transform-sse2.c - third_party/qcms - Git at Google

 //  qcms
 //  Copyright (C) 2009 Mozilla Foundation
 //  Copyright (C) 2015 Intel Corporation
 //
 // Permission is hereby granted, free of charge, to any person obtaining
 // a copy of this software and associated documentation files (the "Software"),
 // to deal in the Software without restriction, including without limitation
 // the rights to use, copy, modify, merge, publish, distribute, sublicense,
 // and/or sell copies of the Software, and to permit persons to whom the Software
 // is furnished to do so, subject to the following conditions:
 //
 // The above copyright notice and this permission notice shall be included in
 // all copies or substantial portions of the Software.
 //
 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 // EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO
 // THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 // NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
 // LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
 // OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
 // WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

 #include <emmintrin.h>

 #include "qcmsint.h"

 /* pre-shuffled: just load these into XMM reg instead of load-scalar/shufps sequence */
 #define FLOATSCALE  (float)(PRECACHE_OUTPUT_SIZE - 1)
 #define CLAMPMAXVAL 1.0f

 static const ALIGN float floatScaleX4[4] =
     { FLOATSCALE, FLOATSCALE, FLOATSCALE, FLOATSCALE};
 static const ALIGN float clampMaxValueX4[4] =
     { CLAMPMAXVAL, CLAMPMAXVAL, CLAMPMAXVAL, CLAMPMAXVAL};

 void qcms_transform_data_rgb_out_lut_sse2(qcms_transform *transform,
                                           unsigned char *src,
                                           unsigned char *dest,
                                           size_t length,
                                           qcms_format_type output_format)
 {
     unsigned int i;
     float (*mat)[4] = transform->matrix;
     char input_back[32];
     /* Ensure we have a buffer that's 16 byte aligned regardless of the original
      * stack alignment. We can't use __attribute__((aligned(16))) or __declspec(align(32))
      * because they don't work on stack variables. gcc 4.4 does do the right thing
      * on x86 but that's too new for us right now. For more info: gcc bug #16660 */
     float const * input = (float*)(((uintptr_t)&input_back[16]) & ~0xf);
     /* share input and output locations to save having to keep the
      * locations in separate registers */
     uint32_t const * output = (uint32_t*)input;

     /* deref *transform now to avoid it in loop */
     const float *igtbl_r = transform->input_gamma_table_r;
     const float *igtbl_g = transform->input_gamma_table_g;
     const float *igtbl_b = transform->input_gamma_table_b;

     /* deref *transform now to avoid it in loop */
     const uint8_t *otdata_r = &transform->output_table_r->data[0];
     const uint8_t *otdata_g = &transform->output_table_g->data[0];
     const uint8_t *otdata_b = &transform->output_table_b->data[0];

     /* input matrix values never change */
     const __m128 mat0  = _mm_load_ps(mat[0]);
     const __m128 mat1  = _mm_load_ps(mat[1]);
     const __m128 mat2  = _mm_load_ps(mat[2]);

     /* these values don't change, either */
     const __m128 max   = _mm_load_ps(clampMaxValueX4);
     const __m128 min   = _mm_setzero_ps();
     const __m128 scale = _mm_load_ps(floatScaleX4);

     /* working variables */
     __m128 vec_r, vec_g, vec_b, result;
     const int r_out = output_format.r;
     const int b_out = output_format.b;

     /* CYA */
     if (!length)
         return;

     /* one pixel is handled outside of the loop */
     length--;

     /* setup for transforming 1st pixel */
     vec_r = _mm_load_ss(&igtbl_r[src[0]]);
     vec_g = _mm_load_ss(&igtbl_g[src[1]]);
     vec_b = _mm_load_ss(&igtbl_b[src[2]]);
     src += 3;

     /* transform all but final pixel */

     for (i=0; i<length; i++)
     {
         /* position values from gamma tables */
         vec_r = _mm_shuffle_ps(vec_r, vec_r, 0);
         vec_g = _mm_shuffle_ps(vec_g, vec_g, 0);
         vec_b = _mm_shuffle_ps(vec_b, vec_b, 0);

         /* gamma * matrix */
         vec_r = _mm_mul_ps(vec_r, mat0);
         vec_g = _mm_mul_ps(vec_g, mat1);
         vec_b = _mm_mul_ps(vec_b, mat2);

