MIPS: DSPr2: Added more fast-paths for ADD operation: - add_8888_8888_8888 - add_8_8 - add_8888_8888

Performance numbers before/after on MIPS-74kc @ 1GHz:

lowlevel-blt-bench results

Referent (before):
        add_8888_8888_8888 =  L1:  17.55  L2:  13.35  M:  8.13 ( 93.95%)  HT:  6.60  VT:  6.64  R:  6.45  RT:  3.47 (  26Kops/s)
        add_8_8            =  L1:  86.07  L2:  84.89  M: 62.36 ( 90.11%)  HT: 36.36  VT: 34.74  R: 29.56  RT: 11.56 (  52Kops/s)
        add_8888_8888      =  L1:  95.59  L2:  73.05  M: 17.62 (101.84%)  HT: 15.46  VT: 15.01  R: 13.94  RT:  6.71 (  42Kops/s)

Optimized:
        add_8888_8888_8888 =  L1:  41.52  L2:  33.21  M: 11.97 (138.45%)  HT: 10.47  VT: 10.19  R:  9.42  RT:  4.86 (  32Kops/s)
        add_8_8            =  L1: 135.06  L2: 104.82  M: 57.13 ( 82.58%)  HT: 34.79  VT: 36.60  R: 28.28  RT: 10.54 (  51Kops/s)
        add_8888_8888      =  L1: 176.36  L2:  67.82  M: 17.48 (101.06%)  HT: 15.16  VT: 14.62  R: 13.88  RT:  8.05 (  45Kops/s)

2 files changed, 212 insertions, 0 deletions

diff --git a/pixman/pixman-mips-dspr2-asm.S b/pixman/pixman-mips-dspr2-asm.S
index 7c8ca30..b5cae16 100644
--- a/pixman/pixman-mips-dspr2-asm.S
+++ b/pixman/pixman-mips-dspr2-asm.S
@@ -1631,6 +1631,208 @@ LEAF_MIPS_DSPR2(pixman_composite_add_8888_n_8888_asm_mips)
 
 END(pixman_composite_add_8888_n_8888_asm_mips)
 
