1 files changed, 0 insertions, 315 deletions
diff --git a/libraries/evas/src/lib/engines/common/evas_op_blend/op_blend_pixel_sse3.c b/libraries/evas/src/lib/engines/common/evas_op_blend/op_blend_pixel_sse3.c
deleted file mode 100644
index 69c597c..0000000
--- a/libraries/evas/src/lib/engines/common/evas_op_blend/op_blend_pixel_sse3.c
+++ /dev/null
@@ -1,315 +0,0 @@
-/* blend pixel --> dst */
-#ifdef BUILD_SSE3
-static void
-_op_blend_p_dp_sse3(DATA32 *s, DATA8 *m __UNUSED__, DATA32 c __UNUSED__, DATA32 *d, int l) {
-   LOOP_ALIGNED_U1_A48_SSE3(d, l,
-      { /* UOP */
-         int alpha = 256 - (*s >> 24);
-         *d = *s + MUL_256(alpha, *d);
-         s++; d++; l--;
-      },
-      { /* A4OP */
-         __m128i s0 = _mm_lddqu_si128((__m128i *)s);
-         __m128i d0 = _mm_load_si128((__m128i *)d);
-         __m128i a0 = sub4_alpha_sse3(s0);
-         __m128i mul0 = mul_256_sse3(a0, d0);
-         d0 = _mm_add_epi32(mul0, s0);
-         _mm_store_si128((__m128i *)d, d0);
-         s += 4; d += 4; l -= 4;
-      },
-      { /* A8OP */
-         __m128i s0 = _mm_lddqu_si128((__m128i *)s);
-         __m128i d0 = _mm_load_si128((__m128i *)d);
-         __m128i s1 = _mm_lddqu_si128((__m128i *)(s+4));
-         __m128i d1 = _mm_load_si128((__m128i *)(d+4));
-         __m128i a0 = sub4_alpha_sse3(s0);
-         __m128i a1 = sub4_alpha_sse3(s1);
-         __m128i mul0 = mul_256_sse3(a0, d0);
-         __m128i mul1 = mul_256_sse3(a1, d1);
-         d0 = _mm_add_epi32(mul0, s0);
-         d1 = _mm_add_epi32(mul1, s1);
-         _mm_store_si128((__m128i *)d, d0);
-         _mm_store_si128((__m128i *)(d+4), d1);
-         s += 8; d += 8; l -= 8;
-      })
-}
-static void
-_op_blend_pas_dp_sse3(DATA32 *s, DATA8 *m __UNUSED__, DATA32 c __UNUSED__, DATA32 *d, int l) {
-   int alpha;
-   const __m128i zero = _mm_setzero_si128();
-   LOOP_ALIGNED_U1_A48_SSE3(d, l,
-      { /* UOP */
-         switch (*s & 0xff000000)
-           {
-           case 0:
-              break;
-           case 0xff000000:
-              *d = *s;
-              break;
-           default:
-              alpha = 256 - (*s >> 24);
-              *d = *s + MUL_256(alpha, *d);
-              break;
-           }
-         s++;  d++; l--;
-      },
-      { /* A4OP */
-         __m128i s0 = _mm_lddqu_si128((__m128i *)s);
-         __m128i d0 = _mm_load_si128((__m128i *)d);
-         __m128i a0 = sub4_alpha_sse3(s0);
-         __m128i mul0 = mul_256_sse3(a0, d0);
-         mul0 = _mm_add_epi32(s0, mul0);
-         __m128i zmask0 = _mm_cmpeq_epi32(_mm_srli_epi32(s0, 24), zero);
-         __m128i imask0  = ~zmask0;
-         mul0 = _mm_and_si128(imask0, mul0);
-         d0 = _mm_and_si128(zmask0, d0);
-         d0 = _mm_add_epi32(mul0, d0);
-         _mm_store_si128((__m128i *)d, d0);
-        s += 4; d += 4; l -= 4;
-      },
-      { /* A8OP */
-         __m128i s0 = _mm_lddqu_si128((__m128i *)s);
-         __m128i d0 = _mm_load_si128((__m128i *)d);