         /* crunch, crunch, crunch */
         vec_r  = _mm_add_ps(vec_g, _mm_add_ps(vec_r, vec_b));
         vec_r  = _mm_max_ps(min, vec_r);
         vec_r  = _mm_min_ps(max, vec_r);
         result = _mm_mul_ps(vec_r, scale);

         /* store calc'd output tables indices */
         _mm_store_si128((__m128i*)output, _mm_cvtps_epi32(result));

         /* load for next loop while store completes */
         vec_r = _mm_load_ss(&igtbl_r[src[0]]);
         vec_g = _mm_load_ss(&igtbl_g[src[1]]);
         vec_b = _mm_load_ss(&igtbl_b[src[2]]);
         src += 3;

         /* use calc'd indices to output RGB values */
         dest[r_out] = otdata_r[output[0]];
         dest[1]     = otdata_g[output[1]];
         dest[b_out] = otdata_b[output[2]];
         dest += 3;
     }

     /* handle final (maybe only) pixel */

     vec_r = _mm_shuffle_ps(vec_r, vec_r, 0);
     vec_g = _mm_shuffle_ps(vec_g, vec_g, 0);
     vec_b = _mm_shuffle_ps(vec_b, vec_b, 0);

     vec_r = _mm_mul_ps(vec_r, mat0);
     vec_g = _mm_mul_ps(vec_g, mat1);
     vec_b = _mm_mul_ps(vec_b, mat2);

     vec_r  = _mm_add_ps(vec_g, _mm_add_ps(vec_r, vec_b));
     vec_r  = _mm_max_ps(min, vec_r);
     vec_r  = _mm_min_ps(max, vec_r);
     result = _mm_mul_ps(vec_r, scale);

     _mm_store_si128((__m128i*)output, _mm_cvtps_epi32(result));

     dest[r_out] = otdata_r[output[0]];
     dest[1]     = otdata_g[output[1]];
     dest[b_out] = otdata_b[output[2]];
 }

 void qcms_transform_data_rgba_out_lut_sse2(qcms_transform *transform,
                                            unsigned char *src,
                                            unsigned char *dest,
                                            size_t length,
                                            qcms_format_type output_format)
 {
     unsigned int i;
     float (*mat)[4] = transform->matrix;
     char input_back[32];
     /* Ensure we have a buffer that's 16 byte aligned regardless of the original
      * stack alignment. We can't use __attribute__((aligned(16))) or __declspec(align(32))
      * because they don't work on stack variables. gcc 4.4 does do the right thing
      * on x86 but that's too new for us right now. For more info: gcc bug #16660 */
     float const * input = (float*)(((uintptr_t)&input_back[16]) & ~0xf);
     /* share input and output locations to save having to keep the
      * locations in separate registers */
     uint32_t const * output = (uint32_t*)input;

     /* deref *transform now to avoid it in loop */
     const float *igtbl_r = transform->input_gamma_table_r;
     const float *igtbl_g = transform->input_gamma_table_g;
     const float *igtbl_b = transform->input_gamma_table_b;

     /* deref *transform now to avoid it in loop */
     const uint8_t *otdata_r = &transform->output_table_r->data[0];
     const uint8_t *otdata_g = &transform->output_table_g->data[0];
     const uint8_t *otdata_b = &transform->output_table_b->data[0];

     /* input matrix values never change */
     const __m128 mat0  = _mm_load_ps(mat[0]);
     const __m128 mat1  = _mm_load_ps(mat[1]);
     const __m128 mat2  = _mm_load_ps(mat[2]);

     /* these values don't change, either */
     const __m128 max   = _mm_load_ps(clampMaxValueX4);
     const __m128 min   = _mm_setzero_ps();
     const __m128 scale = _mm_load_ps(floatScaleX4);

     /* working variables */
     __m128 vec_r, vec_g, vec_b, result;
     const int r_out = output_format.r;
     const int b_out = output_format.b;
     unsigned char alpha;

     /* CYA */
     if (!length)
         return;

     /* one pixel is handled outside of the loop */
     length--;

     /* setup for transforming 1st pixel */
     vec_r = _mm_load_ss(&igtbl_r[src[0]]);
     vec_g = _mm_load_ss(&igtbl_g[src[1]]);
     vec_b = _mm_load_ss(&igtbl_b[src[2]]);
     alpha = src[3];
     src += 4;

