Merge pull request #9714 from tomoaki0705:universalBilateral

imgproc: use universal intrinsic as much as possible (#9714)

* use universal intrinsic as much as possible
  * make SSE3 part as common as possible with universal intrinsic implementation
  * put the reducing part out of the main loop

* follow the comment
  * fix the typo
  * use v_reduce_sum4

* follow the comment again
  * remove all CV_SSE3 part from smooth.cpp
This commit is contained in:
Tomoaki Teshima 2017-09-28 18:30:22 +09:00 committed by Alexander Alekhin
parent 488d4df520
commit 139b32734e

View File

@ -3198,12 +3198,10 @@ public:
{ {
int i, j, cn = dest->channels(), k; int i, j, cn = dest->channels(), k;
Size size = dest->size(); Size size = dest->size();
#if CV_SSE3 #if CV_SIMD128
int CV_DECL_ALIGNED(16) buf[4]; int CV_DECL_ALIGNED(16) buf[4];
float CV_DECL_ALIGNED(16) bufSum[4]; bool haveSIMD128 = hasSIMD128();
static const unsigned int CV_DECL_ALIGNED(16) bufSignMask[] = { 0x80000000, 0x80000000, 0x80000000, 0x80000000 }; #endif
bool haveSSE3 = checkHardwareSupport(CV_CPU_SSE3);
#endif
for( i = range.start; i < range.end; i++ ) for( i = range.start; i < range.end; i++ )
{ {
@ -3217,35 +3215,40 @@ public:
float sum = 0, wsum = 0; float sum = 0, wsum = 0;
int val0 = sptr[j]; int val0 = sptr[j];
k = 0; k = 0;
#if CV_SSE3 #if CV_SIMD128
if( haveSSE3 ) if( haveSIMD128 )
{ {
__m128 _val0 = _mm_set1_ps(static_cast<float>(val0)); v_float32x4 _val0 = v_setall_f32(static_cast<float>(val0));
const __m128 _signMask = _mm_load_ps((const float*)bufSignMask); v_float32x4 vsumw = v_setzero_f32();
v_float32x4 vsumc = v_setzero_f32();
for( ; k <= maxk - 4; k += 4 ) for( ; k <= maxk - 4; k += 4 )
{ {
__m128 _valF = _mm_set_ps(sptr[j + space_ofs[k+3]], sptr[j + space_ofs[k+2]], v_float32x4 _valF = v_float32x4(sptr[j + space_ofs[k]],
sptr[j + space_ofs[k+1]], sptr[j + space_ofs[k]]); sptr[j + space_ofs[k + 1]],
sptr[j + space_ofs[k + 2]],
sptr[j + space_ofs[k + 3]]);
v_float32x4 _val = v_abs(_valF - _val0);
v_store(buf, v_round(_val));
__m128 _val = _mm_andnot_ps(_signMask, _mm_sub_ps(_valF, _val0)); v_float32x4 _cw = v_float32x4(color_weight[buf[0]],
_mm_store_si128((__m128i*)buf, _mm_cvtps_epi32(_val)); color_weight[buf[1]],
color_weight[buf[2]],
color_weight[buf[3]]);
v_float32x4 _sw = v_load(space_weight+k);
v_float32x4 _w = _cw * _sw;
_cw = _w * _valF;
__m128 _cw = _mm_set_ps(color_weight[buf[3]],color_weight[buf[2]], vsumw += _w;
color_weight[buf[1]],color_weight[buf[0]]); vsumc += _cw;
