mesa/src/util/format_r11g11b10f.h

/*
 * Copyright (C) 2011 Marek Olšák <maraeo@gmail.com>
 *
 * 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 (including the next
 * paragraph) 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.
 */

/* Based on code from The OpenGL Programming Guide / 7th Edition, Appendix J.
 * Available here: http://www.opengl-redbook.com/appendices/
 * The algorithm in the book contains a bug though, which is fixed in the code
 * below.
 */

#ifndef FORMAT_R11G11B10F_H
#define FORMAT_R11G11B10F_H

#include <stdint.h>

#include "rounding.h"

#define UF11(e, m)           ((e << 6) | (m))
#define UF11_EXPONENT_BIAS   15
#define UF11_EXPONENT_BITS   0x1F
#define UF11_EXPONENT_SHIFT  6
#define UF11_MANTISSA_BITS   0x3F
#define UF11_MANTISSA_SHIFT  (23 - UF11_EXPONENT_SHIFT)
#define UF11_MAX_EXPONENT    (UF11_EXPONENT_BITS << UF11_EXPONENT_SHIFT)

#define UF10(e, m)           ((e << 5) | (m))
#define UF10_EXPONENT_BIAS   15
#define UF10_EXPONENT_BITS   0x1F
#define UF10_EXPONENT_SHIFT  5
#define UF10_MANTISSA_BITS   0x1F
#define UF10_MANTISSA_SHIFT  (23 - UF10_EXPONENT_SHIFT)
#define UF10_MAX_EXPONENT    (UF10_EXPONENT_BITS << UF10_EXPONENT_SHIFT)

#define F32_INFINITY         0x7f800000

static inline uint32_t f32_to_uf11(float val)
{
   union {
      float f;
      uint32_t ui;
   } f32 = {val};

   uint16_t uf11 = 0;

   /* Decode little-endian 32-bit floating-point value */
   int sign = (f32.ui >> 16) & 0x8000;
   /* Map exponent to the range [-127,128] */
   int exponent = ((f32.ui >> 23) & 0xff) - 127;
   int mantissa = f32.ui & 0x007fffff;

   if (exponent == 128) { /* Infinity or NaN */
      /* From the GL_EXT_packed_float spec:
       *
       *     "Additionally: negative infinity is converted to zero; positive
       *      infinity is converted to positive infinity; and both positive and
       *      negative NaN are converted to positive NaN."
       */
      uf11 = UF11_MAX_EXPONENT;
      if (mantissa) {
         uf11 |= 1; /* NaN */
      } else {
         if (sign)
            uf11 = 0; /* 0.0 */
      }
   } else if (sign) {
      return 0;
   } else if (val > 65024.0f) {
      /* From the GL_EXT_packed_float spec:
       *
       *     "Likewise, finite positive values greater than 65024 (the maximum
       *      finite representable unsigned 11-bit floating-point value) are
       *      converted to 65024."
       */
      uf11 = UF11(30, 63);
   } else if (exponent > -15) { /* Normal value */
      /* Dividing by 2^exponent gives us a number in the range [1, 2).
       * Multiplying by 2^6=64 gives us our mantissa, plus an extra 1 which
       * we'll mask off.
       */
      mantissa = _mesa_lroundevenf(ldexp(val, 6 - exponent));
      if (mantissa >= 2 << UF11_EXPONENT_SHIFT) {
         /* The float32 was rounded upwards into the range of the next
          * exponent, so bump the exponent.
          */
         assert(mantissa == 2 << UF11_EXPONENT_SHIFT);
         mantissa >>= 1;
         exponent++;
      }
      assert((mantissa >> UF11_EXPONENT_SHIFT) == 1);
      mantissa &= UF11_MANTISSA_BITS;
      exponent += UF11_EXPONENT_BIAS;
      uf11 = UF11(exponent, mantissa);
   } else { /* Zero or denormal */
      /* Since exponent <= -15, Multiplying by 2^14 gives us a number in the
       * range [0, 1). Multiplying by 2^6=64 gives us our mantissa.
       */
      mantissa = _mesa_lroundevenf(ldexp(val, 6 + 14));

      /* It's possible that we get a normal after rounding */
      if ((mantissa >> UF11_EXPONENT_SHIFT) != 0) {
         assert(mantissa == (1 << UF11_EXPONENT_SHIFT));
         uf11 = UF11(1, 0);
      } else {
         uf11 = UF11(0, mantissa);
      }
   }

   return uf11;
}

static inline float uf11_to_f32(uint16_t val)
{
   union {
      float f;
      uint32_t ui;
   } f32;

   int exponent = (val & 0x07c0) >> UF11_EXPONENT_SHIFT;
   int mantissa = (val & 0x003f);

   f32.f = 0.0;

   if (exponent == 0) {
      if (mantissa != 0) {
         const float scale = 1.0 / (1 << 20);
         f32.f = scale * mantissa;
      }
   } else if (exponent == 31) {
      f32.ui = F32_INFINITY | mantissa;
   } else {
      float scale, decimal;
      exponent -= 15;
      if (exponent < 0) {
         scale = 1.0f / (1 << -exponent);
      } else {
         scale = (float) (1 << exponent);
      }
      decimal = 1.0f + (float) mantissa / 64;
      f32.f = scale * decimal;
   }

   return f32.f;
}

static inline uint32_t f32_to_uf10(float val)
{
   union {
      float f;
      uint32_t ui;
   } f32 = {val};

   uint16_t uf10 = 0;