     /* transform all but final pixel */

     for (i=0; i<length; i++)
     {
         /* position values from gamma tables */
         vec_r = _mm_shuffle_ps(vec_r, vec_r, 0);
         vec_g = _mm_shuffle_ps(vec_g, vec_g, 0);
         vec_b = _mm_shuffle_ps(vec_b, vec_b, 0);

         /* gamma * matrix */
         vec_r = _mm_mul_ps(vec_r, mat0);
         vec_g = _mm_mul_ps(vec_g, mat1);
         vec_b = _mm_mul_ps(vec_b, mat2);

         /* store alpha for this pixel; load alpha for next */
         dest[3] = alpha;
         alpha   = src[3];

         /* crunch, crunch, crunch */
         vec_r  = _mm_add_ps(vec_g, _mm_add_ps(vec_r, vec_b));
         vec_r  = _mm_max_ps(min, vec_r);
         vec_r  = _mm_min_ps(max, vec_r);
         result = _mm_mul_ps(vec_r, scale);

         /* store calc'd output tables indices */
         _mm_store_si128((__m128i*)output, _mm_cvtps_epi32(result));

         /* load gamma values for next loop while store completes */
         vec_r = _mm_load_ss(&igtbl_r[src[0]]);
         vec_g = _mm_load_ss(&igtbl_g[src[1]]);
         vec_b = _mm_load_ss(&igtbl_b[src[2]]);
         src += 4;

         /* use calc'd indices to output RGB values */
         dest[r_out] = otdata_r[output[0]];
         dest[1]     = otdata_g[output[1]];
         dest[b_out] = otdata_b[output[2]];
         dest += 4;
     }

     /* handle final (maybe only) pixel */

     vec_r = _mm_shuffle_ps(vec_r, vec_r, 0);
     vec_g = _mm_shuffle_ps(vec_g, vec_g, 0);
     vec_b = _mm_shuffle_ps(vec_b, vec_b, 0);

     vec_r = _mm_mul_ps(vec_r, mat0);
     vec_g = _mm_mul_ps(vec_g, mat1);
     vec_b = _mm_mul_ps(vec_b, mat2);

     dest[3] = alpha;

     vec_r  = _mm_add_ps(vec_g, _mm_add_ps(vec_r, vec_b));
     vec_r  = _mm_max_ps(min, vec_r);
     vec_r  = _mm_min_ps(max, vec_r);
     result = _mm_mul_ps(vec_r, scale);

     _mm_store_si128((__m128i*)output, _mm_cvtps_epi32(result));

     dest[r_out] = otdata_r[output[0]];
     dest[1]     = otdata_g[output[1]];
     dest[b_out] = otdata_b[output[2]];
 }

 static inline __m128i __mm_swizzle_epi32(__m128i value, int bgra)
 {
     return bgra ? _mm_shuffle_epi32(value, _MM_SHUFFLE(0, 1, 2, 3)) :
                   _mm_shuffle_epi32(value, _MM_SHUFFLE(0, 3, 2, 1)) ;
 }

 void qcms_transform_data_tetra_clut_rgba_sse2(qcms_transform *transform,
                                               unsigned char *src,
                                               unsigned char *dest,
                                               size_t length,
                                               qcms_format_type output_format)
 {
     const int bgra = output_format.r;

     size_t i;

     const int xy_len_3 = 3 * 1;
     const int x_len_3 = 3 * transform->grid_size;
     const int len_3 = x_len_3 * transform->grid_size;

     const __m128 __255 = _mm_set1_ps(255.0f);
     const __m128 __one = _mm_set1_ps(1.0f);
     const __m128 __000 = _mm_setzero_ps();

     const float* r_table = transform->r_clut;
     const float* g_table = transform->g_clut;
     const float* b_table = transform->b_clut;

     int i3, i2, i1, i0;

     __m128 c3;
     __m128 c2;
     __m128 c1;
     __m128 c0;

     if (!(transform->transform_flags & TRANSFORM_FLAG_CLUT_CACHE))
         qcms_transform_build_clut_cache(transform);

     for (i = 0; i < length; ++i) {
         unsigned char in_r = *src++;
         unsigned char in_g = *src++;
         unsigned char in_b = *src++;

         // initialize the output result with the alpha channel only

         __m128i result = _mm_setr_epi32(*src++, 0, 0, 0);

         // get the input point r.xyz relative to the subcube origin

         float rx = transform->r_cache[in_r];
         float ry = transform->r_cache[in_g];
         float rz = transform->r_cache[in_b];