+LEAF_MIPS_DSPR2(pixman_composite_add_8888_8888_8888_asm_mips)
+/*
+ * a0 - dst  (a8r8g8b8)
+ * a1 - src  (a8r8g8b8)
+ * a2 - mask (a8r8g8b8)
+ * a3 - w
+ */
+
+    SAVE_REGS_ON_STACK 0, s0, s1, s2
+    li       t4, 0x00ff00ff
+    beqz     a3, 3f
+     nop
+    addiu    t1, a3, -1
+    beqz     t1, 2f
+     nop
+1:
+    lw       t0, 0(a1) /* t0 = source      (a8r8g8b8) */
+    lw       t1, 4(a1) /* t1 = source      (a8r8g8b8) */
+    lw       t2, 0(a2) /* t2 = mask        (a8r8g8b8) */
+    lw       t3, 4(a2) /* t3 = mask        (a8r8g8b8) */
+    lw       t5, 0(a0) /* t5 = destination (a8r8g8b8) */
+    lw       t6, 4(a0) /* t6 = destination (a8r8g8b8) */
+    addiu    a1, a1, 8
+    addiu    a2, a2, 8
+    srl      t2, t2, 24
+    srl      t3, t3, 24
+
+    MIPS_2xUN8x4_MUL_2xUN8_ADD_2xUN8x4 t0, t1, \
+                                       t2, t3, \
+                                       t5, t6, \
+                                       t7, t8, \
+                                       t4, t9, s0, s1, s2, t0, t1
+
+    sw       t7, 0(a0)
+    sw       t8, 4(a0)
+    addiu    a3, a3, -2
+    addiu    t1, a3, -1
+    bgtz     t1, 1b
+     addiu   a0, a0, 8
+2:
+    beqz     a3, 3f
+     nop
+    lw       t0, 0(a1) /* t0 = source      (a8r8g8b8) */
+    lw       t1, 0(a2) /* t1 = mask        (a8r8g8b8) */
+    lw       t2, 0(a0) /* t2 = destination (a8r8g8b8) */
+    srl      t1, t1, 24
+
+    MIPS_UN8x4_MUL_UN8_ADD_UN8x4 t0, t1, t2, t3, t4, t5, t6, t7
+
+    sw       t3, 0(a0)
+3:
+    RESTORE_REGS_FROM_STACK 0, s0, s1, s2
+    j        ra
+     nop
+
+END(pixman_composite_add_8888_8888_8888_asm_mips)
+
+LEAF_MIPS_DSPR2(pixman_composite_add_8_8_asm_mips)
+/*
+ * a0 - dst  (a8)
+ * a1 - src  (a8)
+ * a2 - w
+ */
+
+    beqz              a2, 3f
+     nop
+    srl               t9, a2, 2   /* t9 = how many multiples of 4 dst pixels */
+    beqz              t9, 1f      /* branch if less than 4 src pixels */
+     nop
+
+0:
+    beqz              t9, 1f
+     addiu            t9, t9, -1
+    lbu               t0, 0(a1)
+    lbu               t1, 1(a1)
+    lbu               t2, 2(a1)
+    lbu               t3, 3(a1)
+    lbu               t4, 0(a0)
+    lbu               t5, 1(a0)
+    lbu               t6, 2(a0)
+    lbu               t7, 3(a0)
+
+    addiu             a1, a1, 4
+
+    precr_sra.ph.w    t1, t0, 0
+    precr_sra.ph.w    t3, t2, 0
+    precr_sra.ph.w    t5, t4, 0
+    precr_sra.ph.w    t7, t6, 0
+
+    precr.qb.ph       t0, t3, t1
+    precr.qb.ph       t1, t7, t5
+
+    addu_s.qb         t2, t0, t1
+
+    sb                t2, 0(a0)
+    srl               t2, t2, 8
+    sb                t2, 1(a0)
+    srl               t2, t2, 8
+    sb                t2, 2(a0)
+    srl               t2, t2, 8
+    sb                t2, 3(a0)
+    addiu             a2, a2, -4
+    b                 0b
+     addiu            a0, a0, 4
+
+1:
+    beqz              a2, 3f
+     nop
+2:
+    lbu               t0, 0(a1)
+    lbu               t1, 0(a0)
+    addiu             a1, a1, 1
+
+    addu_s.qb         t2, t0, t1
+    sb                t2, 0(a0)
+    addiu             a2, a2, -1
+    bnez              a2, 2b
+     addiu            a0, a0, 1
+
+3:
+    j                 ra
+     nop
+
+END(pixman_composite_add_8_8_asm_mips)
+
+LEAF_MIPS_DSPR2(pixman_composite_add_8888_8888_asm_mips)
+/*
+ * a0 - dst (a8r8g8b8)
+ * a1 - src (a8r8g8b8)
+ * a2 - w
+ */
+
+    beqz         a2, 4f
+     nop
+
+    srl          t9, a2, 2      /* t1 = how many multiples of 4 src pixels */
+    beqz         t9, 3f         /* branch if less than 4 src pixels */
+     nop
+1:
+    addiu        t9, t9, -1
+    beqz         t9, 2f
+     addiu       a2, a2, -4
+
+    lw           t0, 0(a1)
+    lw           t1, 4(a1)
+    lw           t2, 8(a1)
+    lw           t3, 12(a1)
+    lw           t4, 0(a0)
+    lw           t5, 4(a0)
+    lw           t6, 8(a0)
+    lw           t7, 12(a0)
+    addiu        a1, a1, 16
+
+    addu_s.qb    t4, t4, t0
+    addu_s.qb    t5, t5, t1
+    addu_s.qb    t6, t6, t2
+    addu_s.qb    t7, t7, t3
+
+    sw           t4, 0(a0)
+    sw           t5, 4(a0)
+    sw           t6, 8(a0)
+    sw           t7, 12(a0)
+    b            1b
+     addiu       a0, a0, 16
+2:
+    lw           t0, 0(a1)
+    lw           t1, 4(a1)
+    lw           t2, 8(a1)
+    lw           t3, 12(a1)
+    lw           t4, 0(a0)
+    lw           t5, 4(a0)
+    lw           t6, 8(a0)
+    lw           t7, 12(a0)
+    addiu        a1, a1, 16
+
+    addu_s.qb    t4, t4, t0
+    addu_s.qb    t5, t5, t1
+    addu_s.qb    t6, t6, t2
+    addu_s.qb    t7, t7, t3
+
+    sw           t4, 0(a0)
+    sw           t5, 4(a0)
+    sw           t6, 8(a0)
+    sw           t7, 12(a0)
+
+    beqz         a2, 4f
+     addiu       a0, a0, 16
+3:
+    lw           t0, 0(a1)
+    lw           t1, 0(a0)
+    addiu        a1, a1, 4
+    addiu        a2, a2, -1
+    addu_s.qb    t1, t1, t0
+    sw           t1, 0(a0)
+    bnez         a2, 3b
+     addiu       a0, a0, 4
+4:
+    jr           ra
+     nop
+
+END(pixman_composite_add_8888_8888_asm_mips)
+
 LEAF_MIPS_DSPR2(pixman_scaled_bilinear_scanline_8888_8888_SRC_asm_mips)
 /*
  * a0     - *dst
diff --git a/pixman/pixman-mips-dspr2.c b/pixman/pixman-mips-dspr2.c
index 1471750..9da636d 100644
--- a/pixman/pixman-mips-dspr2.c
+++ b/pixman/pixman-mips-dspr2.c
@@ -50,6 +50,10 @@ PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST (DO_FAST_MEMCPY, src_0888_0888,
                                     uint8_t, 3, uint8_t, 3)
 PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST (0, over_8888_8888,
                                     uint32_t, 1, uint32_t, 1)
+PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST (0, add_8_8,
+                                    uint8_t, 1, uint8_t, 1)
+PIXMAN_MIPS_BIND_FAST_PATH_SRC_DST (0, add_8888_8888,
+                                    uint32_t, 1, uint32_t, 1)
 
