#include "common.h" #include "declarations.h" #include "QPUassembler/qpu_assembler.h" //TODO change allocations to pool allocator //TODO // ETC1 is arranged as 64-bit blocks, where each block // is 4x4 pixels. Texture tiling operates on the // 64-bit block the way it would an uncompressed // pixels. // Cube map faces appear as whole miptrees at a page-aligned offset // from the first face's miptree. uint32_t getMemoryTypeIndex(VkPhysicalDeviceMemoryProperties deviceMemoryProperties, uint32_t typeBits, VkMemoryPropertyFlags properties) { // Iterate over all memory types available for the device used in this example for (uint32_t i = 0; i < deviceMemoryProperties.memoryTypeCount; i++) { if ((typeBits & 1) == 1) { if ((deviceMemoryProperties.memoryTypes[i].propertyFlags & properties) == properties) { return i; } } typeBits >>= 1; } assert(0); } void createFullscreenQuad(VkDevice device, VkBuffer* fsqVertexBuffer, VkDeviceMemory* fsqVertexBufferMemory) { VkMemoryRequirements mr; { //create fsq vertex buffer unsigned vboSize = sizeof(float) * 4 * 3 * 2; //4 * 3 x vec2 VkBufferCreateInfo ci = {}; ci.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO; ci.size = vboSize; ci.usage = VK_BUFFER_USAGE_VERTEX_BUFFER_BIT; VkResult res = rpi_vkCreateBuffer(device, &ci, 0, fsqVertexBuffer); rpi_vkGetBufferMemoryRequirements(device, *fsqVertexBuffer, &mr); VkPhysicalDeviceMemoryProperties pdmp; rpi_vkGetPhysicalDeviceMemoryProperties(((_device*)device)->dev, &pdmp); VkMemoryAllocateInfo mai = {}; mai.sType = VK_STRUCTURE_TYPE_MEMORY_ALLOCATE_INFO; mai.allocationSize = mr.size; mai.memoryTypeIndex = getMemoryTypeIndex(pdmp, mr.memoryTypeBits, VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT | VK_MEMORY_PROPERTY_HOST_COHERENT_BIT); res = rpi_vkAllocateMemory(device, &mai, 0, fsqVertexBufferMemory); float vertices[] = { -1, -1, 0, 1, 1, -1, 1, 1, 1, 1, 1, 0, 1, 1, 1, 0, -1, 1, 0, 0, -1, -1, 0, 1 }; void* data; res = rpi_vkMapMemory(device, *fsqVertexBufferMemory, 0, mr.size, 0, &data); memcpy(data, vertices, vboSize); rpi_vkUnmapMemory(device, *fsqVertexBufferMemory); res = rpi_vkBindBufferMemory(device, *fsqVertexBuffer, *fsqVertexBufferMemory, 0); } } void createDescriptorPool(VkDevice device, VkDescriptorPool* descriptorPool) { VkDescriptorPoolSize descriptorPoolSizes[2]; descriptorPoolSizes[0].descriptorCount = 2048; descriptorPoolSizes[0].type = VK_DESCRIPTOR_TYPE_STORAGE_TEXEL_BUFFER; descriptorPoolSizes[1].descriptorCount = 2048; descriptorPoolSizes[1].type = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER; VkDescriptorPoolCreateInfo descriptorPoolCI = {}; descriptorPoolCI.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO; descriptorPoolCI.poolSizeCount = 2; descriptorPoolCI.pPoolSizes = descriptorPoolSizes; descriptorPoolCI.maxSets = 2048; descriptorPoolCI.flags = VK_DESCRIPTOR_POOL_CREATE_FREE_DESCRIPTOR_SET_BIT; rpi_vkCreateDescriptorPool(device, &descriptorPoolCI, 0, descriptorPool); } void createDescriptorSetLayouts(VkDevice device, VkDescriptorSetLayout* bufferDsl, VkDescriptorSetLayout* textureDsl) { assert(device); assert(bufferDsl); assert(textureDsl); //create blit dsl VkDescriptorSetLayoutBinding setLayoutBinding = {}; setLayoutBinding.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_TEXEL_BUFFER; setLayoutBinding.binding = 0; setLayoutBinding.descriptorCount = 1; setLayoutBinding.stageFlags = VK_SHADER_STAGE_FRAGMENT_BIT; VkDescriptorSetLayoutCreateInfo descriptorLayoutCI = {}; descriptorLayoutCI.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO; descriptorLayoutCI.bindingCount = 1; descriptorLayoutCI.pBindings = &setLayoutBinding; rpi_vkCreateDescriptorSetLayout(device, &descriptorLayoutCI, 0, bufferDsl); setLayoutBinding.descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER; rpi_vkCreateDescriptorSetLayout(device, &descriptorLayoutCI, 0, textureDsl); } void createSampler(VkDevice device, VkSampler* nearestTextureSampler, VkSampler* linearTextureSampler) { VkSamplerCreateInfo sampler = {}; sampler.sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO; sampler.magFilter = VK_FILTER_NEAREST; sampler.minFilter = VK_FILTER_NEAREST; sampler.mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST; sampler.addressModeU = VK_SAMPLER_ADDRESS_MODE_REPEAT; sampler.addressModeV = VK_SAMPLER_ADDRESS_MODE_REPEAT; sampler.addressModeW = VK_SAMPLER_ADDRESS_MODE_REPEAT; sampler.mipLodBias = 0.0f; sampler.compareOp = VK_COMPARE_OP_NEVER; sampler.borderColor = VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK; rpi_vkCreateSampler(device, &sampler, 