         // load and LUT scale the subcube maximum vertex

         int xn = transform->ceil_cache[in_r] * len_3;
         int yn = transform->ceil_cache[in_g] * x_len_3;
         int zn = transform->ceil_cache[in_b] * xy_len_3;

         // load and LUT scale the subcube origin vertex

         int x0 = transform->floor_cache[in_r] * len_3;
         int y0 = transform->floor_cache[in_g] * x_len_3;
         int z0 = transform->floor_cache[in_b] * xy_len_3;

         // tetrahedral interpolate the input color r.xyz

 #define TETRA_LOOKUP_CLUT(i3, i2, i1, i0) \
         c0 = _mm_set_ps(b_table[i0], g_table[i0], r_table[i0], 0.f), \
         c1 = _mm_set_ps(b_table[i1], g_table[i1], r_table[i1], 0.f), \
         c2 = _mm_set_ps(b_table[i2], g_table[i2], r_table[i2], 0.f), \
         c3 = _mm_set_ps(b_table[i3], g_table[i3], r_table[i3], 0.f)

         i0 = x0 + y0 + z0;

         if (rx >= ry) {

             if (ry >= rz) {         // rx >= ry && ry >= rz

                 i3 = yn + (i1 = xn);
                 i1 += i0 - x0;
                 i2 = i3 + z0;
                 i3 += zn;

                 TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

                 c3 = _mm_sub_ps(c3, c2);
                 c2 = _mm_sub_ps(c2, c1);
                 c1 = _mm_sub_ps(c1, c0);

             } else if (rx >= rz) {  // rx >= rz && rz >= ry

                 i3 = zn + (i1 = xn);
                 i1 += i0 - x0;
                 i2 = i3 + yn;
                 i3 += y0;

                 TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

                 c2 = _mm_sub_ps(c2, c3);
                 c3 = _mm_sub_ps(c3, c1);
                 c1 = _mm_sub_ps(c1, c0);

             } else {                // rz > rx && rx >= ry

                 i2 = xn + (i3 = zn);
                 i3 += i0 - z0;
                 i1 = i2 + y0;
                 i2 += yn;

                 TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

                 c2 = _mm_sub_ps(c2, c1);
                 c1 = _mm_sub_ps(c1, c3);
                 c3 = _mm_sub_ps(c3, c0);
             }
         } else {

             if (rx >= rz) {         // ry > rx && rx >= rz

                 i3 = xn + (i2 = yn);
                 i2 += i0 - y0;
                 i1 = i3 + z0;
                 i3 += zn;

                 TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

                 c3 = _mm_sub_ps(c3, c1);
                 c1 = _mm_sub_ps(c1, c2);
                 c2 = _mm_sub_ps(c2, c0);

             } else if (ry >= rz) {  // ry >= rz && rz > rx

                 i3 = zn + (i2 = yn);
                 i2 += i0 - y0;
                 i1 = i3 + xn;
                 i3 += x0;

                 TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

                 c1 = _mm_sub_ps(c1, c3);
                 c3 = _mm_sub_ps(c3, c2);
                 c2 = _mm_sub_ps(c2, c0);

             } else {                // rz > ry && ry > rx

                 i2 = yn + (i3 = zn);
                 i3 += i0 - z0;
                 i1 = i2 + xn;
                 i2 += x0;

                 TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

                 c1 = _mm_sub_ps(c1, c2);
                 c2 = _mm_sub_ps(c2, c3);
                 c3 = _mm_sub_ps(c3, c0);
             }
         }

         // output.xyz = column_matrix(c1, c2, c3) x r.xyz + c0.xyz

         c0 = _mm_add_ps(c0, _mm_mul_ps(c1, _mm_set1_ps(rx)));
         c0 = _mm_add_ps(c0, _mm_mul_ps(c2, _mm_set1_ps(ry)));
         c0 = _mm_add_ps(c0, _mm_mul_ps(c3, _mm_set1_ps(rz)));

         // clamp to [0.0..1.0], then scale by 255

         c0 = _mm_max_ps(c0, __000);
         c0 = _mm_min_ps(c0, __one);
         c0 = _mm_mul_ps(c0, __255);

         // int(c0) with float rounding, add alpha

         result = _mm_add_epi32(result, _mm_cvtps_epi32(c0));

         // swizzle and repack in result low bytes

         result = __mm_swizzle_epi32(result, bgra);
         result = _mm_packus_epi16(result, result);
         result = _mm_packus_epi16(result, result);

