MIPS: DSPr2: Added fast-paths for ADD operation: - add_n_8_8 - add_n_8_8888 - add_8_8_8

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

lowlevel-blt-bench results

Referent (before):
        add_n_8_8    =  L1:  41.37  L2:  37.83  M: 30.38 ( 60.45%)  HT: 23.70  VT: 22.85  R: 21.51  RT: 10.32 (  45Kops/s)
        add_n_8_8888 =  L1:  16.01  L2:  14.46  M: 11.64 ( 46.32%)  HT:  5.50  VT:  5.18  R:  5.06  RT:  1.89 (  18Kops/s)
        add_8_8_8    =  L1:  13.26  L2:  12.47  M: 11.16 ( 29.61%)  HT:  8.09  VT:  8.04  R:  7.68  RT:  3.90 (  29Kops/s)

Optimized:
        add_n_8_8    =  L1:  96.03  L2:  79.37  M: 51.89 (103.31%)  HT: 32.59  VT: 31.29  R: 28.52  RT: 11.08 (  46Kops/s)
        add_n_8_8888 =  L1:  53.61  L2:  46.92  M: 23.78 ( 94.70%)  HT: 19.06  VT: 18.64  R: 17.30  RT:  9.15 (  43Kops/s)
        add_8_8_8    =  L1:  89.65  L2:  66.82  M: 37.10 ( 98.48%)  HT: 22.10  VT: 21.74  R: 20.12  RT:  8.12 (  41Kops/s)

3 files changed, 284 insertions, 0 deletions

diff --git a/pixman/pixman-mips-dspr2-asm.S b/pixman/pixman-mips-dspr2-asm.S
index 3a6b26a..614c628 100644
--- a/pixman/pixman-mips-dspr2-asm.S
+++ b/pixman/pixman-mips-dspr2-asm.S
@@ -1209,6 +1209,258 @@ LEAF_MIPS_DSPR2(pixman_composite_over_8888_8888_asm_mips)
 
 END(pixman_composite_over_8888_8888_asm_mips)
 