__m128 _sw = _mm_loadu_ps(space_weight+k);
__m128 _w = _mm_mul_ps(_cw, _sw);
_cw = _mm_mul_ps(_w, _valF);
_sw = _mm_hadd_ps(_w, _cw);
_sw = _mm_hadd_ps(_sw, _sw);
_mm_storel_pi((__m64*)bufSum, _sw);
sum += bufSum[1];
wsum += bufSum[0];
} }
float *bufFloat = (float*)buf;
v_float32x4 sum4 = v_reduce_sum4(vsumw, vsumc, vsumw, vsumc);
v_store(bufFloat, sum4);
sum += bufFloat[1];
wsum += bufFloat[0];
} }
#endif #endif
for( ; k < maxk; k++ ) for( ; k < maxk; k++ )
{ {
int val = sptr[j + space_ofs[k]]; int val = sptr[j + space_ofs[k]];
@ -3265,58 +3268,62 @@ public:
float sum_b = 0, sum_g = 0, sum_r = 0, wsum = 0; float sum_b = 0, sum_g = 0, sum_r = 0, wsum = 0;
int b0 = sptr[j], g0 = sptr[j+1], r0 = sptr[j+2]; int b0 = sptr[j], g0 = sptr[j+1], r0 = sptr[j+2];
k = 0; k = 0;
#if CV_SSE3 #if CV_SIMD128
if( haveSSE3 ) if( haveSIMD128 )
{ {
const __m128i izero = _mm_setzero_si128(); v_float32x4 vsumw = v_setzero_f32();
const __m128 _b0 = _mm_set1_ps(static_cast<float>(b0)); v_float32x4 vsumb = v_setzero_f32();
const __m128 _g0 = _mm_set1_ps(static_cast<float>(g0)); v_float32x4 vsumg = v_setzero_f32();
const __m128 _r0 = _mm_set1_ps(static_cast<float>(r0)); v_float32x4 vsumr = v_setzero_f32();
const __m128 _signMask = _mm_load_ps((const float*)bufSignMask); const v_float32x4 _b0 = v_setall_f32(static_cast<float>(b0));
const v_float32x4 _g0 = v_setall_f32(static_cast<float>(g0));
const v_float32x4 _r0 = v_setall_f32(static_cast<float>(r0));
for( ; k <= maxk - 4; k += 4 ) for( ; k <= maxk - 4; k += 4 )
{ {
const int* const sptr_k0 = reinterpret_cast<const int*>(sptr + j + space_ofs[k]); const uchar* const sptr_k0 = sptr + j + space_ofs[k];
const int* const sptr_k1 = reinterpret_cast<const int*>(sptr + j + space_ofs[k+1]); const uchar* const sptr_k1 = sptr + j + space_ofs[k+1];
const int* const sptr_k2 = reinterpret_cast<const int*>(sptr + j + space_ofs[k+2]); const uchar* const sptr_k2 = sptr + j + space_ofs[k+2];
const int* const sptr_k3 = reinterpret_cast<const int*>(sptr + j + space_ofs[k+3]); const uchar* const sptr_k3 = sptr + j + space_ofs[k+3];
__m128 _b = _mm_cvtepi32_ps(_mm_unpacklo_epi16(_mm_unpacklo_epi8(_mm_cvtsi32_si128(sptr_k0[0]), izero), izero)); v_float32x4 __b = v_cvt_f32(v_reinterpret_as_s32(v_load_expand_q(sptr_k0)));
__m128 _g = _mm_cvtepi32_ps(_mm_unpacklo_epi16(_mm_unpacklo_epi8(_mm_cvtsi32_si128(sptr_k1[0]), izero), izero)); v_float32x4 __g = v_cvt_f32(v_reinterpret_as_s32(v_load_expand_q(sptr_k1)));
__m128 _r = _mm_cvtepi32_ps(_mm_unpacklo_epi16(_mm_unpacklo_epi8(_mm_cvtsi32_si128(sptr_k2[0]), izero), izero)); v_float32x4 __r = v_cvt_f32(v_reinterpret_as_s32(v_load_expand_q(sptr_k2)));
__m128 _z = _mm_cvtepi32_ps(_mm_unpacklo_epi16(_mm_unpacklo_epi8(_mm_cvtsi32_si128(sptr_k3[0]), izero), izero)); v_float32x4 __z = v_cvt_f32(v_reinterpret_as_s32(v_load_expand_q(sptr_k3)));
v_float32x4 _b, _g, _r, _z;