         // store into uint32_t* pixel destination

         *(uint32_t *)dest = _mm_cvtsi128_si32(result);
         dest += 4;
     }
 }
	// qcms
	// Copyright (C) 2009 Mozilla Foundation
	// Copyright (C) 2015 Intel Corporation
	//
	// Permission is hereby granted, free of charge, to any person obtaining
	// a copy of this software and associated documentation files (the "Software"),
	// to deal in the Software without restriction, including without limitation
	// the rights to use, copy, modify, merge, publish, distribute, sublicense,
	// and/or sell copies of the Software, and to permit persons to whom the Software
	// is furnished to do so, subject to the following conditions:
	//
	// The above copyright notice and this permission notice shall be included in
	// all copies or substantial portions of the Software.
	//
	// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
	// EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO
	// THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
	// NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
	// LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
	// OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
	// WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

	#include <emmintrin.h>

	#include "qcmsint.h"

	/* pre-shuffled: just load these into XMM reg instead of load-scalar/shufps sequence */
	#define FLOATSCALE (float)(PRECACHE_OUTPUT_SIZE - 1)
	#define CLAMPMAXVAL 1.0f

	static const ALIGN float floatScaleX4[4] =
	{ FLOATSCALE, FLOATSCALE, FLOATSCALE, FLOATSCALE};
	static const ALIGN float clampMaxValueX4[4] =
	{ CLAMPMAXVAL, CLAMPMAXVAL, CLAMPMAXVAL, CLAMPMAXVAL};

	void qcms_transform_data_rgb_out_lut_sse2(qcms_transform *transform,
	unsigned char *src,
	unsigned char *dest,
	size_t length,
	qcms_format_type output_format)
	{
	unsigned int i;
	float (*mat)[4] = transform->matrix;
	char input_back[32];
	/* Ensure we have a buffer that's 16 byte aligned regardless of the original
	* stack alignment. We can't use __attribute__((aligned(16))) or __declspec(align(32))
	* because they don't work on stack variables. gcc 4.4 does do the right thing
	* on x86 but that's too new for us right now. For more info: gcc bug #16660 */
	float const * input = (float*)(((uintptr_t)&input_back[16]) & ~0xf);
	/* share input and output locations to save having to keep the
	* locations in separate registers */
	uint32_t const * output = (uint32_t*)input;

	/* deref transform now to avoid it in loop /
	const float *igtbl_r = transform->input_gamma_table_r;
	const float *igtbl_g = transform->input_gamma_table_g;
	const float *igtbl_b = transform->input_gamma_table_b;

	/* deref transform now to avoid it in loop /
	const uint8_t *otdata_r = &transform->output_table_r->data[0];
	const uint8_t *otdata_g = &transform->output_table_g->data[0];
	const uint8_t *otdata_b = &transform->output_table_b->data[0];

	/* input matrix values never change */
	const __m128 mat0 = _mm_load_ps(mat[0]);
	const __m128 mat1 = _mm_load_ps(mat[1]);
	const __m128 mat2 = _mm_load_ps(mat[2]);

	/* these values don't change, either */
	const __m128 max = _mm_load_ps(clampMaxValueX4);
	const __m128 min = _mm_setzero_ps();
	const __m128 scale = _mm_load_ps(floatScaleX4);

	/* working variables */
	__m128 vec_r, vec_g, vec_b, result;
	const int r_out = output_format.r;
	const int b_out = output_format.b;

	/* CYA */
	if (!length)
	return;

	/* one pixel is handled outside of the loop */
	length--;

	/* setup for transforming 1st pixel */
	vec_r = _mm_load_ss(&igtbl_r[src[0]]);
	vec_g = _mm_load_ss(&igtbl_g[src[1]]);
	vec_b = _mm_load_ss(&igtbl_b[src[2]]);
	src += 3;

	/* transform all but final pixel */

	for (i=0; i<length; i++)
	{
	/* position values from gamma tables */
	vec_r = _mm_shuffle_ps(vec_r, vec_r, 0);
	vec_g = _mm_shuffle_ps(vec_g, vec_g, 0);
	vec_b = _mm_shuffle_ps(vec_b, vec_b, 0);

	/* gamma * matrix */
	vec_r = _mm_mul_ps(vec_r, mat0);
	vec_g = _mm_mul_ps(vec_g, mat1);
	vec_b = _mm_mul_ps(vec_b, mat2);

