vmx: implement fast path vmx_composite_over_n_8_8888

POWER8, 8 cores, 3.4GHz, RHEL 7.2 ppc64le.

reference memcpy speed = 25008.9MB/s (6252.2MP/s for 32bpp fills)

                Before         After           Change
              ---------------------------------------------
L1              91.32          182.84         +100.22%
L2              94.94          182.83         +92.57%
M               95.55          181.51         +89.96%
HT              88.96          162.09         +82.21%
VT              87.4           168.35         +92.62%
R               83.37          146.23         +75.40%
RT              66.4           91.5           +37.80%
Kops/s          683            859            +25.77%

Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
Acked-by: Pekka Paalanen <pekka.paalanen@collabora.co.uk>
Acked-by: Siarhei Siamashka <siarhei.siamashka@gmail.com>

diff --git a/pixman/pixman-vmx.c b/pixman/pixman-vmx.c
index a5c5db9879e5585f069c7797605b767961a7b699..7ef8bedc7192d1a38db6d89affbf6b492e9be235 100644 (file)
--- a/pixman/pixman-vmx.c
+++ b/pixman/pixman-vmx.c
@@ -59,6 +59,15 @@ splat_alpha (vector unsigned int pix)
 #endif
 }
 
+static force_inline vector unsigned int
+splat_pixel (vector unsigned int pix)
+{
+    return vec_perm (pix, pix,
+                    (vector unsigned char)AVV (
+                        0x00, 0x00, 0x00, 0x00, 0x01, 0x01, 0x01, 0x01,
+                        0x02, 0x02, 0x02, 0x02, 0x03, 0x03, 0x03, 0x03));
+}
+
 static force_inline vector unsigned int
 pix_multiply (vector unsigned int p, vector unsigned int a)
 {
@@ -2508,6 +2517,104 @@ vmx_combine_add_ca (pixman_implementation_t *imp,
     }
 }
 
+static void
+vmx_composite_over_n_8_8888 (pixman_implementation_t *imp,
+                              pixman_composite_info_t *info)
+{
+    PIXMAN_COMPOSITE_ARGS (info);
+    uint32_t src, srca;
+    uint32_t *dst_line, *dst;
+    uint8_t *mask_line;
+    int dst_stride, mask_stride;
+    int32_t w;
+    uint32_t m, d, s, ia;
+
+    vector unsigned int vsrc, valpha, vmask, vdst;
+
+    src = _pixman_image_get_solid (imp, src_image, dest_image->bits.format);
+
+    srca = ALPHA_8(src);
+    if (src == 0)
+       return;
+
+    PIXMAN_IMAGE_GET_LINE (
+       dest_image, dest_x, dest_y, uint32_t, dst_stride, dst_line, 1);
+    PIXMAN_IMAGE_GET_LINE (
+       mask_image, mask_x, mask_y, uint8_t, mask_stride, mask_line, 1);
+
+    vsrc = (vector unsigned int) {src, src, src, src};
+    valpha = splat_alpha(vsrc);
+
+    while (height--)
+    {
+       const uint8_t *pm = mask_line;
+       dst = dst_line;
+       dst_line += dst_stride;
+       mask_line += mask_stride;
+       w = width;
+
+       while (w && (uintptr_t)dst & 15)
+       {
+           s = src;
+           m = *pm++;
+
+           if (m)
+           {
+               d = *dst;
+               UN8x4_MUL_UN8 (s, m);
+               ia = ALPHA_8 (~s);
+               UN8x4_MUL_UN8_ADD_UN8x4 (d, ia, s);
+               *dst = d;
+           }
+
+           w--;
+           dst++;
+       }
+
+       while (w >= 4)
+       {
+           m = *((uint32_t*)pm);
+
+           if (srca == 0xff && m == 0xffffffff)
+           {
+               save_128_aligned(dst, vsrc);
+           }
+           else if (m)
+           {
+               vmask = splat_pixel((vector unsigned int) {m, m, m, m});
+
+               /* dst is 16-byte aligned */
+               vdst = in_over (vsrc, valpha, vmask, load_128_aligned (dst));
+
+               save_128_aligned(dst, vdst);
+           }
+
+           w -= 4;
+           dst += 4;
+           pm += 4;
+       }
+
+       while (w)
+       {
+           s = src;
+           m = *pm++;
+
+           if (m)
+           {
+               d = *dst;
+               UN8x4_MUL_UN8 (s, m);
+               ia = ALPHA_8 (~s);
+               UN8x4_MUL_UN8_ADD_UN8x4 (d, ia, s);
+               *dst = d;
+           }
+
+           w--;
+           dst++;
+       }
+    }
+
+}
+
 static pixman_bool_t
 vmx_fill (pixman_implementation_t *imp,
            uint32_t *               bits,
@@ -3028,6 +3135,10 @@ static const pixman_fast_path_t vmx_fast_paths[] =
     PIXMAN_STD_FAST_PATH (OVER, a8r8g8b8, null, x8r8g8b8, vmx_composite_over_8888_8888),
     PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, a8b8g8r8, vmx_composite_over_8888_8888),
     PIXMAN_STD_FAST_PATH (OVER, a8b8g8r8, null, x8b8g8r8, vmx_composite_over_8888_8888),
+    PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8r8g8b8, vmx_composite_over_n_8_8888),
+    PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8r8g8b8, vmx_composite_over_n_8_8888),
+    PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8b8g8r8, vmx_composite_over_n_8_8888),
+    PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8b8g8r8, vmx_composite_over_n_8_8888),
     PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, a8r8g8b8, vmx_composite_over_n_8888_8888_ca),
     PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8r8g8b8, x8r8g8b8, vmx_composite_over_n_8888_8888_ca),
     PIXMAN_STD_FAST_PATH_CA (OVER, solid, a8b8g8r8, a8b8g8r8, vmx_composite_over_n_8888_8888_ca),