_MM_TRANSPOSE4_PS(_b, _g, _r, _z); v_transpose4x4(__b, __g, __r, __z, _b, _g, _r, _z);
__m128 bt = _mm_andnot_ps(_signMask, _mm_sub_ps(_b,_b0)); v_float32x4 bt = v_abs(_b -_b0);
__m128 gt = _mm_andnot_ps(_signMask, _mm_sub_ps(_g,_g0)); v_float32x4 gt = v_abs(_g -_g0);
__m128 rt = _mm_andnot_ps(_signMask, _mm_sub_ps(_r,_r0)); v_float32x4 rt = v_abs(_r -_r0);
bt =_mm_add_ps(rt, _mm_add_ps(bt, gt)); bt = rt + bt + gt;
_mm_store_si128((__m128i*)buf, _mm_cvtps_epi32(bt)); v_store(buf, v_round(bt));
__m128 _w = _mm_set_ps(color_weight[buf[3]],color_weight[buf[2]], v_float32x4 _w = v_float32x4(color_weight[buf[0]],color_weight[buf[1]],
color_weight[buf[1]],color_weight[buf[0]]); color_weight[buf[2]],color_weight[buf[3]]);
__m128 _sw = _mm_loadu_ps(space_weight+k); v_float32x4 _sw = v_load(space_weight+k);
_w = _mm_mul_ps(_w,_sw); _w *= _sw;
_b = _mm_mul_ps(_b, _w); _b *= _w;
_g = _mm_mul_ps(_g, _w); _g *= _w;
_r = _mm_mul_ps(_r, _w); _r *= _w;
_w = _mm_hadd_ps(_w, _b); vsumw += _w;
_g = _mm_hadd_ps(_g, _r); vsumb += _b;
vsumg += _g;
_w = _mm_hadd_ps(_w, _g); vsumr += _r;
_mm_store_ps(bufSum, _w); }
float *bufFloat = (float*)buf;
wsum += bufSum[0]; v_float32x4 sum4 = v_reduce_sum4(vsumw, vsumb, vsumg, vsumr);
sum_b += bufSum[1]; v_store(bufFloat, sum4);
sum_g += bufSum[2]; wsum += bufFloat[0];
sum_r += bufSum[3]; sum_b += bufFloat[1];
} sum_g += bufFloat[2];
sum_r += bufFloat[3];
} }
#endif #endif
for( ; k < maxk; k++ ) for( ; k < maxk; k++ )
{ {
@ -3515,16 +3522,10 @@ public:
{ {
int i, j, k; int i, j, k;
Size size = dest->size(); Size size = dest->size();
#if CV_SSE3 || CV_NEON #if CV_SIMD128
int CV_DECL_ALIGNED(16) idxBuf[4]; int CV_DECL_ALIGNED(16) idxBuf[4];
float CV_DECL_ALIGNED(16) bufSum32[4]; bool haveSIMD128 = hasSIMD128();
static const unsigned int CV_DECL_ALIGNED(16) bufSignMask[] = { 0x80000000, 0x80000000, 0x80000000, 0x80000000 }; #endif
#endif
#if CV_SSE3
bool haveSSE3 = checkHardwareSupport(CV_CPU_SSE3);
#elif CV_NEON
bool haveNEON = checkHardwareSupport(CV_CPU_NEON);
#endif
for( i = range.start; i < range.end; i++ ) for( i = range.start; i < range.end; i++ )
{ {
@ -3538,84 +3539,49 @@ public:
float sum = 0, wsum = 0; float sum = 0, wsum = 0;
float val0 = sptr[j]; float val0 = sptr[j];
k = 0; k = 0;