 PIXMAN_MIPS_BIND_FAST_PATH_N_MASK_DST (SKIP_ZERO_SRC, over_n_8888_8888_ca,
                                        uint32_t, 1, uint32_t, 1)
@@ -77,6 +81,8 @@ PIXMAN_MIPS_BIND_FAST_PATH_SRC_MASK_DST (add_8_8_8, uint8_t,  1,
                                          uint8_t,  1, uint8_t,  1)
 PIXMAN_MIPS_BIND_FAST_PATH_SRC_MASK_DST (add_8888_8_8888, uint32_t, 1,
                                          uint8_t, 1, uint32_t, 1)
+PIXMAN_MIPS_BIND_FAST_PATH_SRC_MASK_DST (add_8888_8888_8888, uint32_t, 1,
+                                         uint32_t, 1, uint32_t, 1)
 PIXMAN_MIPS_BIND_FAST_PATH_SRC_MASK_DST (add_0565_8_0565, uint16_t, 1,
                                          uint8_t,  1, uint16_t, 1)
 PIXMAN_MIPS_BIND_FAST_PATH_SRC_MASK_DST (over_8888_8_8888, uint32_t, 1,
@@ -291,8 +297,12 @@ static const pixman_fast_path_t mips_dspr2_fast_paths[] =
     PIXMAN_STD_FAST_PATH (ADD,  b5g6r5,   a8,       b5g6r5,   mips_composite_add_0565_8_0565),
     PIXMAN_STD_FAST_PATH (ADD,  a8r8g8b8, a8,       a8r8g8b8, mips_composite_add_8888_8_8888),
     PIXMAN_STD_FAST_PATH (ADD,  a8b8g8r8, a8,       a8b8g8r8, mips_composite_add_8888_8_8888),
+    PIXMAN_STD_FAST_PATH (ADD,  a8r8g8b8, a8r8g8b8, a8r8g8b8, mips_composite_add_8888_8888_8888),
     PIXMAN_STD_FAST_PATH (ADD,  a8r8g8b8, solid,    a8r8g8b8, mips_composite_add_8888_n_8888),
     PIXMAN_STD_FAST_PATH (ADD,  a8b8g8r8, solid,    a8b8g8r8, mips_composite_add_8888_n_8888),
+    PIXMAN_STD_FAST_PATH (ADD,  a8,       null,     a8,       mips_composite_add_8_8),
+    PIXMAN_STD_FAST_PATH (ADD,  a8r8g8b8, null,     a8r8g8b8, mips_composite_add_8888_8888),
+    PIXMAN_STD_FAST_PATH (ADD,  a8b8g8r8, null,     a8b8g8r8, mips_composite_add_8888_8888),
 
     SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, mips_8888_8888),
     SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, mips_8888_8888),