4_MUL_UN8 (d, a);
dest[i] = d;
}
}
static void
vmx128_combine_over_u_no_mask (uint32_t * dest,
const uint32_t *src,
int width)
{
int i;
__vector4 vdest, vsrc;
__vector4 tmp1, tmp2, tmp3, tmp4, edges,
dest_mask, src_mask, store_mask;
COMPUTE_SHIFT_MASKS (dest, src);
/* printf ("%s\n",__PRETTY_FUNCTION__); */
for (i = width / 4; i > 0; i--)
{
LOAD_VECTORS (dest, src);
vdest = over (vsrc, splat_alpha (vsrc), vdest);
STORE_VECTOR (dest);
src += 4;
dest += 4;
}
for (i = width % 4; --i >= 0;)
{
uint32_t s = src[i];
uint32_t d = dest[i];
uint32_t ia = ALPHA_8 (~s);
UN8x4_MUL_UN8_ADD_UN8x4 (d, ia, s);
dest[i] = d;
}
}