0, nearestTextureSampler); sampler.magFilter = VK_FILTER_LINEAR; sampler.minFilter = VK_FILTER_LINEAR; rpi_vkCreateSampler(device, &sampler, 0, linearTextureSampler); } void createRendertarget(VkDevice device, uint32_t baseLayer, uint32_t baseMip, uint32_t width, uint32_t height, VkImage textureImage, VkImageView* textureView, VkRenderPass* offscreenRenderPass, VkFramebuffer* offscreenFramebuffer) { _image* img = textureImage; VkFormat format = img->format; printf("\nCopy Create RT\n"); printf("baseLayer %u\n", baseLayer); printf("baseMip %u\n", baseMip); printf("width %u\n", width); printf("height %u\n", height); //we can't render to an ETC1 texture, so we'll just stick with RGBA8 for now if(img->format == VK_FORMAT_ETC2_R8G8B8_UNORM_BLOCK) { format = VK_FORMAT_R8G8B8A8_UNORM; } VkImageViewCreateInfo view = {}; view.sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO; view.viewType = VK_IMAGE_VIEW_TYPE_2D; view.format = format; view.components.a = VK_COMPONENT_SWIZZLE_A; view.components.b = VK_COMPONENT_SWIZZLE_B; view.components.g = VK_COMPONENT_SWIZZLE_G; view.components.r = VK_COMPONENT_SWIZZLE_R; view.subresourceRange.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT; view.subresourceRange.baseMipLevel = baseMip; view.subresourceRange.baseArrayLayer = baseLayer; view.subresourceRange.layerCount = 1; view.subresourceRange.levelCount = 1; view.image = textureImage; rpi_vkCreateImageView(device, &view, 0, textureView); VkAttachmentDescription attachmentDescription = {}; attachmentDescription.format = format; attachmentDescription.initialLayout = VK_IMAGE_LAYOUT_UNDEFINED; attachmentDescription.finalLayout = VK_IMAGE_LAYOUT_SHADER_READ_ONLY_OPTIMAL; attachmentDescription.loadOp = VK_ATTACHMENT_LOAD_OP_CLEAR; attachmentDescription.storeOp = VK_ATTACHMENT_STORE_OP_STORE; attachmentDescription.stencilLoadOp = VK_ATTACHMENT_LOAD_OP_DONT_CARE; attachmentDescription.stencilStoreOp = VK_ATTACHMENT_STORE_OP_DONT_CARE; attachmentDescription.samples = VK_SAMPLE_COUNT_1_BIT; VkAttachmentReference colorReference = { 0, VK_IMAGE_LAYOUT_COLOR_ATTACHMENT_OPTIMAL }; VkSubpassDescription subpassDescription = {}; subpassDescription.pipelineBindPoint = VK_PIPELINE_BIND_POINT_GRAPHICS; subpassDescription.colorAttachmentCount = 1; subpassDescription.pColorAttachments = &colorReference; VkSubpassDependency dependencies[2]; dependencies[0].srcSubpass = VK_SUBPASS_EXTERNAL; dependencies[0].dstSubpass = 0; dependencies[0].srcStageMask = VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT; dependencies[0].dstStageMask = VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT; dependencies[0].srcAccessMask = VK_ACCESS_SHADER_READ_BIT; dependencies[0].dstAccessMask = VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT; dependencies[0].dependencyFlags = VK_DEPENDENCY_BY_REGION_BIT; dependencies[1].srcSubpass = 0; dependencies[1].dstSubpass = VK_SUBPASS_EXTERNAL; dependencies[1].srcStageMask = VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT; dependencies[1].dstStageMask = VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT; dependencies[1].srcAccessMask = VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT; dependencies[1].dstAccessMask = VK_ACCESS_SHADER_READ_BIT; dependencies[1].dependencyFlags = VK_DEPENDENCY_BY_REGION_BIT; VkRenderPassCreateInfo renderPassInfo = {}; renderPassInfo.sType = VK_STRUCTURE_TYPE_RENDER_PASS_CREATE_INFO; renderPassInfo.attachmentCount = 1; renderPassInfo.pAttachments = &attachmentDescription; renderPassInfo.subpassCount = 1; renderPassInfo.pSubpasses = &subpassDescription; renderPassInfo.dependencyCount = 2; renderPassInfo.pDependencies = dependencies; rpi_vkCreateRenderPass(device, &renderPassInfo, 0, offscreenRenderPass); VkImageView attachments = *textureView; VkFramebufferCreateInfo framebufferCreateInfo = {}; framebufferCreateInfo.renderPass = *offscreenRenderPass; framebufferCreateInfo.attachmentCount = 1; framebufferCreateInfo.pAttachments = &attachments; framebufferCreateInfo.width = width; framebufferCreateInfo.height = height; framebufferCreateInfo.layers = 1; rpi_vkCreateFramebuffer(device, &framebufferCreateInfo, 0, offscreenFramebuffer); } void createPipeline(VkDevice device, uint32_t needTexcoords, uint32_t numVertUniforms, uint32_t numFragUniforms, VkShaderModule blitShaderModule, VkDescriptorSetLayout blitDsl, VkPipelineLayout* blitPipelineLayout, VkRenderPass offscreenRenderPass, VkPipeline* blitPipeline) { VkVertexInputBindingDescription vertexInputBindingDescription = { 