| 1 | /* |
| 2 | Simple DirectMedia Layer |
| 3 | Copyright (C) 1997-2025 Sam Lantinga <slouken@libsdl.org> |
| 4 | |
| 5 | This software is provided 'as-is', without any express or implied |
| 6 | warranty. In no event will the authors be held liable for any damages |
| 7 | arising from the use of this software. |
| 8 | |
| 9 | Permission is granted to anyone to use this software for any purpose, |
| 10 | including commercial applications, and to alter it and redistribute it |
| 11 | freely, subject to the following restrictions: |
| 12 | |
| 13 | 1. The origin of this software must not be misrepresented; you must not |
| 14 | claim that you wrote the original software. If you use this software |
| 15 | in a product, an acknowledgment in the product documentation would be |
| 16 | appreciated but is not required. |
| 17 | 2. Altered source versions must be plainly marked as such, and must not be |
| 18 | misrepresented as being the original software. |
| 19 | 3. This notice may not be removed or altered from any source distribution. |
| 20 | */ |
| 21 | #include "SDL_internal.h" |
| 22 | |
| 23 | #ifdef SDL_VIDEO_RENDER_GPU |
| 24 | |
| 25 | #include "SDL_gpu_util.h" |
| 26 | #include "SDL_pipeline_gpu.h" |
| 27 | |
| 28 | #include "../SDL_sysrender.h" |
| 29 | |
| 30 | static Uint32 SDLCALL HashPipelineCacheKey(void *userdata, const void *key) |
| 31 | { |
| 32 | const GPU_PipelineParameters *params = (const GPU_PipelineParameters *) key; |
| 33 | return SDL_murmur3_32(params, sizeof(*params), 0); |
| 34 | } |
| 35 | |
| 36 | static bool SDLCALL MatchPipelineCacheKey(void *userdata, const void *a, const void *b) |
| 37 | { |
| 38 | return (SDL_memcmp(a, b, sizeof (GPU_PipelineParameters)) == 0); |
| 39 | } |
| 40 | |
| 41 | static void SDLCALL DestroyPipelineCacheHashItem(void *userdata, const void *key, const void *value) |
| 42 | { |
| 43 | SDL_GPUGraphicsPipeline *pipeline = (SDL_GPUGraphicsPipeline *) value; |
| 44 | SDL_GPUDevice *device = (SDL_GPUDevice *) userdata; |
| 45 | SDL_ReleaseGPUGraphicsPipeline(device, pipeline); |
| 46 | SDL_free((GPU_PipelineParameters *) key); |
| 47 | } |
| 48 | |
| 49 | bool GPU_InitPipelineCache(GPU_PipelineCache *cache, SDL_GPUDevice *device) |
| 50 | { |
| 51 | cache->table = SDL_CreateHashTable(0, false, HashPipelineCacheKey, MatchPipelineCacheKey, DestroyPipelineCacheHashItem, device); |
| 52 | return (cache->table != NULL); |
| 53 | } |
| 54 | |
| 55 | void GPU_DestroyPipelineCache(GPU_PipelineCache *cache) |
| 56 | { |
| 57 | SDL_DestroyHashTable(cache->table); |
| 58 | } |
| 59 | |
| 60 | static SDL_GPUGraphicsPipeline *MakePipeline(SDL_GPUDevice *device, GPU_Shaders *shaders, const GPU_PipelineParameters *params) |
| 61 | { |
| 62 | SDL_GPUColorTargetDescription ad; |
| 63 | SDL_zero(ad); |
| 64 | ad.format = params->attachment_format; |
| 65 | |
| 66 | SDL_BlendMode blend = params->blend_mode; |
| 67 | ad.blend_state.enable_blend = blend != 0; |
| 68 | ad.blend_state.color_write_mask = 0xF; |
| 69 | ad.blend_state.alpha_blend_op = GPU_ConvertBlendOperation(SDL_GetBlendModeAlphaOperation(blend)); |
| 70 | ad.blend_state.dst_alpha_blendfactor = GPU_ConvertBlendFactor(SDL_GetBlendModeDstAlphaFactor(blend)); |
| 71 | ad.blend_state.src_alpha_blendfactor = GPU_ConvertBlendFactor(SDL_GetBlendModeSrcAlphaFactor(blend)); |
| 72 | ad.blend_state.color_blend_op = GPU_ConvertBlendOperation(SDL_GetBlendModeColorOperation(blend)); |
| 73 | ad.blend_state.dst_color_blendfactor = GPU_ConvertBlendFactor(SDL_GetBlendModeDstColorFactor(blend)); |
| 74 | ad.blend_state.src_color_blendfactor = GPU_ConvertBlendFactor(SDL_GetBlendModeSrcColorFactor(blend)); |
| 75 | |
| 76 | SDL_GPUGraphicsPipelineCreateInfo pci; |
| 77 | SDL_zero(pci); |
| 78 | pci.target_info.has_depth_stencil_target = false; |
