From 2e855a2b4a2bb7b3d2ed1826cb4426d14080ca67 Mon Sep 17 00:00:00 2001 From: Siarhei Siamashka Date: Tue, 2 Nov 2010 15:25:51 +0200 Subject: [PATCH] ARM: NEON optimization for scaled src_8888_0565 with nearest filter Benchmark from ARM Cortex-A8 @720MHz: == before == op=1, src_fmt=20028888, dst_fmt=10020565, speed=42.51 MPix/s == after == op=1, src_fmt=20028888, dst_fmt=10020565, speed=55.61 MPix/s == unscaled == op=1, src_fmt=20028888, dst_fmt=10020565, speed=117.99 MPix/s --- pixman/pixman-arm-neon-asm.S | 10 ++++++++++ pixman/pixman-arm-neon.c | 7 +++++++ 2 files changed, 17 insertions(+) diff --git a/pixman/pixman-arm-neon-asm.S b/pixman/pixman-arm-neon-asm.S index ea3337a..4d6b03b 100644 --- a/pixman/pixman-arm-neon-asm.S +++ b/pixman/pixman-arm-neon-asm.S @@ -1999,3 +1999,13 @@ generate_composite_function_nearest_scanline \ 4, /* dst_r_basereg */ \ 0, /* src_basereg */ \ 24 /* mask_basereg */ + +generate_composite_function_nearest_scanline \ + pixman_scaled_nearest_scanline_8888_0565_SRC_asm_neon, 32, 0, 16, \ + FLAG_DST_WRITEONLY | FLAG_DEINTERLEAVE_32BPP, \ + 8, /* number of pixels, processed in a single block */ \ + default_init, \ + default_cleanup, \ + pixman_composite_src_8888_0565_process_pixblock_head, \ + pixman_composite_src_8888_0565_process_pixblock_tail, \ + pixman_composite_src_8888_0565_process_pixblock_tail_head diff --git a/pixman/pixman-arm-neon.c b/pixman/pixman-arm-neon.c index a200ed1..bf921e6 100644 --- a/pixman/pixman-arm-neon.c +++ b/pixman/pixman-arm-neon.c @@ -101,6 +101,8 @@ PIXMAN_ARM_BIND_SCALED_NEAREST_SRC_DST (neon, 8888_8888, OVER, uint32_t, uint32_t) PIXMAN_ARM_BIND_SCALED_NEAREST_SRC_DST (neon, 8888_0565, OVER, uint32_t, uint16_t) +PIXMAN_ARM_BIND_SCALED_NEAREST_SRC_DST (neon, 8888_0565, SRC, + uint32_t, uint16_t) void pixman_composite_src_n_8_asm_neon (int32_t w, @@ -284,6 +286,11 @@ static const pixman_fast_path_t arm_neon_fast_paths[] = PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (OVER, a8r8g8b8, r5g6b5, neon_8888_0565), PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (OVER, a8b8g8r8, b5g6r5, neon_8888_0565), + PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, a8r8g8b8, r5g6b5, neon_8888_0565), + PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, x8r8g8b8, r5g6b5, neon_8888_0565), + PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, a8b8g8r8, b5g6r5, neon_8888_0565), + PIXMAN_ARM_SIMPLE_NEAREST_FAST_PATH (SRC, x8b8g8r8, b5g6r5, neon_8888_0565), + { PIXMAN_OP_NONE }, }; -- 2.7.4