#if CV_SSE3 #if CV_SIMD128
if( haveSSE3 ) if( haveSIMD128 )
{ {
__m128 psum = _mm_setzero_ps(); v_float32x4 vecwsum = v_setzero_f32();
const __m128 _val0 = _mm_set1_ps(sptr[j]); v_float32x4 vecvsum = v_setzero_f32();
const __m128 _scale_index = _mm_set1_ps(scale_index); const v_float32x4 _val0 = v_setall_f32(sptr[j]);
const __m128 _signMask = _mm_load_ps((const float*)bufSignMask); const v_float32x4 _scale_index = v_setall_f32(scale_index);
for( ; k <= maxk - 4 ; k += 4 ) for (; k <= maxk - 4; k += 4)
{ {
__m128 _sw = _mm_loadu_ps(space_weight + k); v_float32x4 _sw = v_load(space_weight + k);
__m128 _val = _mm_set_ps(sptr[j + space_ofs[k+3]], sptr[j + space_ofs[k+2]], v_float32x4 _val = v_float32x4(sptr[j + space_ofs[k]],
sptr[j + space_ofs[k+1]], sptr[j + space_ofs[k]]); sptr[j + space_ofs[k + 1]],
__m128 _alpha = _mm_mul_ps(_mm_andnot_ps( _signMask, _mm_sub_ps(_val,_val0)), _scale_index); sptr[j + space_ofs[k + 2]],
sptr[j + space_ofs[k + 3]]);
v_float32x4 _alpha = v_abs(_val - _val0) * _scale_index;
__m128i _idx = _mm_cvtps_epi32(_alpha); v_int32x4 _idx = v_round(_alpha);
_mm_store_si128((__m128i*)idxBuf, _idx); v_store(idxBuf, _idx);
_alpha = _mm_sub_ps(_alpha, _mm_cvtepi32_ps(_idx)); _alpha -= v_cvt_f32(_idx);
__m128 _explut = _mm_set_ps(expLUT[idxBuf[3]], expLUT[idxBuf[2]], v_float32x4 _explut = v_float32x4(expLUT[idxBuf[0]],
expLUT[idxBuf[1]], expLUT[idxBuf[0]]); expLUT[idxBuf[1]],
__m128 _explut1 = _mm_set_ps(expLUT[idxBuf[3]+1], expLUT[idxBuf[2]+1], expLUT[idxBuf[2]],
expLUT[idxBuf[1]+1], expLUT[idxBuf[0]+1]); expLUT[idxBuf[3]]);
v_float32x4 _explut1 = v_float32x4(expLUT[idxBuf[0] + 1],
expLUT[idxBuf[1] + 1],
expLUT[idxBuf[2] + 1],
expLUT[idxBuf[3] + 1]);
__m128 _w = _mm_mul_ps(_sw, _mm_add_ps(_explut, _mm_mul_ps(_alpha, _mm_sub_ps(_explut1, _explut)))); v_float32x4 _w = _sw * (_explut + (_alpha * (_explut1 - _explut)));
_val = _mm_mul_ps(_w, _val); _val *= _w;
_sw = _mm_hadd_ps(_w, _val); vecwsum += _w;
_sw = _mm_hadd_ps(_sw, _sw); vecvsum += _val;
psum = _mm_add_ps(_sw, psum);
} }
_mm_storel_pi((__m64*)bufSum32, psum); float *bufFloat = (float*)idxBuf;
v_float32x4 sum4 = v_reduce_sum4(vecwsum, vecvsum, vecwsum, vecvsum);
sum = bufSum32[1]; v_store(bufFloat, sum4);
wsum = bufSum32[0]; sum += bufFloat[1];
wsum += bufFloat[0];
} }
#elif CV_NEON #endif
if( haveNEON )
{
float32x2_t psum = vdup_n_f32(0.0f);
const volatile float32x4_t _val0 = vdupq_n_f32(sptr[j]);
const float32x4_t _scale_index = vdupq_n_f32(scale_index);
const uint32x4_t _signMask = vld1q_u32(bufSignMask);