0, sizeof(float) * 2 * 2, VK_VERTEX_INPUT_RATE_VERTEX }; VkVertexInputAttributeDescription vertexInputAttributeDescription[2]; if(!needTexcoords) { vertexInputAttributeDescription[0].binding = 0; vertexInputAttributeDescription[0].location = 0; vertexInputAttributeDescription[0].offset = 0; vertexInputAttributeDescription[0].format = VK_FORMAT_R32G32_SFLOAT; } else { vertexInputAttributeDescription[0].binding = 0; vertexInputAttributeDescription[0].location = 0; vertexInputAttributeDescription[0].offset = 0; vertexInputAttributeDescription[0].format = VK_FORMAT_R32G32_SFLOAT; vertexInputAttributeDescription[1].binding = 0; vertexInputAttributeDescription[1].location = 1; vertexInputAttributeDescription[1].offset = sizeof(float) * 2; vertexInputAttributeDescription[1].format = VK_FORMAT_R32G32_SFLOAT; } VkPipelineVertexInputStateCreateInfo vertexInputInfo = {}; vertexInputInfo.sType = VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO; vertexInputInfo.vertexAttributeDescriptionCount = needTexcoords ? 2 : 1; vertexInputInfo.pVertexAttributeDescriptions = vertexInputAttributeDescription; vertexInputInfo.vertexBindingDescriptionCount = 1; vertexInputInfo.pVertexBindingDescriptions = &vertexInputBindingDescription; VkPipelineInputAssemblyStateCreateInfo pipelineIACreateInfo = {}; pipelineIACreateInfo.sType = VK_STRUCTURE_TYPE_PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO; pipelineIACreateInfo.topology = VK_PRIMITIVE_TOPOLOGY_TRIANGLE_LIST; VkPipelineRasterizationStateCreateInfo rastCreateInfo = {}; rastCreateInfo.sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_STATE_CREATE_INFO; rastCreateInfo.polygonMode = VK_POLYGON_MODE_FILL; rastCreateInfo.cullMode = VK_CULL_MODE_NONE; rastCreateInfo.frontFace = VK_FRONT_FACE_COUNTER_CLOCKWISE; rastCreateInfo.lineWidth = 1.0f; VkPipelineMultisampleStateCreateInfo pipelineMSCreateInfo = {}; pipelineMSCreateInfo.sType = VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO; VkPipelineColorBlendAttachmentState blendAttachState = {}; blendAttachState.colorWriteMask = 0xf; blendAttachState.blendEnable = false; VkPipelineColorBlendStateCreateInfo blendCreateInfo = {}; blendCreateInfo.sType = VK_STRUCTURE_TYPE_PIPELINE_COLOR_BLEND_STATE_CREATE_INFO; blendCreateInfo.attachmentCount = 1; blendCreateInfo.pAttachments = &blendAttachState; VkPipelineDepthStencilStateCreateInfo depthStencilState = {}; depthStencilState.depthTestEnable = false; depthStencilState.stencilTestEnable = false; //create blit pipeline VkPushConstantRange pushConstantRanges[2]; pushConstantRanges[0].offset = 0; pushConstantRanges[0].size = numVertUniforms * 4; //n * 32bits pushConstantRanges[0].stageFlags = VK_SHADER_STAGE_VERTEX_BIT; pushConstantRanges[1].offset = 0; pushConstantRanges[1].size = numFragUniforms * 4; //n * 32bits pushConstantRanges[1].stageFlags = VK_SHADER_STAGE_FRAGMENT_BIT; VkPipelineShaderStageCreateInfo shaderStageCreateInfo[2] = {}; shaderStageCreateInfo[0].sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO; shaderStageCreateInfo[0].stage = VK_SHADER_STAGE_VERTEX_BIT; shaderStageCreateInfo[0].module = blitShaderModule; shaderStageCreateInfo[0].pName = "main"; shaderStageCreateInfo[1].sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO; shaderStageCreateInfo[1].stage = VK_SHADER_STAGE_FRAGMENT_BIT; shaderStageCreateInfo[1].module = blitShaderModule; shaderStageCreateInfo[1].pName = "main"; VkPipelineLayoutCreateInfo pipelineLayoutCI = {}; pipelineLayoutCI.sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO; pipelineLayoutCI.setLayoutCount = 1; pipelineLayoutCI.pSetLayouts = &blitDsl; pipelineLayoutCI.pushConstantRangeCount = 2; pipelineLayoutCI.pPushConstantRanges = &pushConstantRanges[0]; rpi_vkCreatePipelineLayout(device, &pipelineLayoutCI, 0, blitPipelineLayout); VkDynamicState dynState = VK_DYNAMIC_STATE_VIEWPORT; VkPipelineDynamicStateCreateInfo pdsci = {}; pdsci.sType = VK_STRUCTURE_TYPE_PIPELINE_DYNAMIC_STATE_CREATE_INFO; pdsci.dynamicStateCount = 1; pdsci.pDynamicStates = &dynState; VkPipelineViewportStateCreateInfo pvsci = {}; pvsci.sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_STATE_CREATE_INFO; pvsci.viewportCount = 0; pvsci.scissorCount = 0; VkGraphicsPipelineCreateInfo pipelineInfo = {}; pipelineInfo.sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO; pipelineInfo.stageCount = 2; pipelineInfo.pStages = &shaderStageCreateInfo[0]; pipelineInfo.pVertexInputState = &vertexInputInfo; pipelineInfo.pInputAssemblyState = &pipelineIACreateInfo; pipelineInfo.pViewportState = &pvsci; pipelineInfo.pDynamicState = &pdsci; pipelineInfo.pRasterizationState = &rastCreateInfo; pipelineInfo.pMultisampleState = &pipelineMSCreateInfo; pipelineInfo.pColorBlendState = &blendCreateInfo; pipelineInfo.renderPass = offscreenRenderPass; pipelineInfo.basePipelineIndex = -1; pipelineInfo.pDepthStencilState = &depthStencilState; pipelineInfo.layout = *blitPipelineLayout; VkResult res = rpi_vkCreateGraphicsPipelines(device, VK_NULL_HANDLE, 1, &pipelineInfo, NULL, blitPipeline); } void createBufferToTextureShaderModule(VkDevice device, VkShaderModule* blitShaderModule) { char vs_asm_code[] = ///0x40000000 = 2.0 ///uni = 1.0 ///rb0 = 2 - 1 = 1 "sig_small_imm ; rx0 = fsub.ws.always(b, a, uni, 0x40000000) ; nop = nop(r0, r0) ;\n" ///set up VPM read for subsequent reads ///0x00201a00: 0000 0000 0010 0000 0001 1010 0000 0000 ///addr: 0 ///size: 32bit ///packed ///horizontal ///stride=1 ///vectors to read = 2 (how many components) "sig_load_imm ; vr_setup = load32.always(0x00201a00) ; nop = load32.always() ;\n" ///uni = viewportXScale ///r0 = vpm * uni "sig_none ; nop = nop(r0, r0, vpm_read, uni) ; r0 = fmul.always(a, b) ;\n" ///r1 = r0 * rb0 (1) "sig_none ; nop = nop(r0, r0, nop, rb0) ; r1 = fmul.always(r0, b) ;\n" ///uni = viewportYScale ///ra0.16a = int(r1), r2 = vpm * uni "sig_none ; rx0.16a = ftoi.always(r1, r1, vpm_read, uni) ; r2 = fmul.always(a, b) ;\n" ///r3 = r2 * rb0 "sig_none ; nop = nop(r0, r0, nop, rb0) ; r3 = fmul.always(r2, b) ;\n" ///ra0.16b = int(r3) "sig_none ; rx0.16b = ftoi.always(r3, r3) ; nop = nop(r0, r0) ;\n" ///set up VPM write for subsequent writes ///0x00001a00: 0000 0000 0000 0000 0001 1010 0000 0000 ///addr: 0 ///size: 32bit ///horizontal ///stride = 1 "sig_load_imm ; vw_setup = load32.always.ws(0x00001a00) ; nop = load32.always() ;\n" ///shaded vertex format for PSE /// Ys and Xs ///vpm = ra0 "sig_none ; vpm = or.always(a, a, ra0, nop) ; nop = nop(r0, r0);\n" /// Zs ///uni = 0.5 ///vpm = uni "sig_none ; vpm = or.always(a, a, uni, nop) ; nop = nop(r0, r0);\n" /// 1.0 / Wc ///vpm = rb0 (1) "sig_none ; vpm = or.always(b, b, nop, rb0) ; nop = nop(r0, r0);\n" ///END "sig_end ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "\0"; char cs_asm_code[] = ///uni = 1.0 ///r3 = 2.0 - uni "sig_small_imm ; r3 = fsub.always(b, a, uni, 0x40000000) ; nop = nop(r0, r0);\n" "sig_load_imm ; vr_setup = load32.always(0x00201a00) ; nop = load32.always() ;\n" ///r2 = vpm "sig_none ; r2 = or.always(a, a, vpm_read, nop) ; nop = nop(r0, r0);\n" "sig_load_imm ; vw_setup = load32.always.ws(0x00001a00) ; nop = load32.always() ;\n" ///shaded coordinates format for PTB /// write Xc ///r1 = vpm, vpm = r2 "sig_none ; r1 = or.always(a, a, vpm_read, nop) ; vpm = v8min.always(r2, r2);\n" /// write Yc ///uni = viewportXscale ///vpm = r1, r2 = r2 * uni "sig_none ; vpm = or.always(r1, r1, uni, nop) ; r2 = fmul.always(r2, a);\n" ///uni = viewportYscale ///r1 = r1 * uni "sig_none ; nop = nop(r0, r0, uni, nop) ; r1 = fmul.always(r1, a);\n" ///r0 = r2 * r3 "sig_none ; nop = nop(r0, r0) ; r0 = fmul.always(r2, r3);\n" ///ra0.16a = r0, r1 = r1 * r3 "sig_none ; rx0.16a = ftoi.always(r0, r0) ; r1 = fmul.always(r1, r3) ;\n" ///ra0.16b = r1 "sig_none ; rx0.16b = ftoi.always(r1, r1) ; nop = nop(r0, r0) ;\n" ///write Zc ///vpm = 0 "sig_small_imm ; vpm = or.always(b, b, nop, 0) ; nop = nop(r0, r0) ;\n" ///write Wc ///vpm = 1.0 "sig_small_imm ; vpm = or.always(b, b, nop, 0x3f800000) ; nop = nop(r0, r0) ;\n" ///write Ys and Xs ///vpm = ra0 "sig_none ; vpm = or.always(a, a, ra0, nop) ; nop = nop(r0, r0) ;\n" ///write Zs ///uni = 0.5 ///vpm = uni "sig_none ; vpm = or.always(a, a, uni, nop) ; nop = nop(r0, r0) ;\n" ///write 1/Wc ///vpm = r3 "sig_none ; vpm = or.always(r3, r3) ; nop = nop(r0, r0) ;\n" ///END "sig_end ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "\0"; //clever: use small immedate -1 interpreted as 0xffffffff (white) to set color to white //"sig_small_imm ; tlb_color_all = or.always(b, b, nop, -1) ; nop = nop(r0, r0) ;" //8bit access //abcd //BGRA /** "General-memory lookups are performed by writing to just the ā€˜sā€™ parameter, using the absolute memory address. In this case no uniform is read. General-memory lookups always return a 32-bit value, and the bottom two bits of the address are ignored." /**/ //blit buffer to texture (generic buffer read) char blit_fs_asm_code[] = ///"sig_load_imm ; r2 = load32.always(0x44f00000) ; nop = load32() ;" //width = 1920.0 "sig_none ; r2 = or.always(b, b, nop, uni) ; nop = nop(r0, r0) ;" //width "sig_none ; r1 = itof.always(b, b, x_pix, y_pix) ; nop = nop(r0, r0) ;" //FragCoord Y "sig_none ; r0 = itof.always(a, a, x_pix, y_pix) ; r1 = fmul.always(r1, r2) ;" //FragCoord X, r1 = Y * width "sig_none ; r0 = fadd.always(r0, r1) ; r0 = nop(r0, r0) ;" //r0 = Y * width + X "sig_none ; r0 = nop(r0, r0, nop, uni) ; r0 = fmul.always(r0, b) ;" //r0 = (Y * width + X) * pixelBpp "sig_small_imm ; nop = nop(r0, r0, nop, 0x3e000000) ; r0 = fmul.always(r0, b) ;" //r0 = ((Y * width + X) * pixelBpp) / 8 "sig_none ; r0 = ftoi.always(r0, r0) ; nop = nop(r0, r0) ;" //convert to integer ///write general mem access address ///first argument must be clamped to [0...bufsize-4] ///eg must do min(max(x,0), uni) ///second argument must be a uniform (containing base address, which is 0) ///writing tmu0_s signals that all coordinates are written "sig_small_imm ; r0 = max.always(r0, b, nop, 0) ; nop = nop(r0, r0) ;" //clamp general access "sig_none ; r0 = min.always(r0, b, nop, uni) ; nop = nop(r0, r0) ;" //uni = width * height * pixelBytes - pixelBytes "sig_none ; tmu0_s = add.always(r0, b, nop, uni) ; nop = nop(r0, r0) ;" //uni = 0 ///suspend thread (after 2 nops) to wait for TMU request to finish "sig_thread_switch ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" ///read TMU0 request result to R4 "sig_load_tmu0 ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" ///when thread has been awakened, MOV from R4 to R0 "sig_none ; r0 = fmax.pm.always.8a(r4, r4) ; nop = nop(r0, r0) ;" "sig_none ; r1 = fmax.pm.always.8b(r4, r4) ; r0.8a = v8min.always(r0, r0) ;" "sig_none ; r2 = fmax.pm.always.8c(r4, r4) ; r0.8b = v8min.always(r1, r1) ;" "sig_none ; r3 = fmax.pm.always.8d(r4, r4) ; r0.8c = v8min.always(r2, r2) ;" "sig_none ; nop = nop.pm(r0, r0) ; r0.8d = v8min.always(r3, r3) ;" "sig_none ; tlb_color_all = or.always(r0, r0) ; nop = nop(r0, r0) ;" "sig_end ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" "sig_unlock_score ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" "\0"; char* blit_asm_strings[] = { (char*)cs_asm_code, (char*)vs_asm_code, (char*)blit_fs_asm_code, 0 }; VkRpiAssemblyMappingEXT vertexMappings[] = { //vertex shader uniforms { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 0, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 4, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 8, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 12, //resource offset }, }; VkRpiAssemblyMappingEXT fragmentMappings[] = { //fragment shader uniforms { VK_RPI_ASSEMBLY_MAPPING_TYPE_DESCRIPTOR, VK_DESCRIPTOR_TYPE_STORAGE_TEXEL_BUFFER, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 0, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 0, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 4, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 8, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 12, //resource offset } }; uint32_t spirv[6]; uint64_t* asm_ptrs[4] = {}; uint32_t asm_sizes[4] = {}; VkRpiAssemblyMappingEXT* asm_mappings[4] = {}; uint32_t asm_mappings_sizes[4] = {}; VkRpiShaderModuleAssemblyCreateInfoEXT shaderModuleCreateInfo = {}; shaderModuleCreateInfo.instructions = asm_ptrs; shaderModuleCreateInfo.numInstructions = asm_sizes; shaderModuleCreateInfo.mappings = asm_mappings; shaderModuleCreateInfo.numMappings = asm_mappings_sizes; asm_mappings[VK_RPI_ASSEMBLY_TYPE_VERTEX] = vertexMappings; asm_mappings_sizes[VK_RPI_ASSEMBLY_TYPE_VERTEX] = sizeof(vertexMappings) / sizeof(VkRpiAssemblyMappingEXT); asm_mappings[VK_RPI_ASSEMBLY_TYPE_FRAGMENT] = fragmentMappings; asm_mappings_sizes[VK_RPI_ASSEMBLY_TYPE_FRAGMENT] = sizeof(fragmentMappings) / sizeof(VkRpiAssemblyMappingEXT); //TODO use allocator { //assemble cs code asm_sizes[0] = get_num_instructions(cs_asm_code); uint32_t size = sizeof(uint64_t)*asm_sizes[0]; //TODO this alloc feels kinda useless, we just copy the data anyway to kernel space //why not map kernel space mem to user space instead? asm_ptrs[0] = (uint64_t*)malloc(size); assemble_qpu_asm(cs_asm_code, asm_ptrs[0]); } { //assemble vs code asm_sizes[1] = get_num_instructions(vs_asm_code); uint32_t size = sizeof(uint64_t)*asm_sizes[1]; //TODO this alloc feels kinda useless, we just copy the data anyway to kernel space //why not map kernel space mem to user space instead? asm_ptrs[1] = (uint64_t*)malloc(size); assemble_qpu_asm(vs_asm_code, asm_ptrs[1]); } { //assemble fs code asm_sizes[2] = get_num_instructions(blit_fs_asm_code); uint32_t size = sizeof(uint64_t)*asm_sizes[2]; //TODO this alloc feels kinda useless, we just copy the data anyway to kernel