-         __m128i s1 = _mm_lddqu_si128((__m128i *)(s+4));
-         __m128i d1 = _mm_load_si128((__m128i *)(d+4));
-         __m128i a0 = sub4_alpha_sse3(s0);
-         __m128i a1 = sub4_alpha_sse3(s1);
-         __m128i mul0 = mul_256_sse3(a0, d0);
-         __m128i mul1 = mul_256_sse3(a1, d1);
-         mul0 = _mm_add_epi32(s0, mul0);
-         mul1 = _mm_add_epi32(s1, mul1);
-         __m128i zmask0 = _mm_cmpeq_epi32(_mm_srli_epi32(s0, 24), zero);
-         __m128i zmask1 = _mm_cmpeq_epi32(_mm_srli_epi32(s1, 24),  zero);
-         __m128i imask0 = ~zmask0;
-         __m128i imask1 = ~zmask1;
-         mul0 = _mm_and_si128(imask0, mul0);
-         d0 = _mm_and_si128(zmask0, d0);
-         mul1 = _mm_and_si128(imask1, mul1);
-         d1 = _mm_and_si128(zmask1, d1);
-         d0 = _mm_add_epi32(mul0, d0);
-         d1 = _mm_add_epi32(mul1, d1);
-         _mm_store_si128((__m128i *)d, d0);
-         _mm_store_si128((__m128i *)(d+4), d1);
-         s += 8; d += 8; l -= 8;
-      })
-}
-#define _op_blend_pan_dp_sse3 NULL
-#define _op_blend_p_dpan_sse3 _op_blend_p_dp_sse3
-#define _op_blend_pas_dpan_sse3 _op_blend_pas_dp_sse3
-#define _op_blend_pan_dpan_sse3 _op_blend_pan_dp_sse3
-static void
-init_blend_pixel_span_funcs_sse3(void)
-{
-   op_blend_span_funcs[SP][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_p_dp_sse3;
-   op_blend_span_funcs[SP_AS][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_pas_dp_sse3;
-   op_blend_span_funcs[SP_AN][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_pan_dp_sse3;
-// FIXME: BUGGY BUGGY Core i5 750 (32bit), 4.5.2 (Ubuntu/Linaro 4.5.2-8ubuntu4), ello (text and rectangle)
-//   op_blend_span_funcs[SP][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_p_dpan_sse3;
-   op_blend_span_funcs[SP_AS][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_pas_dpan_sse3;
-   op_blend_span_funcs[SP_AN][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_pan_dpan_sse3;
-}
-#define _op_blend_pt_p_dp_sse3 NULL
-#define _op_blend_pt_pas_dp_sse3 _op_blend_pt_p_dp_sse3
-#define _op_blend_pt_pan_dp_sse3 NULL
-#define _op_blend_pt_p_dpan_sse3 _op_blend_pt_p_dp_sse3
-#define _op_blend_pt_pan_dpan_sse3 _op_blend_pt_pan_dp_sse3
-#define _op_blend_pt_pas_dpan_sse3 _op_blend_pt_pas_dp_sse3
-static void
-init_blend_pixel_pt_funcs_sse3(void)
-{
-   op_blend_pt_funcs[SP][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_pt_p_dp_sse3;
-   op_blend_pt_funcs[SP_AS][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_pt_pas_dp_sse3;
-   op_blend_pt_funcs[SP_AN][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_pt_pan_dp_sse3;
-   op_blend_pt_funcs[SP][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_pt_p_dpan_sse3;
-   op_blend_pt_funcs[SP_AS][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_pt_pas_dpan_sse3;
-   op_blend_pt_funcs[SP_AN][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_pt_pan_dpan_sse3;
-}
-/*-----*/
-/* blend_rel pixel -> dst */
