diff options
Diffstat (limited to 'meta-oe/recipes-graphics/xorg-lib/pixman-0.20.0/0008-ARM-optimization-for-scaled-src_0565_0565-operation-.patch')
| -rw-r--r-- | meta-oe/recipes-graphics/xorg-lib/pixman-0.20.0/0008-ARM-optimization-for-scaled-src_0565_0565-operation-.patch | 172 |
1 files changed, 172 insertions, 0 deletions
diff --git a/meta-oe/recipes-graphics/xorg-lib/pixman-0.20.0/0008-ARM-optimization-for-scaled-src_0565_0565-operation-.patch b/meta-oe/recipes-graphics/xorg-lib/pixman-0.20.0/0008-ARM-optimization-for-scaled-src_0565_0565-operation-.patch new file mode 100644 index 0000000000..6efdb621ad --- /dev/null +++ b/meta-oe/recipes-graphics/xorg-lib/pixman-0.20.0/0008-ARM-optimization-for-scaled-src_0565_0565-operation-.patch @@ -0,0 +1,172 @@ +From e1191ad6563a1fb02a45982b1c4d7fed3c655e97 Mon Sep 17 00:00:00 2001 +From: Siarhei Siamashka <siarhei.siamashka@nokia.com> +Date: Mon, 4 Oct 2010 01:56:59 +0300 +Subject: [PATCH 8/8] ARM optimization for scaled src_0565_0565 operation with nearest filter + +The code actually uses only armv4t instructions. + +Benchmark from ARM11: + + == before == + op=1, src_fmt=10020565, dst_fmt=10020565, speed=34.86 MPix/s + + == after == + op=1, src_fmt=10020565, dst_fmt=10020565, speed=36.62 MPix/s + +Benchmark from ARM Cortex-A8: + + == before == + op=1, src_fmt=10020565, dst_fmt=10020565, speed=89.55 MPix/s + + == after == + op=1, src_fmt=10020565, dst_fmt=10020565, speed=94.91 MPix/s +--- + pixman/pixman-arm-simd-asm.S | 66 ++++++++++++++++++++++++++++++++++++++++++ + pixman/pixman-arm-simd.c | 37 +++++++++++++++++++++++ + 2 files changed, 103 insertions(+), 0 deletions(-) + +diff --git a/pixman/pixman-arm-simd-asm.S b/pixman/pixman-arm-simd-asm.S +index a3d2d40..b6f69db 100644 +--- a/pixman/pixman-arm-simd-asm.S ++++ b/pixman/pixman-arm-simd-asm.S +@@ -1,5 +1,6 @@ + /* + * Copyright © 2008 Mozilla Corporation ++ * Copyright © 2010 Nokia Corporation + * + * Permission to use, copy, modify, distribute, and sell this software and its + * documentation for any purpose is hereby granted without fee, provided that +@@ -328,3 +329,68 @@ pixman_asm_function pixman_composite_over_n_8_8888_asm_armv6 + pop {r4, r5, r6, r7, r8, r9, r10, r11} + bx lr + .endfunc ++ ++/* ++ * Note: This function is actually primarily optimized for ARM Cortex-A8 ++ * pipeline. In order to get good performance on ARM9/ARM11 cores (which ++ * don't have efficient write combining), it needs to be changed to use ++ * 16-byte aligned writes using STM instruction. ++ */ ++pixman_asm_function pixman_scaled_nearest_scanline_565_565_SRC_asm_armv6 ++ DST .req r0 ++ SRC .req r1 ++ W .req r2 ++ VX .req r3 ++ UNIT_X .req r12 ++ TMP1 .req r4 ++ TMP2 .req r5 ++ MASK .req r6 ++ ldr UNIT_X, [sp] ++ push {r4, r5, r6, r7} ++ mvn MASK, #1 ++ ++ /* define helper macro */ ++ .macro scale_2_pixels ++ ldrh TMP1, [SRC, TMP1] ++ and