	/* crunch, crunch, crunch */
	vec_r = _mm_add_ps(vec_g, _mm_add_ps(vec_r, vec_b));
	vec_r = _mm_max_ps(min, vec_r);
	vec_r = _mm_min_ps(max, vec_r);
	result = _mm_mul_ps(vec_r, scale);

	/* store calc'd output tables indices */
	_mm_store_si128((__m128i*)output, _mm_cvtps_epi32(result));

	/* load for next loop while store completes */
	vec_r = _mm_load_ss(&igtbl_r[src[0]]);
	vec_g = _mm_load_ss(&igtbl_g[src[1]]);
	vec_b = _mm_load_ss(&igtbl_b[src[2]]);
	src += 3;

	/* use calc'd indices to output RGB values */
	dest[r_out] = otdata_r[output[0]];
	dest[1] = otdata_g[output[1]];
	dest[b_out] = otdata_b[output[2]];
	dest += 3;
	}

	/* handle final (maybe only) pixel */

	vec_r = _mm_shuffle_ps(vec_r, vec_r, 0);
	vec_g = _mm_shuffle_ps(vec_g, vec_g, 0);
	vec_b = _mm_shuffle_ps(vec_b, vec_b, 0);

	vec_r = _mm_mul_ps(vec_r, mat0);
	vec_g = _mm_mul_ps(vec_g, mat1);
	vec_b = _mm_mul_ps(vec_b, mat2);

	vec_r = _mm_add_ps(vec_g, _mm_add_ps(vec_r, vec_b));
	vec_r = _mm_max_ps(min, vec_r);
	vec_r = _mm_min_ps(max, vec_r);
	result = _mm_mul_ps(vec_r, scale);

	_mm_store_si128((__m128i*)output, _mm_cvtps_epi32(result));

	dest[r_out] = otdata_r[output[0]];
	dest[1] = otdata_g[output[1]];
	dest[b_out] = otdata_b[output[2]];
	}

	void qcms_transform_data_rgba_out_lut_sse2(qcms_transform *transform,
	unsigned char *src,
	unsigned char *dest,
	size_t length,
	qcms_format_type output_format)
	{
	unsigned int i;
	float (*mat)[4] = transform->matrix;
	char input_back[32];
	/* Ensure we have a buffer that's 16 byte aligned regardless of the original
	* stack alignment. We can't use __attribute__((aligned(16))) or __declspec(align(32))
	* because they don't work on stack variables. gcc 4.4 does do the right thing
	* on x86 but that's too new for us right now. For more info: gcc bug #16660 */
	float const * input = (float*)(((uintptr_t)&input_back[16]) & ~0xf);
	/* share input and output locations to save having to keep the
	* locations in separate registers */
	uint32_t const * output = (uint32_t*)input;

	/* deref transform now to avoid it in loop /
	const float *igtbl_r = transform->input_gamma_table_r;
	const float *igtbl_g = transform->input_gamma_table_g;
	const float *igtbl_b = transform->input_gamma_table_b;

	/* deref transform now to avoid it in loop /
	const uint8_t *otdata_r = &transform->output_table_r->data[0];
	const uint8_t *otdata_g = &transform->output_table_g->data[0];
	const uint8_t *otdata_b = &transform->output_table_b->data[0];

	/* input matrix values never change */
	const __m128 mat0 = _mm_load_ps(mat[0]);
	const __m128 mat1 = _mm_load_ps(mat[1]);
	const __m128 mat2 = _mm_load_ps(mat[2]);

	/* these values don't change, either */
	const __m128 max = _mm_load_ps(clampMaxValueX4);
	const __m128 min = _mm_setzero_ps();
	const __m128 scale = _mm_load_ps(floatScaleX4);

	/* working variables */
	__m128 vec_r, vec_g, vec_b, result;
	const int r_out = output_format.r;
	const int b_out = output_format.b;
	unsigned char alpha;

	/* CYA */
	if (!length)
	return;

	/* one pixel is handled outside of the loop */
	length--;

	/* setup for transforming 1st pixel */
	vec_r = _mm_load_ss(&igtbl_r[src[0]]);
	vec_g = _mm_load_ss(&igtbl_g[src[1]]);
	vec_b = _mm_load_ss(&igtbl_b[src[2]]);
	alpha = src[3];
	src += 4;