-static void
-_op_blend_rel_p_dp_sse3(DATA32 *s, DATA8 *m __UNUSED__, DATA32 c, DATA32 *d, int l) {
-   const __m128i ones = _mm_set_epi32(1, 1, 1, 1);
-   LOOP_ALIGNED_U1_A48_SSE3(d, l,
-      { /* UOP */
-         int alpha = 256 - (*s >> 24);
-         c = 1 + (*d >> 24);
-         *d = MUL_256(c, *s) + MUL_256(alpha, *d);
-         d++; s++; l--;
-      },
-      { /*A4OP */
-         __m128i s0 = _mm_lddqu_si128((__m128i *)s);
-         __m128i d0 = _mm_load_si128((__m128i *)d);
-         __m128i c0 = _mm_add_epi32(_mm_srli_epi32(d0, 24), ones);
-         __m128i a0 = sub4_alpha_sse3(s0);
-         d0 = _mm_add_epi32(mul_256_sse3(c0, s0), mul_256_sse3(a0, d0));
-         _mm_store_si128((__m128i *)d, d0);
-         d += 4; s += 4; l -= 4;
-      },
-      { /* A8OP */
-         __m128i s0 = _mm_lddqu_si128((__m128i *)s);
-         __m128i d0 = _mm_load_si128 ((__m128i *)d);
-         __m128i s1 = _mm_lddqu_si128((__m128i *)(s+4));
-         __m128i d1 = _mm_load_si128 ((__m128i *)(d+4));
-         __m128i c0 = _mm_add_epi32(_mm_srli_epi32(d0, 24), ones);
-         __m128i c1 = _mm_add_epi32(_mm_srli_epi32(d1, 24), ones);
-         __m128i a0 = sub4_alpha_sse3(s0);
-         __m128i a1 = sub4_alpha_sse3(s1);
-         d0 = _mm_add_epi32(mul_256_sse3(c0, s0), mul_256_sse3(a0, d0));
-         d1 = _mm_add_epi32(mul_256_sse3(c1, s1), mul_256_sse3(a1, d1));
-         _mm_store_si128((__m128i *)d, d0);
-         _mm_store_si128((__m128i *)(d+4), d1);
-         d += 8; s += 8; l -= 8;
-      })
-}
-static void
-_op_blend_rel_pan_dp_sse3(DATA32 *s, DATA8 *m __UNUSED__, DATA32 c, DATA32 *d, int l) {
-   const __m128i ones = _mm_set_epi32(1, 1, 1, 1);
-   LOOP_ALIGNED_U1_A48_SSE3(d, l,
-      { /* UOP */
-         c = 1 + (*d >> 24);
-         *d++ = MUL_256(c, *s);
-         s++; l--;
-      },
-      { /* A4OP */
-         __m128i s0 = _mm_lddqu_si128((__m128i *)s);
-         __m128i d0 = _mm_load_si128((__m128i *)d);
-         __m128i c0 = _mm_add_epi32(_mm_srli_epi32(d0, 24), ones);
-         d0 = mul_256_sse3(c0, s0);
-         _mm_store_si128((__m128i *)d, d0);
-         d += 4; s += 4; l -= 4;
-      },
-      { /* A8OP */
-         __m128i s0 = _mm_lddqu_si128((__m128i *)s);
-         __m128i d0 = _mm_load_si128 ((__m128i *)d);
-         __m128i s1 = _mm_lddqu_si128((__m128i *)(s+4));
-         __m128i d1 = _mm_load_si128 ((__m128i *)(d+4));
-         __m128i c0 = _mm_add_epi32(_mm_srli_epi32(d0, 24), ones);
-         __m128i c1 = _mm_add_epi32(_mm_srli_epi32(d1, 24), ones);
-         d0 = mul_256_sse3(c0, s0);
-         d1 = mul_256_sse3(c1, s1);
-         _mm_store_si128((__m128i *)d, d0);
-         _mm_store_si128((__m128i *)(d+4), d1);
-         d += 8; s += 8; l -= 8;
-      })
-}
-#define _op_blend_rel_pas_dp_sse3 _op_blend_rel_p_dp_sse3
-#define _op_blend_rel_p_dpan_sse3 _op_blend_p_dpan_sse3
-#define _op_blend_rel_pan_dpan_sse3 _op_blend_pan_dpan_sse3
-#define _op_blend_rel_pas_dpan_sse3 _op_blend_pas_dpan_sse3