for( ; k <= maxk - 4 ; k += 4 )
{
float32x4_t _sw = vld1q_f32(space_weight + k);
float CV_DECL_ALIGNED(16) _data[] = {sptr[j + space_ofs[k]], sptr[j + space_ofs[k+1]],
sptr[j + space_ofs[k+2]], sptr[j + space_ofs[k+3]],};
float32x4_t _val = vld1q_f32(_data);
float32x4_t _alpha = vsubq_f32(_val, _val0);
_alpha = vreinterpretq_f32_u32(vbicq_u32(vreinterpretq_u32_f32(_alpha), _signMask));
_alpha = vmulq_f32(_alpha, _scale_index);
int32x4_t _idx = vcvtq_s32_f32(_alpha);
vst1q_s32(idxBuf, _idx);
_alpha = vsubq_f32(_alpha, vcvtq_f32_s32(_idx));
bufSum32[0] = expLUT[idxBuf[0]];
bufSum32[1] = expLUT[idxBuf[1]];
bufSum32[2] = expLUT[idxBuf[2]];
bufSum32[3] = expLUT[idxBuf[3]];
float32x4_t _explut = vld1q_f32(bufSum32);
bufSum32[0] = expLUT[idxBuf[0]+1];
bufSum32[1] = expLUT[idxBuf[1]+1];
bufSum32[2] = expLUT[idxBuf[2]+1];
bufSum32[3] = expLUT[idxBuf[3]+1];
float32x4_t _explut1 = vld1q_f32(bufSum32);
float32x4_t _w = vmulq_f32(_sw, vaddq_f32(_explut, vmulq_f32(_alpha, vsubq_f32(_explut1, _explut))));
_val = vmulq_f32(_w, _val);
float32x2_t _wval = vpadd_f32(vpadd_f32(vget_low_f32(_w),vget_high_f32(_w)), vpadd_f32(vget_low_f32(_val), vget_high_f32(_val)));
psum = vadd_f32(_wval, psum);
}
sum = vget_lane_f32(psum, 1);
wsum = vget_lane_f32(psum, 0);
}
#endif
for( ; k < maxk; k++ ) for( ; k < maxk; k++ )
{ {
@ -3638,129 +3604,70 @@ public:
float sum_b = 0, sum_g = 0, sum_r = 0, wsum = 0; float sum_b = 0, sum_g = 0, sum_r = 0, wsum = 0;
float b0 = sptr[j], g0 = sptr[j+1], r0 = sptr[j+2]; float b0 = sptr[j], g0 = sptr[j+1], r0 = sptr[j+2];
k = 0; k = 0;
#if CV_SSE3 #if CV_SIMD128
if( haveSSE3 ) if( haveSIMD128 )
{ {
__m128 sum = _mm_setzero_ps(); v_float32x4 sumw = v_setzero_f32();
const __m128 _b0 = _mm_set1_ps(b0); v_float32x4 sumb = v_setzero_f32();
const __m128 _g0 = _mm_set1_ps(g0); v_float32x4 sumg = v_setzero_f32();
const __m128 _r0 = _mm_set1_ps(r0); v_float32x4 sumr = v_setzero_f32();
const __m128 _scale_index = _mm_set1_ps(scale_index); const v_float32x4 _b0 = v_setall_f32(b0);
const __m128 _signMask = _mm_load_ps((const float*)bufSignMask); const v_float32x4 _g0 = v_setall_f32(g0);
const v_float32x4 _r0 = v_setall_f32(r0);
const v_float32x4 _scale_index = v_setall_f32(scale_index);
for( ; k <= maxk-4; k += 4 ) for( ; k <= maxk-4; k += 4 )
{ {
__m128 _sw = _mm_loadu_ps(space_weight + k); v_float32x4 _sw = v_load(space_weight + k);
const float* const sptr_k0 = sptr + j + space_ofs[k]; const float* const sptr_k0 = sptr + j + space_ofs[k];
const float* const sptr_k1 = sptr + j + space_ofs[k+1]; const float* const sptr_k1 = sptr + j + space_ofs[k+1];