	/* transform all but final pixel */

	for (i=0; i<length; i++)
	{
	/* position values from gamma tables */
	vec_r = _mm_shuffle_ps(vec_r, vec_r, 0);
	vec_g = _mm_shuffle_ps(vec_g, vec_g, 0);
	vec_b = _mm_shuffle_ps(vec_b, vec_b, 0);

	/* gamma * matrix */
	vec_r = _mm_mul_ps(vec_r, mat0);
	vec_g = _mm_mul_ps(vec_g, mat1);
	vec_b = _mm_mul_ps(vec_b, mat2);

	/* store alpha for this pixel; load alpha for next */
	dest[3] = alpha;
	alpha = src[3];

	/* crunch, crunch, crunch */
	vec_r = _mm_add_ps(vec_g, _mm_add_ps(vec_r, vec_b));
	vec_r = _mm_max_ps(min, vec_r);
	vec_r = _mm_min_ps(max, vec_r);
	result = _mm_mul_ps(vec_r, scale);

	/* store calc'd output tables indices */
	_mm_store_si128((__m128i*)output, _mm_cvtps_epi32(result));

	/* load gamma values for next loop while store completes */
	vec_r = _mm_load_ss(&igtbl_r[src[0]]);
	vec_g = _mm_load_ss(&igtbl_g[src[1]]);
	vec_b = _mm_load_ss(&igtbl_b[src[2]]);
	src += 4;

	/* use calc'd indices to output RGB values */
	dest[r_out] = otdata_r[output[0]];
	dest[1] = otdata_g[output[1]];
	dest[b_out] = otdata_b[output[2]];
	dest += 4;
	}

	/* handle final (maybe only) pixel */

	vec_r = _mm_shuffle_ps(vec_r, vec_r, 0);
	vec_g = _mm_shuffle_ps(vec_g, vec_g, 0);
	vec_b = _mm_shuffle_ps(vec_b, vec_b, 0);

	vec_r = _mm_mul_ps(vec_r, mat0);
	vec_g = _mm_mul_ps(vec_g, mat1);
	vec_b = _mm_mul_ps(vec_b, mat2);

	dest[3] = alpha;

	vec_r = _mm_add_ps(vec_g, _mm_add_ps(vec_r, vec_b));
	vec_r = _mm_max_ps(min, vec_r);
	vec_r = _mm_min_ps(max, vec_r);
	result = _mm_mul_ps(vec_r, scale);

	_mm_store_si128((__m128i*)output, _mm_cvtps_epi32(result));

	dest[r_out] = otdata_r[output[0]];
	dest[1] = otdata_g[output[1]];
	dest[b_out] = otdata_b[output[2]];
	}

	static inline __m128i __mm_swizzle_epi32(__m128i value, int bgra)
	{
	return bgra ? _mm_shuffle_epi32(value, _MM_SHUFFLE(0, 1, 2, 3)) :
	_mm_shuffle_epi32(value, _MM_SHUFFLE(0, 3, 2, 1)) ;
	}

	void qcms_transform_data_tetra_clut_rgba_sse2(qcms_transform *transform,
	unsigned char *src,
	unsigned char *dest,
	size_t length,
	qcms_format_type output_format)
	{
	const int bgra = output_format.r;

	size_t i;

	const int xy_len_3 = 3 * 1;
	const int x_len_3 = 3 * transform->grid_size;
	const int len_3 = x_len_3 * transform->grid_size;

	const __m128 __255 = _mm_set1_ps(255.0f);
	const __m128 __one = _mm_set1_ps(1.0f);
	const __m128 __000 = _mm_setzero_ps();

	const float* r_table = transform->r_clut;
	const float* g_table = transform->g_clut;
	const float* b_table = transform->b_clut;

	int i3, i2, i1, i0;

	__m128 c3;
	__m128 c2;
	__m128 c1;
	__m128 c0;

	if (!(transform->transform_flags & TRANSFORM_FLAG_CLUT_CACHE))
	qcms_transform_build_clut_cache(transform);

	for (i = 0; i < length; ++i) {
	unsigned char in_r = *src++;
	unsigned char in_g = *src++;
	unsigned char in_b = *src++;

	// initialize the output result with the alpha channel only

	__m128i result = _mm_setr_epi32(*src++, 0, 0, 0);

	// get the input point r.xyz relative to the subcube origin

	float rx = transform->r_cache[in_r];
	float ry = transform->r_cache[in_g];
	float rz = transform->r_cache[in_b];