-static void
-init_blend_rel_pixel_span_funcs_sse3(void)
-{
-   op_blend_rel_span_funcs[SP][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_p_dp_sse3;
-   op_blend_rel_span_funcs[SP_AS][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_pas_dp_sse3;
-   op_blend_rel_span_funcs[SP_AN][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_pan_dp_sse3;
-   op_blend_rel_span_funcs[SP][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_p_dpan_sse3;
-   op_blend_rel_span_funcs[SP_AS][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_pas_dpan_sse3;
-   op_blend_rel_span_funcs[SP_AN][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_pan_dpan_sse3;
-}
-#define _op_blend_rel_pt_p_dp_sse3 NULL
-#define _op_blend_rel_pt_pan_dp_sse3 NULL
-#define _op_blend_rel_pt_pas_dp_sse3 _op_blend_rel_pt_p_dp_sse3
-#define _op_blend_rel_pt_p_dpan_sse3 _op_blend_pt_p_dpan_sse3
-#define _op_blend_rel_pt_pan_dpan_sse3 _op_blend_pt_pan_dpan_sse3
-#define _op_blend_rel_pt_pas_dpan_sse3 _op_blend_pt_pas_dpan_sse3
-static void
-init_blend_rel_pixel_pt_funcs_sse3(void)
-{
-   op_blend_rel_pt_funcs[SP][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_pt_p_dp_sse3;
-   op_blend_rel_pt_funcs[SP_AS][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_pt_pas_dp_sse3;
-   op_blend_rel_pt_funcs[SP_AN][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_pt_pan_dp_sse3;
-   op_blend_rel_pt_funcs[SP][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_pt_p_dpan_sse3;
-   op_blend_rel_pt_funcs[SP_AS][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_pt_pas_dpan_sse3;
-   op_blend_rel_pt_funcs[SP_AN][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_pt_pan_dpan_sse3;
-}
-#endif

diff --git a/libraries/evas/src/lib/engines/common/evas_op_blend/op_blend_pixel_sse3.c b/libraries/evas/src/lib/engines/common/evas_op_blend/op_blend_pixel_sse3.c deleted file mode 100644 index 69c597c..0000000 --- a/libraries/evas/src/lib/engines/common/evas_op_blend/op_blend_pixel_sse3.c +++ /dev/null
@@ -1,315 +0,0 @@
1	/* blend pixel --> dst */
2
3	#ifdef BUILD_SSE3
4
5	static void
6	_op_blend_p_dp_sse3(DATA32 s, DATA8 m __UNUSED__, DATA32 c __UNUSED__, DATA32 *d, int l) {
7
8	LOOP_ALIGNED_U1_A48_SSE3(d, l,
9	{ /* UOP */
10
11	int alpha = 256 - (*s >> 24);
12	d = s + MUL_256(alpha, *d);
13	s++; d++; l--;
14	},
15	{ /* A4OP */
16
17	__m128i s0 = _mm_lddqu_si128((__m128i *)s);
18	__m128i d0 = _mm_load_si128((__m128i *)d);
19
20	__m128i a0 = sub4_alpha_sse3(s0);
21	__m128i mul0 = mul_256_sse3(a0, d0);
22	d0 = _mm_add_epi32(mul0, s0);
23
24	_mm_store_si128((__m128i *)d, d0);
25
26	s += 4; d += 4; l -= 4;
27	},
28	{ /* A8OP */
29
30	__m128i s0 = _mm_lddqu_si128((__m128i *)s);
31	__m128i d0 = _mm_load_si128((__m128i *)d);
32
33	__m128i s1 = _mm_lddqu_si128((__m128i *)(s+4));
34	__m128i d1 = _mm_load_si128((__m128i *)(d+4));
35
36	__m128i a0 = sub4_alpha_sse3(s0);
37	__m128i a1 = sub4_alpha_sse3(s1);
38
