src/panfrost/util/pan_lower_framebuffer.c

   1 /*
   2  * Copyright (C) 2020 Collabora, Ltd.
   3  *
   4  * Permission is hereby granted, free of charge, to any person obtaining a
   5  * copy of this software and associated documentation files (the "Software"),
   6  * to deal in the Software without restriction, including without limitation
   7  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
   8  * and/or sell copies of the Software, and to permit persons to whom the
   9  * Software is furnished to do so, subject to the following conditions:
  10  *
  11  * The above copyright notice and this permission notice (including the next
  12  * paragraph) shall be included in all copies or substantial portions of the
  13  * Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
  18  * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
  19  * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
  20  * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS
  21  * IN THE SOFTWARE.
  22  *
  23  * Authors (Collabora):
  24  *      Alyssa Rosenzweig <alyssa.rosenzweig@collabora.com>
  25  */
  26
  27 /**
  28  * Implements framebuffer format conversions in software for Midgard/Bifrost
  29  * blend shaders. This pass is designed for a single render target; Midgard
  30  * duplicates blend shaders for MRT to simplify everything. A particular
  31  * framebuffer format may be categorized as 1) typed load available, 2) typed
  32  * unpack available, or 3) software unpack only, and likewise for stores. The
  33  * first two types are handled in the compiler backend directly, so this module
  34  * is responsible for identifying type 3 formats (hardware dependent) and
  35  * inserting appropriate ALU code to perform the conversion from the packed
  36  * type to a designated unpacked type, and vice versa.
  37  *
  38  * The unpacked type depends on the format:
  39  *
  40  *      - For 32-bit float formats, 32-bit floats.
  41  *      - For other floats, 16-bit floats.
  42  *      - For 32-bit ints, 32-bit ints.
  43  *      - For 8-bit ints, 8-bit ints.
  44  *      - For other ints, 16-bit ints.
  45  *
  46  * The rationale is to optimize blending and logic op instructions by using the
  47  * smallest precision necessary to store the pixel losslessly.
  48  */
  49
  50 #include "compiler/nir/nir.h"
  51 #include "compiler/nir/nir_builder.h"
  52 #include "compiler/nir/nir_format_convert.h"
  53 #include "util/format/u_format.h"
  54 #include "pan_lower_framebuffer.h"
  55
  56 /* Determines the unpacked type best suiting a given format, so the rest of the
  57  * pipeline may be adjusted accordingly */
  58
  59 nir_alu_type
  60 pan_unpacked_type_for_format(const struct util_format_description *desc)
  61 {
  62         int c = util_format_get_first_non_void_channel(desc->format);
  63
  64         if (c == -1)
  65                 unreachable("Void format not renderable");
  66
  67         bool large = (desc->channel[c].size > 16);
  68         bool bit8 = (desc->channel[c].size == 8);
  69         assert(desc->channel[c].size <= 32);
  70
  71         if (desc->channel[c].normalized)
  72                 return large ? nir_type_float32 : nir_type_float16;
  73
  74         switch (desc->channel[c].type) {
  75         case UTIL_FORMAT_TYPE_UNSIGNED:
  76                 return bit8 ? nir_type_uint8 :
  77                         large ? nir_type_uint32 : nir_type_uint16;
  78         case UTIL_FORMAT_TYPE_SIGNED:
  79                 return bit8 ? nir_type_int8 :
  80                         large ? nir_type_int32 : nir_type_int16;
  81         case UTIL_FORMAT_TYPE_FLOAT:
  82                 return large ? nir_type_float32 : nir_type_float16;
  83         default:
  84                 unreachable("Format not renderable");
  85         }
  86 }