	// load and LUT scale the subcube maximum vertex

	int xn = transform->ceil_cache[in_r] * len_3;
	int yn = transform->ceil_cache[in_g] * x_len_3;
	int zn = transform->ceil_cache[in_b] * xy_len_3;

	// load and LUT scale the subcube origin vertex

	int x0 = transform->floor_cache[in_r] * len_3;
	int y0 = transform->floor_cache[in_g] * x_len_3;
	int z0 = transform->floor_cache[in_b] * xy_len_3;

	// tetrahedral interpolate the input color r.xyz

	#define TETRA_LOOKUP_CLUT(i3, i2, i1, i0) \
	c0 = _mm_set_ps(b_table[i0], g_table[i0], r_table[i0], 0.f), \
	c1 = _mm_set_ps(b_table[i1], g_table[i1], r_table[i1], 0.f), \
	c2 = _mm_set_ps(b_table[i2], g_table[i2], r_table[i2], 0.f), \
	c3 = _mm_set_ps(b_table[i3], g_table[i3], r_table[i3], 0.f)

	i0 = x0 + y0 + z0;

	if (rx >= ry) {

	if (ry >= rz) { // rx >= ry && ry >= rz

	i3 = yn + (i1 = xn);
	i1 += i0 - x0;
	i2 = i3 + z0;
	i3 += zn;

	TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

	c3 = _mm_sub_ps(c3, c2);
	c2 = _mm_sub_ps(c2, c1);
	c1 = _mm_sub_ps(c1, c0);

	} else if (rx >= rz) { // rx >= rz && rz >= ry

	i3 = zn + (i1 = xn);
	i1 += i0 - x0;
	i2 = i3 + yn;
	i3 += y0;

	TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

	c2 = _mm_sub_ps(c2, c3);
	c3 = _mm_sub_ps(c3, c1);
	c1 = _mm_sub_ps(c1, c0);

	} else { // rz > rx && rx >= ry

	i2 = xn + (i3 = zn);
	i3 += i0 - z0;
	i1 = i2 + y0;
	i2 += yn;

	TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

	c2 = _mm_sub_ps(c2, c1);
	c1 = _mm_sub_ps(c1, c3);
	c3 = _mm_sub_ps(c3, c0);
	}
	} else {

	if (rx >= rz) { // ry > rx && rx >= rz

	i3 = xn + (i2 = yn);
	i2 += i0 - y0;
	i1 = i3 + z0;
	i3 += zn;

	TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

	c3 = _mm_sub_ps(c3, c1);
	c1 = _mm_sub_ps(c1, c2);
	c2 = _mm_sub_ps(c2, c0);

	} else if (ry >= rz) { // ry >= rz && rz > rx

	i3 = zn + (i2 = yn);
	i2 += i0 - y0;
	i1 = i3 + xn;
	i3 += x0;

	TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

	c1 = _mm_sub_ps(c1, c3);
	c3 = _mm_sub_ps(c3, c2);
	c2 = _mm_sub_ps(c2, c0);

	} else { // rz > ry && ry > rx

	i2 = yn + (i3 = zn);
	i3 += i0 - z0;
	i1 = i2 + xn;
	i2 += x0;

	TETRA_LOOKUP_CLUT(i3, i2, i1, i0);

	c1 = _mm_sub_ps(c1, c2);
	c2 = _mm_sub_ps(c2, c3);
	c3 = _mm_sub_ps(c3, c0);
	}
	}

	// output.xyz = column_matrix(c1, c2, c3) x r.xyz + c0.xyz

	c0 = _mm_add_ps(c0, _mm_mul_ps(c1, _mm_set1_ps(rx)));
	c0 = _mm_add_ps(c0, _mm_mul_ps(c2, _mm_set1_ps(ry)));
	c0 = _mm_add_ps(c0, _mm_mul_ps(c3, _mm_set1_ps(rz)));

	// clamp to [0.0..1.0], then scale by 255

	c0 = _mm_max_ps(c0, __000);
	c0 = _mm_min_ps(c0, __one);
	c0 = _mm_mul_ps(c0, __255);

	// int(c0) with float rounding, add alpha

	result = _mm_add_epi32(result, _mm_cvtps_epi32(c0));

	// swizzle and repack in result low bytes

	result = __mm_swizzle_epi32(result, bgra);
	result = _mm_packus_epi16(result, result);
	result = _mm_packus_epi16(result, result);

	// store into uint32_t* pixel destination

	(uint32_t )dest = _mm_cvtsi128_si32(result);
	dest += 4;
	}
	}