space //why not map kernel space mem to user space instead? asm_ptrs[2] = (uint64_t*)malloc(size); assemble_qpu_asm(blit_fs_asm_code, asm_ptrs[2]); } spirv[0] = 0x07230203; spirv[1] = 0x00010000; spirv[2] = 0x14E45250; spirv[3] = 1; spirv[4] = (uint32_t)&shaderModuleCreateInfo; //words start here spirv[5] = 1 << 16; VkShaderModuleCreateInfo smci = {}; smci.sType = VK_STRUCTURE_TYPE_SHADER_MODULE_CREATE_INFO; smci.codeSize = sizeof(uint32_t)*6; smci.pCode = spirv; rpi_vkCreateShaderModule(device, &smci, 0, blitShaderModule); assert(blitShaderModule); for(uint32_t c = 0; c < 4; ++c) { free(asm_ptrs[c]); } } void createTextureToTextureShaderModule(VkDevice device, VkShaderModule* blitShaderModule) { char vs_asm_code[] = ///0x40000000 = 2.0 ///uni = 1.0 ///rb0 = 2 - 1 = 1 "sig_small_imm ; rx0 = fsub.ws.always(b, a, uni, 0x40000000) ; nop = nop(r0, r0) ;\n" ///set up VPM read for subsequent reads ///0x00201a00: 0000 0000 0010 0000 0001 1010 0000 0000 ///addr: 0 ///size: 32bit ///packed ///horizontal ///stride=1 ///vectors to read = 4 "sig_load_imm ; vr_setup = load32.always(0x00401a00) ; nop = load32.always() ;\n" ///uni = viewportXScale ///r0 = vpm * uni "sig_none ; nop = nop(r0, r0, vpm_read, uni) ; r0 = fmul.always(a, b) ;\n" ///r1 = r0 * rb0 (1) "sig_none ; nop = nop(r0, r0, nop, rb0) ; r1 = fmul.always(r0, b) ;\n" ///uni = viewportYScale ///ra0.16a = int(r1), r2 = vpm * uni "sig_none ; rx0.16a = ftoi.always(r1, r1, vpm_read, uni) ; r2 = fmul.always(a, b) ;\n" ///r3 = r2 * rb0 ///r0 = vpm "sig_none ; r0 = or.always(a, a, vpm_read, rb0) ; r3 = fmul.always(r2, b) ;\n" ///ra0.16b = int(r3) ///r1 = vpm "sig_none ; rx0.16b = ftoi.always(r3, r3, vpm_read, nop) ; r1 = v8min.always(a, a) ;\n" ///set up VPM write for subsequent writes ///0x00001a00: 0000 0000 0000 0000 0001 1010 0000 0000 ///addr: 0 ///size: 32bit ///horizontal ///stride = 1 "sig_load_imm ; vw_setup = load32.always.ws(0x00001a00) ; nop = load32.always() ;\n" ///shaded vertex format for PSE /// Ys and Xs ///vpm = ra0 "sig_none ; vpm = or.always(a, a, ra0, nop) ; nop = nop(r0, r0);\n" /// Zs ///uni = 0.5 ///vpm = uni "sig_none ; vpm = or.always(a, a, uni, nop) ; nop = nop(r0, r0);\n" /// 1.0 / Wc ///vpm = rb0 (1) "sig_none ; vpm = or.always(b, b, nop, rb0) ; nop = nop(r0, r0);\n" ///vpm = r0 "sig_none ; vpm = or.always(r0, r0) ; nop = nop(r0, r0);\n" ///vpm = r1 "sig_none ; vpm = or.always(r1, r1) ; nop = nop(r0, r0);\n" ///END "sig_end ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "\0"; char cs_asm_code[] = ///uni = 1.0 ///r3 = 2.0 - uni "sig_small_imm ; r3 = fsub.always(b, a, uni, 0x40000000) ; nop = nop(r0, r0);\n" "sig_load_imm ; vr_setup = load32.always(0x00201a00) ; nop = load32.always() ;\n" ///r2 = vpm "sig_none ; r2 = or.always(a, a, vpm_read, nop) ; nop = nop(r0, r0);\n" "sig_load_imm ; vw_setup = load32.always.ws(0x00001a00) ; nop = load32.always() ;\n" ///shaded coordinates format for PTB /// write Xc ///r1 = vpm, vpm = r2 "sig_none ; r1 = or.always(a, a, vpm_read, nop) ; vpm = v8min.always(r2, r2);\n" /// write Yc ///uni = viewportXscale ///vpm = r1, r2 = r2 * uni "sig_none ; vpm = or.always(r1, r1, uni, nop) ; r2 = fmul.always(r2, a);\n" ///uni = viewportYscale ///r1 = r1 * uni "sig_none ; nop = nop(r0, r0, uni, nop) ; r1 = fmul.always(r1, a);\n" ///r0 = r2 * r3 "sig_none ; nop = nop(r0, r0) ; r0 = fmul.always(r2, r3);\n" ///ra0.16a = r0, r1 = r1 * r3 "sig_none ; rx0.16a = ftoi.always(r0, r0) ; r1 = fmul.always(r1, r3) ;\n" ///ra0.16b = r1 "sig_none ; rx0.16b = ftoi.always(r1, r1) ; nop = nop(r0, r0) ;\n" ///write Zc ///vpm = 0 "sig_small_imm ; vpm = or.always(b, b, nop, 0) ; nop = nop(r0, r0) ;\n" ///write Wc ///vpm = 1.0 "sig_small_imm ; vpm = or.always(b, b, nop, 0x3f800000) ; nop = nop(r0, r0) ;\n" ///write Ys and Xs ///vpm = ra0 "sig_none ; vpm = or.always(a, a, ra0, nop) ; nop = nop(r0, r0) ;\n" ///write Zs ///uni = 0.5 ///vpm = uni "sig_none ; vpm = or.always(a, a, uni, nop) ; nop = nop(r0, r0) ;\n" ///write 1/Wc ///vpm = r3 "sig_none ; vpm = or.always(r3, r3) ; nop = nop(r0, r0) ;\n" ///END "sig_end ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;\n" "\0"; //sample texture char sample_fs_asm_code[] = ///r0 = varyingX * W "sig_none ; nop = nop(r0, r0, pay_zw, vary) ; r0 = fmul.always(a, b) ;" ///r2 = r0 + r5 (C) ///r0 = varyingY * W "sig_none ; r2 = fadd.always(r0, r5, pay_zw, vary) ; r0 = fmul.always(a, b) ;" ///r3 = r0 + r5 (C) "sig_none ; r3 = fadd.pm.always(r0, r5, nop, uni) ; r0 = v8min.always(b, b) ;" ///write texture addresses (x, y) ///writing tmu0_s signals