dany/libwebp
1
0
Fork 0
mirror of https://github.com/webmproject/libwebp.git synced 2025-07-17 22:39:52 +02:00
Code Issues Actions Packages Projects Releases Wiki Activity

10% faster table-less SSE2/NEON version of YUV->RGB conversion

Browse Source 
* Precision is slightly different
* also implemented in SSE2 the missing WebPUpsamplers for MODE_ARGB, MODE_Argb, MODE_RGB565, etc.
* removing yuv_tables_sse2.h saved ~8k of binary size
* the mips32/mips_dsp_r2 code is disabled for now, since it has drifted away
* the NEON code is somewhat tricky

Change-Id: Icf205faa62cf46c2825d79f3af6725dc1ec7f052
This commit is contained in:
Pascal Massimino
2015-12-08 12:31:47 +01:00
committed by James Zern
parent bd91af200a
commit ac761a3738
11 changed files with 286 additions and 887 deletions
Download Patch File Download Diff File Expand all files Collapse all files

68
src/dsp/upsampling_neon.c
View File

@ -89,7 +89,8 @@ static void Upsample16Pixels(const uint8_t *r1, const uint8_t *r2,
//-----------------------------------------------------------------------------
// YUV->RGB conversion
static const int16_t kCoeffs[4] = { kYScale, kVToR, kUToG, kVToG };
// note: we represent the 33050 large constant as 32768 + 282
static const int16_t kCoeffs1[4] = { 19077, 26149, 6419, 13320 };
#define v255 vdup_n_u8(255)
@ -117,38 +118,35 @@ static const int16_t kCoeffs[4] = { kYScale, kVToR, kUToG, kVToG };
vst4_u8(out, b_g_r_v255); \
} while (0)
#define CONVERT8(FMT, XSTEP, N, src_y, src_uv, out, cur_x) { \
#define CONVERT8(FMT, XSTEP, N, src_y, src_uv, out, cur_x) do { \
int i; \
for (i = 0; i < N; i += 8) { \
const int off = ((cur_x) + i) * XSTEP; \
uint8x8_t y = vld1_u8((src_y) + (cur_x) + i); \
uint8x8_t u = vld1_u8((src_uv) + i); \
uint8x8_t v = vld1_u8((src_uv) + i + 16); \
const int16x8_t yy = vreinterpretq_s16_u16(vsubl_u8(y, u16)); \
const int16x8_t uu = vreinterpretq_s16_u16(vsubl_u8(u, u128)); \
const int16x8_t vv = vreinterpretq_s16_u16(vsubl_u8(v, u128)); \
int32x4_t yl = vmull_lane_s16(vget_low_s16(yy), cf16, 0); \
int32x4_t yh = vmull_lane_s16(vget_high_s16(yy), cf16, 0); \
const int32x4_t rl = vmlal_lane_s16(yl, vget_low_s16(vv), cf16, 1);\
const int32x4_t rh = vmlal_lane_s16(yh, vget_high_s16(vv), cf16, 1);\
int32x4_t gl = vmlsl_lane_s16(yl, vget_low_s16(uu), cf16, 2); \
int32x4_t gh = vmlsl_lane_s16(yh, vget_high_s16(uu), cf16, 2); \
const int32x4_t bl = vmovl_s16(vget_low_s16(uu)); \
const int32x4_t bh = vmovl_s16(vget_high_s16(uu)); \
gl = vmlsl_lane_s16(gl, vget_low_s16(vv), cf16, 3); \
gh = vmlsl_lane_s16(gh, vget_high_s16(vv), cf16, 3); \
yl = vmlaq_lane_s32(yl, bl, cf32, 0); \
yh = vmlaq_lane_s32(yh, bh, cf32, 0); \
/* vrshrn_n_s32() already incorporates the rounding constant */ \