39	__m128i mul0 = mul_256_sse3(a0, d0);
40	__m128i mul1 = mul_256_sse3(a1, d1);
41
42	d0 = _mm_add_epi32(mul0, s0);
43	d1 = _mm_add_epi32(mul1, s1);
44
45	_mm_store_si128((__m128i *)d, d0);
46	_mm_store_si128((__m128i *)(d+4), d1);
47
48	s += 8; d += 8; l -= 8;
49	})
50	}
51
52	static void
53	_op_blend_pas_dp_sse3(DATA32 s, DATA8 m __UNUSED__, DATA32 c __UNUSED__, DATA32 *d, int l) {
54
55	int alpha;
56
57	const __m128i zero = _mm_setzero_si128();
58
59	LOOP_ALIGNED_U1_A48_SSE3(d, l,
60	{ /* UOP */
61	switch (*s & 0xff000000)
62	{
63	case 0:
64	break;
65	case 0xff000000:
66	d = s;
67	break;
68	default:
69	alpha = 256 - (*s >> 24);
70	d = s + MUL_256(alpha, *d);
71	break;
72	}
73	s++; d++; l--;
74	},
75	{ /* A4OP */
76
77	__m128i s0 = _mm_lddqu_si128((__m128i *)s);
78	__m128i d0 = _mm_load_si128((__m128i *)d);
79
80	__m128i a0 = sub4_alpha_sse3(s0);
81	__m128i mul0 = mul_256_sse3(a0, d0);
82
83	mul0 = _mm_add_epi32(s0, mul0);
84
85	__m128i zmask0 = _mm_cmpeq_epi32(_mm_srli_epi32(s0, 24), zero);
86	__m128i imask0 = ~zmask0;
87
88	mul0 = _mm_and_si128(imask0, mul0);
89	d0 = _mm_and_si128(zmask0, d0);
90
91	d0 = _mm_add_epi32(mul0, d0);
92
93	_mm_store_si128((__m128i *)d, d0);
94
95	s += 4; d += 4; l -= 4;
96	},
97	{ /* A8OP */
98
99	__m128i s0 = _mm_lddqu_si128((__m128i *)s);
100	__m128i d0 = _mm_load_si128((__m128i *)d);
101
102	__m128i s1 = _mm_lddqu_si128((__m128i *)(s+4));
103	__m128i d1 = _mm_load_si128((__m128i *)(d+4));
104
105	__m128i a0 = sub4_alpha_sse3(s0);
106	__m128i a1 = sub4_alpha_sse3(s1);
107
108	__m128i mul0 = mul_256_sse3(a0, d0);
109	__m128i mul1 = mul_256_sse3(a1, d1);
110
111	mul0 = _mm_add_epi32(s0, mul0);
112	mul1 = _mm_add_epi32(s1, mul1);
113
114	__m128i zmask0 = _mm_cmpeq_epi32(_mm_srli_epi32(s0, 24), zero);
115	__m128i zmask1 = _mm_cmpeq_epi32(_mm_srli_epi32(s1, 24), zero);
116
117	__m128i imask0 = ~zmask0;
118	__m128i imask1 = ~zmask1;
119
120	mul0 = _mm_and_si128(imask0, mul0);
121	d0 = _mm_and_si128(zmask0, d0);
122
123	mul1 = _mm_and_si128(imask1, mul1);
124	d1 = _mm_and_si128(zmask1, d1);
125
126	d0 = _mm_add_epi32(mul0, d0);
127	d1 = _mm_add_epi32(mul1, d1);
128
129	_mm_store_si128((__m128i *)d, d0);
130	_mm_store_si128((__m128i *)(d+4), d1);
131
132	s += 8; d += 8; l -= 8;
133	})
134	}
135
136	#define _op_blend_pan_dp_sse3 NULL
137
138	#define _op_blend_p_dpan_sse3 _op_blend_p_dp_sse3
139	#define _op_blend_pas_dpan_sse3 _op_blend_pas_dp_sse3
140	#define _op_blend_pan_dpan_sse3 _op_blend_pan_dp_sse3
141
142	static void
143	init_blend_pixel_span_funcs_sse3(void)
144	{
145	op_blend_span_funcs[SP][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_p_dp_sse3;
146	op_blend_span_funcs[SP_AS][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_pas_dp_sse3;
147	op_blend_span_funcs[SP_AN][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_pan_dp_sse3;
148
149	// FIXME: BUGGY BUGGY Core i5 750 (32bit), 4.5.2 (Ubuntu/Linaro 4.5.2-8ubuntu4), ello (text and rectangle)