const float* const sptr_k2 = sptr + j + space_ofs[k+2]; const float* const sptr_k2 = sptr + j + space_ofs[k+2];
const float* const sptr_k3 = sptr + j + space_ofs[k+3]; const float* const sptr_k3 = sptr + j + space_ofs[k+3];
__m128 _b = _mm_loadu_ps(sptr_k0); v_float32x4 _v0 = v_load(sptr_k0);
__m128 _g = _mm_loadu_ps(sptr_k1); v_float32x4 _v1 = v_load(sptr_k1);
__m128 _r = _mm_loadu_ps(sptr_k2); v_float32x4 _v2 = v_load(sptr_k2);
__m128 _z = _mm_loadu_ps(sptr_k3); v_float32x4 _v3 = v_load(sptr_k3);
_MM_TRANSPOSE4_PS(_b, _g, _r, _z); v_float32x4 _b, _g, _r, _dummy;
__m128 _bt = _mm_andnot_ps(_signMask,_mm_sub_ps(_b,_b0)); v_transpose4x4(_v0, _v1, _v2, _v3, _b, _g, _r, _dummy);
__m128 _gt = _mm_andnot_ps(_signMask,_mm_sub_ps(_g,_g0));
__m128 _rt = _mm_andnot_ps(_signMask,_mm_sub_ps(_r,_r0));
__m128 _alpha = _mm_mul_ps(_scale_index, _mm_add_ps(_rt,_mm_add_ps(_bt, _gt))); v_float32x4 _bt = v_abs(_b - _b0);
v_float32x4 _gt = v_abs(_g - _g0);
v_float32x4 _rt = v_abs(_r - _r0);
v_float32x4 _alpha = _scale_index * (_bt + _gt + _rt);
__m128i _idx = _mm_cvtps_epi32(_alpha); v_int32x4 _idx = v_round(_alpha);
_mm_store_si128((__m128i*)idxBuf, _idx); v_store((int*)idxBuf, _idx);
_alpha = _mm_sub_ps(_alpha, _mm_cvtepi32_ps(_idx)); v_float32x4 _explut = v_float32x4(expLUT[idxBuf[0]],
expLUT[idxBuf[1]],
expLUT[idxBuf[2]],
expLUT[idxBuf[3]]);
v_float32x4 _explut1 = v_float32x4(expLUT[idxBuf[0] + 1],
expLUT[idxBuf[1] + 1],
expLUT[idxBuf[2] + 1],
expLUT[idxBuf[3] + 1]);
__m128 _explut = _mm_set_ps(expLUT[idxBuf[3]], expLUT[idxBuf[2]], expLUT[idxBuf[1]], expLUT[idxBuf[0]]); v_float32x4 _w = _sw * (_explut + (_alpha * (_explut1 - _explut)));
__m128 _explut1 = _mm_set_ps(expLUT[idxBuf[3]+1], expLUT[idxBuf[2]+1], expLUT[idxBuf[1]+1], expLUT[idxBuf[0]+1]);
__m128 _w = _mm_mul_ps(_sw, _mm_add_ps(_explut, _mm_mul_ps(_alpha, _mm_sub_ps(_explut1, _explut)))); _b *= _w;
_g *= _w;
_b = _mm_mul_ps(_b, _w); _r *= _w;
_g = _mm_mul_ps(_g, _w); sumw += _w;
_r = _mm_mul_ps(_r, _w); sumb += _b;
sumg += _g;
_w = _mm_hadd_ps(_w, _b); sumr += _r;
_g = _mm_hadd_ps(_g, _r);
_w = _mm_hadd_ps(_w, _g);
sum = _mm_add_ps(sum, _w);
} }
_mm_store_ps(bufSum32, sum); v_float32x4 sum4 = v_reduce_sum4(sumw, sumb, sumg, sumr);
wsum = bufSum32[0]; float *bufFloat = (float*)idxBuf;
sum_b = bufSum32[1]; v_store(bufFloat, sum4);
sum_g = bufSum32[2]; wsum += bufFloat[0];
sum_r = bufSum32[3]; sum_b += bufFloat[1];
sum_g += bufFloat[2];
sum_r += bufFloat[3];
} }
#elif CV_NEON #endif
if( haveNEON )