TMP2, MASK, VX, lsr #15 ++ add VX, VX, UNIT_X ++ strh TMP1, [DST], #2 ++ ++ ldrh TMP2, [SRC, TMP2] ++ and TMP1, MASK, VX, lsr #15 ++ add VX, VX, UNIT_X ++ strh TMP2, [DST], #2 ++ .endm ++ ++ /* now do the scaling */ ++ and TMP1, MASK, VX, lsr #15 ++ add VX, VX, UNIT_X ++ subs W, #4 ++ blt 2f ++1: /* main loop, process 4 pixels per iteration */ ++ scale_2_pixels ++ scale_2_pixels ++ subs W, W, #4 ++ bge 1b ++2: ++ tst W, #2 ++ beq 2f ++ scale_2_pixels ++2: ++ tst W, #1 ++ ldrneh TMP1, [SRC, TMP1] ++ strneh TMP1, [DST], #2 ++ /* cleanup helper macro */ ++ .purgem scale_2_pixels ++ .unreq DST ++ .unreq SRC ++ .unreq W ++ .unreq VX ++ .unreq UNIT_X ++ .unreq TMP1 ++ .unreq TMP2 ++ .unreq MASK ++ /* return */ ++ pop {r4, r5, r6, r7} ++ bx lr ++.endfunc +diff --git a/pixman/pixman-arm-simd.c b/pixman/pixman-arm-simd.c +index d466a31..f6f464c 100644 +--- a/pixman/pixman-arm-simd.c ++++ b/pixman/pixman-arm-simd.c +@@ -29,6 +29,7 @@ + + #include "pixman-private.h" + #include "pixman-arm-common.h" ++#include "pixman-fast-path.h" + + #if 0 /* This code was moved to 'pixman-arm-simd-asm.S' */ + +@@ -375,6 +376,35 @@ pixman_composite_over_n_8_8888_asm_armv6 (int32_t width, + + #endif + ++void ++pixman_scaled_nearest_scanline_565_565_SRC_asm_armv6 (uint16_t * dst, ++ uint16_t * src, ++ int32_t w, ++ pixman_fixed_t vx, ++ pixman_fixed_t unit_x); ++ ++static force_inline void ++scaled_nearest_scanline_armv6_565_565_SRC (uint16_t * dst, ++ uint16_t * src, ++ int32_t w, ++ pixman_fixed_t vx, ++ pixman_fixed_t unit_x, ++ pixman_fixed_t max_vx) ++{ ++ pixman_scaled_nearest_scanline_565_565_SRC_asm_armv6 (dst, src, w, ++ vx, unit_x); ++} ++ ++FAST_NEAREST_MAINLOOP (armv6_565_565_cover_SRC, ++ scaled_nearest_scanline_armv6_565_565_SRC, ++ uint16_t, uint16_t, COVER); ++FAST_NEAREST_MAINLOOP (armv6_565_565_none_SRC, ++ scaled_nearest_scanline_armv6_565_565_SRC, ++ uint16_t, uint16_t, NONE); ++FAST_NEAREST_MAINLOOP (armv6_565_565_pad_SRC, ++ scaled_nearest_scanline_armv6_565_565_SRC, ++ uint16_t, uint16_t, PAD); ++ + PIXMAN_ARM_BIND_FAST_PATH_SRC_DST (armv6, add_8_8, + uint8_t, 1, uint8_t, 1) + PIXMAN_ARM_BIND_FAST_PATH_SRC_DST (armv6, over_8888_8888, +@@ -404,6 +434,13 @@ static const pixman_fast_path_t arm_simd_fast_paths[] = + PIXMAN_STD_FAST_PATH (OVER, solid, a8, a8b8g8r8, armv6_composite_over_n_8_8888), + PIXMAN_STD_FAST_PATH (OVER, solid, a8, x8b8g8r8, armv6_composite_over_n_8_8888), + ++ SIMPLE_NEAREST_FAST_PATH_COVER (SRC, r5g6b5, r5g6b5, armv6_565_565), ++ SIMPLE_NEAREST_FAST_PATH_COVER (SRC, b5g6r5, b5g6r5, armv6_565_565), ++ SIMPLE_NEAREST_FAST_PATH_NONE (SRC, r5g6b5, r5g6b5, armv6_565_565), ++ SIMPLE_NEAREST_FAST_PATH_NONE (SRC, b5g6r5, b5g6r5, armv6_565_565), ++ SIMPLE_NEAREST_FAST_PATH_PAD (SRC, r5g6b5, r5g6b5, armv6_565_565), ++ SIMPLE_NEAREST_FAST_PATH_PAD (SRC, b5g6r5, b5g6r5, armv6_565_565), ++ + { PIXMAN_OP_NONE }, + }; + +-- +1.6.6.1 + |