150	// op_blend_span_funcs[SP][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_p_dpan_sse3;
151	op_blend_span_funcs[SP_AS][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_pas_dpan_sse3;
152	op_blend_span_funcs[SP_AN][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_pan_dpan_sse3;
153	}
154
155	#define _op_blend_pt_p_dp_sse3 NULL
156
157	#define _op_blend_pt_pas_dp_sse3 _op_blend_pt_p_dp_sse3
158	#define _op_blend_pt_pan_dp_sse3 NULL
159
160	#define _op_blend_pt_p_dpan_sse3 _op_blend_pt_p_dp_sse3
161	#define _op_blend_pt_pan_dpan_sse3 _op_blend_pt_pan_dp_sse3
162	#define _op_blend_pt_pas_dpan_sse3 _op_blend_pt_pas_dp_sse3
163
164	static void
165	init_blend_pixel_pt_funcs_sse3(void)
166	{
167	op_blend_pt_funcs[SP][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_pt_p_dp_sse3;
168	op_blend_pt_funcs[SP_AS][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_pt_pas_dp_sse3;
169	op_blend_pt_funcs[SP_AN][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_pt_pan_dp_sse3;
170
171	op_blend_pt_funcs[SP][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_pt_p_dpan_sse3;
172	op_blend_pt_funcs[SP_AS][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_pt_pas_dpan_sse3;
173	op_blend_pt_funcs[SP_AN][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_pt_pan_dpan_sse3;
174	}
175
176	/-----/
177
178	/* blend_rel pixel -> dst */
179
180	static void
181	_op_blend_rel_p_dp_sse3(DATA32 s, DATA8 m __UNUSED__, DATA32 c, DATA32 *d, int l) {
182
183	const __m128i ones = _mm_set_epi32(1, 1, 1, 1);
184
185	LOOP_ALIGNED_U1_A48_SSE3(d, l,
186	{ /* UOP */
187
188	int alpha = 256 - (*s >> 24);
189	c = 1 + (*d >> 24);
190	d = MUL_256(c, s) + MUL_256(alpha, *d);
191	d++; s++; l--;
192	},
193	{ /A4OP /
194
195	__m128i s0 = _mm_lddqu_si128((__m128i *)s);
196	__m128i d0 = _mm_load_si128((__m128i *)d);
197
198	__m128i c0 = _mm_add_epi32(_mm_srli_epi32(d0, 24), ones);
199	__m128i a0 = sub4_alpha_sse3(s0);
200
201	d0 = _mm_add_epi32(mul_256_sse3(c0, s0), mul_256_sse3(a0, d0));
202
203	_mm_store_si128((__m128i *)d, d0);
204
205	d += 4; s += 4; l -= 4;
206	},
207	{ /* A8OP */
208
209	__m128i s0 = _mm_lddqu_si128((__m128i *)s);
210	__m128i d0 = _mm_load_si128 ((__m128i *)d);
211
212	__m128i s1 = _mm_lddqu_si128((__m128i *)(s+4));
213	__m128i d1 = _mm_load_si128 ((__m128i *)(d+4));
214
215	__m128i c0 = _mm_add_epi32(_mm_srli_epi32(d0, 24), ones);
216	__m128i c1 = _mm_add_epi32(_mm_srli_epi32(d1, 24), ones);
217
218	__m128i a0 = sub4_alpha_sse3(s0);
219	__m128i a1 = sub4_alpha_sse3(s1);
220
221	d0 = _mm_add_epi32(mul_256_sse3(c0, s0), mul_256_sse3(a0, d0));
222	d1 = _mm_add_epi32(mul_256_sse3(c1, s1), mul_256_sse3(a1, d1));
223
224	_mm_store_si128((__m128i *)d, d0);
225	_mm_store_si128((__m128i *)(d+4), d1);
226
227	d += 8; s += 8; l -= 8;
228	})
229	}
230
231	static void
232	_op_blend_rel_pan_dp_sse3(DATA32 s, DATA8 m __UNUSED__, DATA32 c, DATA32 *d, int l) {
233
234	const __m128i ones = _mm_set_epi32(1, 1, 1, 1);
235
236	LOOP_ALIGNED_U1_A48_SSE3(d, l,