y = vqmovun_s16(vcombine_s16(vrshrn_n_s32(rl, YUV_FIX2), \
vrshrn_n_s32(rh, YUV_FIX2))); \
u = vqmovun_s16(vcombine_s16(vrshrn_n_s32(gl, YUV_FIX2), \
vrshrn_n_s32(gh, YUV_FIX2))); \
v = vqmovun_s16(vcombine_s16(vrshrn_n_s32(yl, YUV_FIX2), \
vrshrn_n_s32(yh, YUV_FIX2))); \
STORE_ ## FMT(out + off, y, u, v); \
const uint8x8_t y = vld1_u8((src_y) + (cur_x) + i); \
const uint8x8_t u = vld1_u8((src_uv) + i + 0); \
const uint8x8_t v = vld1_u8((src_uv) + i + 16); \
const int16x8_t Y0 = vreinterpretq_s16_u16(vshll_n_u8(y, 7)); \
const int16x8_t U0 = vreinterpretq_s16_u16(vshll_n_u8(u, 7)); \
const int16x8_t V0 = vreinterpretq_s16_u16(vshll_n_u8(v, 7)); \
const int16x8_t Y1 = vqdmulhq_lane_s16(Y0, coeff1, 0); \
const int16x8_t R0 = vqdmulhq_lane_s16(V0, coeff1, 1); \
const int16x8_t G0 = vqdmulhq_lane_s16(U0, coeff1, 2); \
const int16x8_t G1 = vqdmulhq_lane_s16(V0, coeff1, 3); \
const int16x8_t B0 = vqdmulhq_n_s16(U0, 282); \
const int16x8_t R1 = vqaddq_s16(Y1, R_Rounder); \
const int16x8_t G2 = vqaddq_s16(Y1, G_Rounder); \
const int16x8_t B1 = vqaddq_s16(Y1, B_Rounder); \
const int16x8_t R2 = vqaddq_s16(R0, R1); \
const int16x8_t G3 = vqaddq_s16(G0, G1); \
const int16x8_t B2 = vqaddq_s16(B0, B1); \
const int16x8_t G4 = vqsubq_s16(G2, G3); \
const int16x8_t B3 = vqaddq_s16(B2, U0); \
const uint8x8_t R = vqshrun_n_s16(R2, YUV_FIX2); \
const uint8x8_t G = vqshrun_n_s16(G4, YUV_FIX2); \
const uint8x8_t B = vqshrun_n_s16(B3, YUV_FIX2); \
STORE_ ## FMT(out + off, R, G, B); \
} \
}
} while (0)
#define CONVERT1(FUNC, XSTEP, N, src_y, src_uv, rgb, cur_x) { \
int i; \
@ -163,9 +161,9 @@ static const int16_t kCoeffs[4] = { kYScale, kVToR, kUToG, kVToG };
#define CONVERT2RGB_8(FMT, XSTEP, top_y, bottom_y, uv, \
top_dst, bottom_dst, cur_x, len) { \
CONVERT8(FMT, XSTEP, len, top_y, uv, top_dst, cur_x) \
CONVERT8(FMT, XSTEP, len, top_y, uv, top_dst, cur_x); \
if (bottom_y != NULL) { \
CONVERT8(FMT, XSTEP, len, bottom_y, (uv) + 32, bottom_dst, cur_x) \
CONVERT8(FMT, XSTEP, len, bottom_y, (uv) + 32, bottom_dst, cur_x); \
} \
}
@ -195,10 +193,10 @@ static void FUNC_NAME(const uint8_t *top_y, const uint8_t *bottom_y, \
const int u_diag = ((top_u[0] + cur_u[0]) >> 1) + 1; \
const int v_diag = ((top_v[0] + cur_v[0]) >> 1) + 1; \
\
const int16x4_t cf16 = vld1_s16(kCoeffs); \
const int32x2_t cf32 = vdup_n_s32(kUToB); \
const uint8x8_t u16 = vdup_n_u8(16); \
const uint8x8_t u128 = vdup_n_u8(128); \
const int16x4_t coeff1 = vld1_s16(kCoeffs1); \
const int16x8_t R_Rounder = vdupq_n_s16(-14234); \
const int16x8_t G_Rounder = vdupq_n_s16(8708); \
const int16x8_t B_Rounder = vdupq_n_s16(-17685); \
\
/* Treat the first pixel in regular way */ \
assert(top_y != NULL); \