+LEAF_MIPS_DSPR2(pixman_composite_add_8_8_8_asm_mips)
+/*
+ * a0 - dst  (a8)
+ * a1 - src  (a8)
+ * a2 - mask (a8)
+ * a3 - w
+ */
+
+    SAVE_REGS_ON_STACK 0, v0, v1
+    li                t9, 0x00ff00ff
+    beqz              a3, 3f
+     nop
+
+    srl               v0, a3, 2   /* v0 = how many multiples of 4 dst pixels */
+    beqz              v0, 1f      /* branch if less than 4 src pixels */
+     nop
+
+0:
+    beqz              v0, 1f
+     addiu            v0, v0, -1
+    lbu               t0, 0(a2)
+    lbu               t1, 1(a2)
+    lbu               t2, 2(a2)
+    lbu               t3, 3(a2)
+    lbu               t4, 0(a0)
+    lbu               t5, 1(a0)
+    lbu               t6, 2(a0)
+    lbu               t7, 3(a0)
+
+    addiu             a2, a2, 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
+
+    lbu               t4, 0(a1)
+    lbu               v1, 1(a1)
+    lbu               t7, 2(a1)
+    lbu               t8, 3(a1)
+
+    addiu             a1, a1, 4
+
+    precr_sra.ph.w    v1, t4, 0
+    precr_sra.ph.w    t8, t7, 0
+
+    muleu_s.ph.qbl    t2, t0, t8
+    muleu_s.ph.qbr    t3, t0, v1
+    shra_r.ph         t4, t2, 8
+    shra_r.ph         t5, t3, 8
+    and               t4, t4, t9
+    and               t5, t5, t9
+    addq.ph           t2, t2, t4
+    addq.ph           t3, t3, t5
+    shra_r.ph         t2, t2, 8
+    shra_r.ph         t3, t3, 8
+    precr.qb.ph       t0, t2, t3
+
+    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             a3, a3, -4
+    b                 0b
+     addiu            a0, a0, 4
+
+1:
+    beqz              a3, 3f
+     nop
+2:
+    lbu               t8, 0(a1)
+    lbu               t0, 0(a2)
+    lbu               t1, 0(a0)
+    addiu             a1, a1, 1
+    addiu             a2, a2, 1
+
+    mul               t2, t0, t8
+    shra_r.ph         t3, t2, 8
+    andi              t3, t3, 0xff
+    addq.ph           t2, t2, t3
+    shra_r.ph         t2, t2, 8
+    andi              t2, t2, 0xff
+
+    addu_s.qb         t2, t2, t1
+    sb                t2, 0(a0)
+    addiu             a3, a3, -1
+    bnez              a3, 2b
+     addiu            a0, a0, 1
+
+3:
+    RESTORE_REGS_FROM_STACK 0, v0, v1
+    j                 ra
+     nop
+
+END(pixman_composite_add_8_8_8_asm_mips)
+
+LEAF_MIPS_DSPR2(pixman_composite_add_n_8_8_asm_mips)
+/*
+ * a0 - dst  (a8)
+ * a1 - src  (32bit constant)
+ * a2 - mask (a8)
+ * a3 - w
+ */
+
+    SAVE_REGS_ON_STACK 0, v0
+    li                t9, 0x00ff00ff
+    beqz              a3, 3f
+     nop
+
+    srl               v0, a3, 2   /* v0 = how many multiples of 4 dst pixels */
+    beqz              v0, 1f      /* branch if less than 4 src pixels */
+     nop
+
+    srl               t8, a1, 24
+    replv.ph          t8, t8
+
+0:
+    beqz              v0, 1f
+     addiu            v0, v0, -1
+    lbu               t0, 0(a2)
+    lbu               t1, 1(a2)
+    lbu               t2, 2(a2)
+    lbu               t3, 3(a2)
+    lbu               t4, 0(a0)
+    lbu               t5, 1(a0)
+    lbu               t6, 2(a0)
+    lbu               t7, 3(a0)
+
+    addiu             a2, a2, 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
+
+    muleu_s.ph.qbl    t2, t0, t8
+    muleu_s.ph.qbr    t3, t0, t8
+    shra_r.ph         t4, t2, 8
+    shra_r.ph         t5, t3, 8
+    and               t4, t4, t9
+    and               t5, t5, t9
+    addq.ph           t2, t2, t4
+    addq.ph           t3, t3, t5
+    shra_r.ph         t2, t2, 8
+    shra_r.ph         t3, t3, 8
+    precr.qb.ph       t0, t2, t3
+
+    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             a3, a3, -4
+    b                 0b
+     addiu            a0, a0, 4
+
+1:
+    beqz              a3, 3f
+     nop
+    srl               t8, a1, 24
+2:
+    lbu               t0, 0(a2)
+    lbu               t1, 0(a0)
+    addiu             a2, a2, 1
+
+    mul               t2, t0, t8
+    shra_r.ph         t3, t2, 8
+    andi              t3, t3, 0xff
+    addq.ph           t2, t2, t3
+    shra_r.ph         t2, t2, 8
+    andi              t2, t2, 0xff
+
+    addu_s.qb         t2, t2, t1
+    sb                t2, 0(a0)
+    addiu             a3, a3, -1
+    bnez              a3, 2b
+     addiu            a0, a0, 1
+
+3:
+    RESTORE_REGS_FROM_STACK 0, v0
+    j                 ra
+     nop
+
+END(pixman_composite_add_n_8_8_asm_mips)
+
+LEAF_MIPS_DSPR2(pixman_composite_add_n_8_8888_asm_mips)
+/*
+ * a0 - dst  (a8r8g8b8)
+ * a1 - src  (32bit constant)
+ * a2 - mask (a8)
+ * 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:
+                       /* a1 = source      (32bit constant) */
+    lbu      t0, 0(a2) /* t0 = mask        (a8) */
+    lbu      t1, 1(a2) /* t1 = mask        (a8) */
+    lw       t2, 0(a0) /* t2 = destination (a8r8g8b8) */
+    lw       t3, 4(a0) /* t3 = destination (a8r8g8b8) */
+    addiu    a2, a2, 2
+
+    MIPS_2xUN8x4_MUL_2xUN8_ADD_2xUN8x4 a1, a1, \
+                                       t0, t1, \
+                                       t2, t3, \
+                                       t5, t6, \
+                                       t4, t7, t8, t9, s0, s1, s2
+
+    sw       t5, 0(a0)
+    sw       t6, 4(a0)
+    addiu    a3, a3, -2
+    addiu    t1, a3, -1
+    bgtz     t1, 1b
+     addiu   a0, a0, 8
+2:
+    beqz     a3, 3f
+     nop
+                       /* a1 = source      (32bit constant) */
+    lbu      t0, 0(a2) /* t0 = mask        (a8) */
+    lw       t1, 0(a0) /* t1 = destination (a8r8g8b8) */
+
+    MIPS_UN8x4_MUL_UN8_ADD_UN8x4 a1, t0, t1, t2, t4, t3, t5, t6
+
+    sw       t2, 0(a0)
+3:
+    RESTORE_REGS_FROM_STACK 0, s0, s1, s2
+    j        ra
+     nop
+
+END(pixman_composite_add_n_8_8888_asm_mips)
+
 LEAF_MIPS_DSPR2(pixman_scaled_bilinear_scanline_8888_8888_SRC_asm_mips)
 /*
  * a0     - *dst
diff --git a/pixman/pixman-mips-dspr2-asm.h b/pixman/pixman-mips-dspr2-asm.h
index 7327dc6..b330c0f 100644
--- a/pixman/pixman-mips-dspr2-asm.h
+++ b/pixman/pixman-mips-dspr2-asm.h
@@ -600,6 +600,28 @@ LEAF_MIPS32R2(symbol)                                   \
     addu_s.qb          \out_8888, \out_8888, \d_8888
 .endm
 