237	{ /* UOP */
238
239	c = 1 + (*d >> 24);
240	d++ = MUL_256(c, s);
241	s++; l--;
242	},
243	{ /* A4OP */
244
245	__m128i s0 = _mm_lddqu_si128((__m128i *)s);
246	__m128i d0 = _mm_load_si128((__m128i *)d);
247
248	__m128i c0 = _mm_add_epi32(_mm_srli_epi32(d0, 24), ones);
249	d0 = mul_256_sse3(c0, s0);
250
251	_mm_store_si128((__m128i *)d, d0);
252
253	d += 4; s += 4; l -= 4;
254	},
255	{ /* A8OP */
256
257	__m128i s0 = _mm_lddqu_si128((__m128i *)s);
258	__m128i d0 = _mm_load_si128 ((__m128i *)d);
259
260	__m128i s1 = _mm_lddqu_si128((__m128i *)(s+4));
261	__m128i d1 = _mm_load_si128 ((__m128i *)(d+4));
262
263	__m128i c0 = _mm_add_epi32(_mm_srli_epi32(d0, 24), ones);
264	__m128i c1 = _mm_add_epi32(_mm_srli_epi32(d1, 24), ones);
265
266	d0 = mul_256_sse3(c0, s0);
267	d1 = mul_256_sse3(c1, s1);
268
269	_mm_store_si128((__m128i *)d, d0);
270	_mm_store_si128((__m128i *)(d+4), d1);
271
272	d += 8; s += 8; l -= 8;
273	})
274	}
275
276	#define _op_blend_rel_pas_dp_sse3 _op_blend_rel_p_dp_sse3
277
278	#define _op_blend_rel_p_dpan_sse3 _op_blend_p_dpan_sse3
279	#define _op_blend_rel_pan_dpan_sse3 _op_blend_pan_dpan_sse3
280	#define _op_blend_rel_pas_dpan_sse3 _op_blend_pas_dpan_sse3
281
282	static void
283	init_blend_rel_pixel_span_funcs_sse3(void)
284	{
285	op_blend_rel_span_funcs[SP][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_p_dp_sse3;
286	op_blend_rel_span_funcs[SP_AS][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_pas_dp_sse3;
287	op_blend_rel_span_funcs[SP_AN][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_pan_dp_sse3;
288
289	op_blend_rel_span_funcs[SP][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_p_dpan_sse3;
290	op_blend_rel_span_funcs[SP_AS][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_pas_dpan_sse3;
291	op_blend_rel_span_funcs[SP_AN][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_pan_dpan_sse3;
292	}
293
294	#define _op_blend_rel_pt_p_dp_sse3 NULL
295	#define _op_blend_rel_pt_pan_dp_sse3 NULL
296
297	#define _op_blend_rel_pt_pas_dp_sse3 _op_blend_rel_pt_p_dp_sse3
298
299	#define _op_blend_rel_pt_p_dpan_sse3 _op_blend_pt_p_dpan_sse3
300	#define _op_blend_rel_pt_pan_dpan_sse3 _op_blend_pt_pan_dpan_sse3
301	#define _op_blend_rel_pt_pas_dpan_sse3 _op_blend_pt_pas_dpan_sse3
302
303	static void
304	init_blend_rel_pixel_pt_funcs_sse3(void)
305	{
306	op_blend_rel_pt_funcs[SP][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_pt_p_dp_sse3;
307	op_blend_rel_pt_funcs[SP_AS][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_pt_pas_dp_sse3;
308	op_blend_rel_pt_funcs[SP_AN][SM_N][SC_N][DP][CPU_SSE3] = _op_blend_rel_pt_pan_dp_sse3;
309
310	op_blend_rel_pt_funcs[SP][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_pt_p_dpan_sse3;
311	op_blend_rel_pt_funcs[SP_AS][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_pt_pas_dpan_sse3;
312	op_blend_rel_pt_funcs[SP_AN][SM_N][SC_N][DP_AN][CPU_SSE3] = _op_blend_rel_pt_pan_dpan_sse3;
313	}
314
315	#endif