{
float32x4_t sum = vdupq_n_f32(0.0f);
const float32x4_t _b0 = vdupq_n_f32(b0);
const float32x4_t _g0 = vdupq_n_f32(g0);
const float32x4_t _r0 = vdupq_n_f32(r0);
const float32x4_t _scale_index = vdupq_n_f32(scale_index);
const uint32x4_t _signMask = vld1q_u32(bufSignMask);
for( ; k <= maxk-4; k += 4 )
{
float32x4_t _sw = vld1q_f32(space_weight + k);
const float* const sptr_k0 = sptr + j + space_ofs[k];
const float* const sptr_k1 = sptr + j + space_ofs[k+1];
const float* const sptr_k2 = sptr + j + space_ofs[k+2];
const float* const sptr_k3 = sptr + j + space_ofs[k+3];
float32x4_t _v0 = vld1q_f32(sptr_k0);
float32x4_t _v1 = vld1q_f32(sptr_k1);
float32x4_t _v2 = vld1q_f32(sptr_k2);
float32x4_t _v3 = vld1q_f32(sptr_k3);
float32x4x2_t v01 = vtrnq_f32(_v0, _v1);
float32x4x2_t v23 = vtrnq_f32(_v2, _v3);
float32x4_t _b = vcombine_f32(vget_low_f32(v01.val[0]), vget_low_f32(v23.val[0]));
float32x4_t _g = vcombine_f32(vget_low_f32(v01.val[1]), vget_low_f32(v23.val[1]));
float32x4_t _r = vcombine_f32(vget_high_f32(v01.val[0]), vget_high_f32(v23.val[0]));
float32x4_t _bt = vreinterpretq_f32_u32(vbicq_u32(vreinterpretq_u32_f32(vsubq_f32(_b, _b0)), _signMask));
float32x4_t _gt = vreinterpretq_f32_u32(vbicq_u32(vreinterpretq_u32_f32(vsubq_f32(_g, _g0)), _signMask));
float32x4_t _rt = vreinterpretq_f32_u32(vbicq_u32(vreinterpretq_u32_f32(vsubq_f32(_r, _r0)), _signMask));
float32x4_t _alpha = vmulq_f32(_scale_index, vaddq_f32(_bt, vaddq_f32(_gt, _rt)));
int32x4_t _idx = vcvtq_s32_f32(_alpha);
vst1q_s32((int*)idxBuf, _idx);
bufSum32[0] = expLUT[idxBuf[0]];
bufSum32[1] = expLUT[idxBuf[1]];
bufSum32[2] = expLUT[idxBuf[2]];
bufSum32[3] = expLUT[idxBuf[3]];
float32x4_t _explut = vld1q_f32(bufSum32);
bufSum32[0] = expLUT[idxBuf[0]+1];
bufSum32[1] = expLUT[idxBuf[1]+1];
bufSum32[2] = expLUT[idxBuf[2]+1];
bufSum32[3] = expLUT[idxBuf[3]+1];
float32x4_t _explut1 = vld1q_f32(bufSum32);
float32x4_t _w = vmulq_f32(_sw, vaddq_f32(_explut, vmulq_f32(_alpha, vsubq_f32(_explut1, _explut))));
_b = vmulq_f32(_b, _w);
_g = vmulq_f32(_g, _w);
_r = vmulq_f32(_r, _w);
float32x2_t _wb = vpadd_f32(vpadd_f32(vget_low_f32(_w),vget_high_f32(_w)), vpadd_f32(vget_low_f32(_b), vget_high_f32(_b)));
float32x2_t _gr = vpadd_f32(vpadd_f32(vget_low_f32(_g),vget_high_f32(_g)), vpadd_f32(vget_low_f32(_r), vget_high_f32(_r)));
_w = vcombine_f32(_wb, _gr);
sum = vaddq_f32(sum, _w);
}
vst1q_f32(bufSum32, sum);
wsum = bufSum32[0];
sum_b = bufSum32[1];
sum_g = bufSum32[2];
sum_r = bufSum32[3];
}
#endif
for(; k < maxk; k++ ) for(; k < maxk; k++ )
{ {