+.macro MIPS_2xUN8x4_MUL_2xUN8_ADD_2xUN8x4 s1_8888,   \
+                             s2_8888,   \
+                             m1_8,      \
+                             m2_8,      \
+                             d1_8888,   \
+                             d2_8888,   \
+                             out1_8888, \
+                             out2_8888, \
+                             maskLSR,   \
+                             scratch1,  scratch2, scratch3, \
+                             scratch4, scratch5, scratch6
+    MIPS_2xUN8x4_MUL_2xUN8 \s1_8888,   \s2_8888, \
+                           \m1_8,      \m2_8, \
+                           \out1_8888, \out2_8888, \
+                           \maskLSR, \
+                           \scratch1,  \scratch2, \scratch3, \
+                           \scratch4,  \scratch5, \scratch6
+
+    addu_s.qb             \out1_8888, \out1_8888, \d1_8888
+    addu_s.qb             \out2_8888, \out2_8888, \d2_8888
+.endm
+
 .macro BILINEAR_INTERPOLATE_SINGLE_PIXEL tl, tr, bl, br,         \
                                          scratch1, scratch2,     \
                                          alpha, red, green, blue \
diff --git a/pixman/pixman-mips-dspr2.c b/pixman/pixman-mips-dspr2.c
index e80bbb6..30d2a85 100644
--- a/pixman/pixman-mips-dspr2.c
+++ b/pixman/pixman-mips-dspr2.c
@@ -59,6 +59,10 @@ PIXMAN_MIPS_BIND_FAST_PATH_N_MASK_DST (SKIP_ZERO_SRC, over_n_8_8888,
                                        uint8_t, 1, uint32_t, 1)
 PIXMAN_MIPS_BIND_FAST_PATH_N_MASK_DST (SKIP_ZERO_SRC, over_n_8_0565,
                                        uint8_t, 1, uint16_t, 1)
+PIXMAN_MIPS_BIND_FAST_PATH_N_MASK_DST (SKIP_ZERO_SRC, add_n_8_8,
+                                       uint8_t, 1, uint8_t, 1)
+PIXMAN_MIPS_BIND_FAST_PATH_N_MASK_DST (SKIP_ZERO_SRC, add_n_8_8888,
+                                       uint8_t, 1, uint32_t, 1)
 
 PIXMAN_MIPS_BIND_FAST_PATH_SRC_N_DST (SKIP_ZERO_MASK, over_8888_n_8888,
                                       uint32_t, 1, uint32_t, 1)
@@ -67,6 +71,8 @@ PIXMAN_MIPS_BIND_FAST_PATH_SRC_N_DST (SKIP_ZERO_MASK, over_8888_n_0565,
 PIXMAN_MIPS_BIND_FAST_PATH_SRC_N_DST (SKIP_ZERO_MASK, over_0565_n_0565,
                                       uint16_t, 1, uint16_t, 1)
 
+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 (over_8888_8_8888, uint32_t, 1,
                                          uint8_t, 1, uint32_t, 1)
 PIXMAN_MIPS_BIND_FAST_PATH_SRC_MASK_DST (over_8888_8_0565, uint32_t, 1,
@@ -271,6 +277,10 @@ static const pixman_fast_path_t mips_dspr2_fast_paths[] =
     PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null,     x8r8g8b8, mips_composite_over_8888_8888),
     PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null,     a8b8g8r8, mips_composite_over_8888_8888),
     PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null,     x8b8g8r8, mips_composite_over_8888_8888),
+    PIXMAN_STD_FAST_PATH (ADD,  solid,    a8,       a8,       mips_composite_add_n_8_8),
+    PIXMAN_STD_FAST_PATH (ADD,  solid,    a8,       a8r8g8b8, mips_composite_add_n_8_8888),
+    PIXMAN_STD_FAST_PATH (ADD,  solid,    a8,       a8b8g8r8, mips_composite_add_n_8_8888),
+    PIXMAN_STD_FAST_PATH (ADD,  a8,       a8,       a8,       mips_composite_add_8_8_8),
 
     SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, a8r8g8b8, mips_8888_8888),
     SIMPLE_BILINEAR_FAST_PATH (SRC, a8r8g8b8, x8r8g8b8, mips_8888_8888),