   /* Decode little-endian 32-bit floating-point value */
   int sign = (f32.ui >> 16) & 0x8000;
   /* Map exponent to the range [-127,128] */
   int exponent = ((f32.ui >> 23) & 0xff) - 127;
   int mantissa = f32.ui & 0x007fffff;

   if (exponent == 128) {
      /* From the GL_EXT_packed_float spec:
       *
       *     "Additionally: negative infinity is converted to zero; positive
       *      infinity is converted to positive infinity; and both positive and
       *      negative NaN are converted to positive NaN."
       */
      uf10 = UF10_MAX_EXPONENT;
      if (mantissa) {
         uf10 |= 1; /* NaN */
      } else {
         if (sign)
            uf10 = 0; /* 0.0 */
      }
   } else if (sign) {
      return 0;
   } else if (val > 64512.0f) {
      /* From the GL_EXT_packed_float spec:
       *
       *     "Likewise, finite positive values greater than 64512 (the maximum
       *      finite representable unsigned 10-bit floating-point value) are
       *      converted to 64512."
       */
      uf10 = UF10(30, 31);
   } else if (exponent > -15) { /* Normal value */
      /* Dividing by 2^exponent gives us a number in the range [1, 2).
       * Multiplying by 2^5=32 gives us our mantissa, plus an extra 1 which
       * we'll mask off.
       */
      mantissa = _mesa_lroundevenf(ldexp(val, 5 - exponent));
      if (mantissa >= 2 << UF10_EXPONENT_SHIFT) {
         /* The float32 was rounded upwards into the range of the next
          * exponent, so bump the exponent.
          */
         assert(mantissa == 2 << UF10_EXPONENT_SHIFT);
         mantissa >>= 1;
         exponent++;
      }
      assert((mantissa >> UF10_EXPONENT_SHIFT) == 1);
      mantissa &= UF10_MANTISSA_BITS;
      exponent += UF10_EXPONENT_BIAS;
      uf10 = UF10(exponent, mantissa);
   } else { /* Zero or denormal */
      /* Since exponent <= -15, Multiplying by 2^14 gives us a number in the
       * range [0, 1). Multiplying by 2^5=32 gives us our mantissa.
       */
      mantissa = _mesa_lroundevenf(ldexp(val, 5 + 14));

      /* It's possible that we get a normal after rounding */
      if ((mantissa >> UF10_EXPONENT_SHIFT) != 0) {
         assert(mantissa == (1 << UF10_EXPONENT_SHIFT));
         uf10 = UF10(1, 0);
      } else {
         uf10 = UF10(0, mantissa);
      }
   }

   return uf10;
}

static inline float uf10_to_f32(uint16_t val)
{
   union {
      float f;
      uint32_t ui;
   } f32;

   int exponent = (val & 0x03e0) >> UF10_EXPONENT_SHIFT;
   int mantissa = (val & 0x001f);

   f32.f = 0.0;

   if (exponent == 0) {
      if (mantissa != 0) {
         const float scale = 1.0 / (1 << 19);
         f32.f = scale * mantissa;
      }
   } else if (exponent == 31) {
      f32.ui = F32_INFINITY | mantissa;
   } else {
      float scale, decimal;
      exponent -= 15;
      if (exponent < 0) {
         scale = 1.0f / (1 << -exponent);
      }
      else {
         scale = (float) (1 << exponent);
      }
      decimal = 1.0f + (float) mantissa / 32;
      f32.f = scale * decimal;
   }

   return f32.f;
}

static inline uint32_t float3_to_r11g11b10f(const float rgb[3])
{
   return ( f32_to_uf11(rgb[0]) & 0x7ff) |
          ((f32_to_uf11(rgb[1]) & 0x7ff) << 11) |
          ((f32_to_uf10(rgb[2]) & 0x3ff) << 22);
}

static inline void r11g11b10f_to_float3(uint32_t rgb, float retval[3])
{
   retval[0] = uf11_to_f32( rgb        & 0x7ff);
   retval[1] = uf11_to_f32((rgb >> 11) & 0x7ff);
   retval[2] = uf10_to_f32((rgb >> 22) & 0x3ff);
}

#endif /* FORMAT_R11G11B10F_H */