| 79 | pci.target_info.num_color_targets = 1; |
| 80 | pci.target_info.color_target_descriptions = &ad; |
| 81 | pci.vertex_shader = GPU_GetVertexShader(shaders, params->vert_shader); |
| 82 | pci.fragment_shader = GPU_GetFragmentShader(shaders, params->frag_shader); |
| 83 | pci.multisample_state.sample_count = SDL_GPU_SAMPLECOUNT_1; |
| 84 | pci.multisample_state.enable_mask = false; |
| 85 | pci.primitive_type = params->primitive_type; |
| 86 | |
| 87 | pci.rasterizer_state.cull_mode = SDL_GPU_CULLMODE_NONE; |
| 88 | pci.rasterizer_state.fill_mode = SDL_GPU_FILLMODE_FILL; |
| 89 | pci.rasterizer_state.front_face = SDL_GPU_FRONTFACE_COUNTER_CLOCKWISE; |
| 90 | |
| 91 | SDL_GPUVertexBufferDescription vertex_buffer_desc; |
| 92 | SDL_zero(vertex_buffer_desc); |
| 93 | |
| 94 | Uint32 num_attribs = 0; |
| 95 | SDL_GPUVertexAttribute attribs[4]; |
| 96 | SDL_zero(attribs); |
| 97 | |
| 98 | bool have_attr_color = false; |
| 99 | bool have_attr_uv = false; |
| 100 | |
| 101 | switch (params->vert_shader) { |
| 102 | case VERT_SHADER_TRI_TEXTURE: |
| 103 | have_attr_uv = true; |
| 104 | SDL_FALLTHROUGH; |
| 105 | case VERT_SHADER_TRI_COLOR: |
| 106 | have_attr_color = true; |
| 107 | SDL_FALLTHROUGH; |
| 108 | default: |
| 109 | break; |
| 110 | } |
| 111 | |
| 112 | // Position |
| 113 | attribs[num_attribs].location = num_attribs; |
| 114 | attribs[num_attribs].format = SDL_GPU_VERTEXELEMENTFORMAT_FLOAT2; |
| 115 | attribs[num_attribs].offset = vertex_buffer_desc.pitch; |
| 116 | vertex_buffer_desc.pitch += 2 * sizeof(float); |
| 117 | num_attribs++; |
| 118 | |
| 119 | if (have_attr_color) { |
| 120 | // Color |
| 121 | attribs[num_attribs].location = num_attribs; |
| 122 | attribs[num_attribs].format = SDL_GPU_VERTEXELEMENTFORMAT_FLOAT4; |
| 123 | attribs[num_attribs].offset = vertex_buffer_desc.pitch; |
| 124 | vertex_buffer_desc.pitch += 4 * sizeof(float); |
| 125 | num_attribs++; |
| 126 | } |
| 127 | |
| 128 | if (have_attr_uv) { |
| 129 | // UVs |
| 130 | attribs[num_attribs].location = num_attribs; |
| 131 | attribs[num_attribs].format = SDL_GPU_VERTEXELEMENTFORMAT_FLOAT2; |
| 132 | attribs[num_attribs].offset = vertex_buffer_desc.pitch; |
| 133 | vertex_buffer_desc.pitch += 2 * sizeof(float); |
| 134 | num_attribs++; |
| 135 | } |
| 136 | |
| 137 | pci.vertex_input_state.num_vertex_attributes = num_attribs; |
| 138 | pci.vertex_input_state.vertex_attributes = attribs; |
| 139 | pci.vertex_input_state.num_vertex_buffers = 1; |
| 140 | pci.vertex_input_state.vertex_buffer_descriptions = &vertex_buffer_desc; |
| 141 | |
| 142 | return SDL_CreateGPUGraphicsPipeline(device, &pci); |
| 143 | } |
| 144 | |
| 145 | SDL_GPUGraphicsPipeline *GPU_GetPipeline(GPU_PipelineCache *cache, GPU_Shaders *shaders, SDL_GPUDevice *device, const GPU_PipelineParameters *params) |
| 146 | { |
| 147 | SDL_GPUGraphicsPipeline *pipeline = NULL; |
| 148 | if (!SDL_FindInHashTable(cache->table, params, (const void **) &pipeline)) { |
| 149 | bool inserted = false; |
| 150 | // !!! FIXME: why don't we have an SDL_alloc_copy function/macro? |
| 151 | GPU_PipelineParameters *paramscpy = (GPU_PipelineParameters *) SDL_malloc(sizeof (*paramscpy)); |
| 152 | if (paramscpy) { |
| 153 | SDL_copyp(paramscpy, params); |
| 154 | pipeline = MakePipeline(device, shaders, params); |
| 155 | if (pipeline) { |
| 156 | inserted = SDL_InsertIntoHashTable(cache->table, paramscpy, pipeline, false); |
| 157 | } |
| 158 | } |
| 159 | |
| 160 | if (!inserted) { |
| 161 | SDL_free(paramscpy); |
| 162 | if (pipeline) { |
| 163 | SDL_ReleaseGPUGraphicsPipeline(device, pipeline); |
| 164 | pipeline = NULL; |
| 165 | } |
| 166 | } |
| 167 | } |
| 168 | |
| 169 | return pipeline; |
| 170 | } |
| 171 | |
| 172 | #endif // SDL_VIDEO_RENDER_GPU |
| 173 | |