that all coordinates are written "sig_none ; tmu0_b = or.always(r0, r0) ; nop = nop(r0, r0) ;" "sig_none ; tmu0_t = or.always(r3, r3) ; nop = nop(r0, r0) ;" "sig_none ; tmu0_s = or.always(r2, r2) ; nop = nop(r0, r0) ;" ///suspend thread (after 2 nops) to wait for TMU request to finish "sig_thread_switch ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" ///read TMU0 request result to R4 "sig_load_tmu0 ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" ///when thread has been awakened, MOV from R4 to R0 "sig_none ; r0 = fmax.pm.always.8a(r4, r4) ; nop = nop(r0, r0) ;" "sig_none ; r1 = fmax.pm.always.8b(r4, r4) ; r0.8a = v8min.always(r0, r0) ;" "sig_none ; r2 = fmax.pm.always.8c(r4, r4) ; r0.8b = v8min.always(r1, r1) ;" "sig_none ; r3 = fmax.pm.always.8d(r4, r4) ; r0.8c = v8min.always(r2, r2) ;" "sig_none ; nop = nop.pm(r0, r0) ; r0.8d = v8min.always(r3, r3) ;" ///"sig_small_imm; r0 = or.always(b, b, nop, -1) ; nop = nop(r0, r0) ;" "sig_none ; tlb_color_all = or.always(r0, r0) ; nop = nop(r0, r0) ;" "sig_end ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" "sig_none ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" "sig_unlock_score ; nop = nop(r0, r0) ; nop = nop(r0, r0) ;" "\0"; char* blit_asm_strings[] = { (char*)cs_asm_code, (char*)vs_asm_code, (char*)sample_fs_asm_code, 0 }; VkRpiAssemblyMappingEXT vertexMappings[] = { //vertex shader uniforms { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 0, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 4, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 8, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 12, //resource offset }, }; VkRpiAssemblyMappingEXT fragmentMappings[] = { //fragment shader uniforms { VK_RPI_ASSEMBLY_MAPPING_TYPE_PUSH_CONSTANT, VK_DESCRIPTOR_TYPE_MAX_ENUM, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 0, //resource offset }, { VK_RPI_ASSEMBLY_MAPPING_TYPE_DESCRIPTOR, VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER, //descriptor type 0, //descriptor set # 0, //descriptor binding # 0, //descriptor array element # 0, //resource offset }, }; uint32_t spirv[6]; uint64_t* asm_ptrs[4] = {}; uint32_t asm_sizes[4] = {}; VkRpiAssemblyMappingEXT* asm_mappings[4] = {}; uint32_t asm_mappings_sizes[4] = {}; VkRpiShaderModuleAssemblyCreateInfoEXT shaderModuleCreateInfo = {}; shaderModuleCreateInfo.instructions = asm_ptrs; shaderModuleCreateInfo.numInstructions = asm_sizes; shaderModuleCreateInfo.mappings = asm_mappings; shaderModuleCreateInfo.numMappings = asm_mappings_sizes; asm_mappings[VK_RPI_ASSEMBLY_TYPE_VERTEX] = vertexMappings; asm_mappings_sizes[VK_RPI_ASSEMBLY_TYPE_VERTEX] = sizeof(vertexMappings) / sizeof(VkRpiAssemblyMappingEXT); asm_mappings[VK_RPI_ASSEMBLY_TYPE_FRAGMENT] = fragmentMappings; asm_mappings_sizes[VK_RPI_ASSEMBLY_TYPE_FRAGMENT] = sizeof(fragmentMappings) / sizeof(VkRpiAssemblyMappingEXT); //TODO use allocator { //assemble cs code asm_sizes[0] = get_num_instructions(cs_asm_code); uint32_t size = sizeof(uint64_t)*asm_sizes[0]; //TODO this alloc feels kinda useless, we just copy the data anyway to kernel space //why not map kernel space mem to user space instead? asm_ptrs[0] = (uint64_t*)malloc(size); assemble_qpu_asm(cs_asm_code, asm_ptrs[0]); } { //assemble vs code asm_sizes[1] = get_num_instructions(vs_asm_code); uint32_t size = sizeof(uint64_t)*asm_sizes[1]; //TODO this alloc feels kinda useless, we just copy the data anyway to kernel space //why not map kernel space mem to user space instead? asm_ptrs[1] = (uint64_t*)malloc(size); assemble_qpu_asm(vs_asm_code, asm_ptrs[1]); } { //assemble fs code asm_sizes[2] = get_num_instructions(sample_fs_asm_code); uint32_t size = sizeof(uint64_t)*asm_sizes[2]; //TODO this alloc feels kinda useless, we just copy the data anyway to kernel space //why not map kernel space mem to user space instead? asm_ptrs[2] = (uint64_t*)malloc(size); assemble_qpu_asm(sample_fs_asm_code, asm_ptrs[2]); } spirv[0] = 0x07230203; spirv[1] = 0x00010000; spirv[2] = 0x14E45250; spirv[3] = 1; spirv[4] = (uint32_t)&shaderModuleCreateInfo; //words start here spirv[5] = 1 << 16; VkShaderModuleCreateInfo smci = {}; smci.sType = VK_STRUCTURE_TYPE_SHADER_MODULE_CREATE_INFO; smci.codeSize = sizeof(uint32_t)*6; smci.pCode = spirv; rpi_vkCreateShaderModule(device, &smci, 0, blitShaderModule); assert(blitShaderModule); for(uint32_t c = 0; c < 4; ++c) { free(asm_ptrs[c]); } } void setupEmulationResources(VkDevice device) { //create resources that won't change _device* dev = device; createFullscreenQuad(device, &dev->emulFsqVertexBuffer, &dev->emulFsqVertexBufferMemory); createDescriptorPool(device, &dev->emulDescriptorPool); createDescriptorSetLayouts(device, &dev->emulBufferDsl, &dev->emulTextureDsl); createSampler(device, &dev->emulNearestTextureSampler, &dev->emulLinearTextureSampler); createBufferToTextureShaderModule(device, &dev->emulBufferToTextureShaderModule); createTextureToTextureShaderModule(device, &dev->emulTextureToTextureShaderModule); } VKAPI_ATTR void VKAPI_CALL rpi_vkCmdCopyBufferToImage( VkCommandBuffer commandBuffer, VkBuffer srcBuffer, VkImage dstImage, VkImageLayout dstImageLayout, uint32_t regionCount, const VkBufferImageCopy* pRegions) { _commandBuffer* cmdBuf = commandBuffer; _device* device = cmdBuf->dev; _buffer* buf = srcBuffer; _image* img = dstImage; for(uint32_t c = 0; c < regionCount; ++c) { //TODO support this assert(!pRegions[c].bufferRowLength); assert(!pRegions[c].bufferImageHeight); uint32_t width = pRegions[c].imageExtent.width, height = pRegions[c].imageExtent.height; uint32_t pixelBpp = getFormatBpp(img->format); VkBufferView texelBufferView; VkBufferViewCreateInfo bvci = {}; bvci.sType = VK_STRUCTURE_TYPE_BUFFER_VIEW_CREATE_INFO; bvci.buffer = buf; bvci.format = img->format; bvci.offset = pRegions[c].bufferOffset; bvci.range = (width * height * pixelBpp) >> 3; rpi_vkCreateBufferView(device, &bvci, 0, &texelBufferView); VkDescriptorSet blitDescriptorSet; VkImageView textureView; VkRenderPass offscreenRenderPass; VkFramebuffer offscreenFramebuffer; VkPipeline blitPipeline; VkPipelineLayout blitPipelineLayout; //create blit descriptor set VkDescriptorSetAllocateInfo allocInfo = {}; allocInfo.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_ALLOCATE_INFO; allocInfo.descriptorPool = device->emulDescriptorPool; allocInfo.descriptorSetCount = 1; allocInfo.pSetLayouts = &device->emulBufferDsl; rpi_vkAllocateDescriptorSets(device, &allocInfo, &blitDescriptorSet); VkWriteDescriptorSet writeDescriptorSet = {}; writeDescriptorSet.sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET; writeDescriptorSet.dstSet = blitDescriptorSet; writeDescriptorSet.dstBinding = 0; writeDescriptorSet.descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_TEXEL_BUFFER; writeDescriptorSet.pTexelBufferView = &texelBufferView; writeDescriptorSet.descriptorCount = 1; rpi_vkUpdateDescriptorSets(device, 1, &writeDescriptorSet, 0, 0); createRendertarget(device, pRegions[c].imageSubresource.baseArrayLayer, pRegions[c].imageSubresource.mipLevel, width, height, img, &textureView, &offscreenRenderPass, &offscreenFramebuffer); createPipeline(device, 0, 4, 5, device->emulBufferToTextureShaderModule, device->emulBufferDsl, &blitPipelineLayout, offscreenRenderPass, &blitPipeline); //offscreen rendering VkClearValue offscreenClearValues = { .color = { 1.0f, 0.0f, 1.0f, 1.0f } }; VkRenderPassBeginInfo renderPassInfo = {}; renderPassInfo.sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO; renderPassInfo.renderArea.offset.x = 0; renderPassInfo.renderArea.offset.y = 0; renderPassInfo.renderArea.extent.width = width; renderPassInfo.renderArea.extent.height = height; renderPassInfo.framebuffer = offscreenFramebuffer; renderPassInfo.renderPass = offscreenRenderPass; renderPassInfo.clearValueCount = 1; renderPassInfo.pClearValues = &offscreenClearValues; rpi_vkCmdBeginRenderPass(commandBuffer, &renderPassInfo, VK_SUBPASS_CONTENTS_INLINE); rpi_vkCmdBindPipeline(commandBuffer, VK_PIPELINE_BIND_POINT_GRAPHICS, blitPipeline); VkViewport vp = {}; vp.x = 0.0f; vp.y = 0.0f; vp.width = (float)width; vp.height = (float)height; vp.minDepth = 0.0f; vp.maxDepth = 1.0f; rpi_vkCmdSetViewport(commandBuffer, 0, 1, &vp); VkDeviceSize offsets = 0; rpi_vkCmdBindVertexBuffers(commandBuffer, 0, 1, &device->emulFsqVertexBuffer, &offsets ); rpi_vkCmdBindDescriptorSets(commandBuffer, VK_PIPELINE_BIND_POINT_GRAPHICS, blitPipelineLayout, 0, 1, &blitDescriptorSet, 0, 0); float Wcoeff = 1.0f; //1.0f / Wc = 2.0 - Wcoeff float viewportScaleX = (float)(width) * 0.5f * 16.0f; float viewportScaleY = -1.0f * (float)(height) * 0.5f * 16.0f; float Zs = 0.5f; uint32_t vertConstants[4]; vertConstants[0] = *(uint32_t*)&Wcoeff; vertConstants[1] = *(uint32_t*)&viewportScaleX; vertConstants[2] = *(uint32_t*)&viewportScaleY; vertConstants[3] = *(uint32_t*)&Zs; rpi_vkCmdPushConstants(commandBuffer, blitPipelineLayout, VK_SHADER_STAGE_VERTEX_BIT, 0, sizeof(vertConstants), &vertConstants); float w = width; float bppfloat = pixelBpp; uint32_t size = ((width * height * pixelBpp) >> 3) - ((pixelBpp > 32 ? pixelBpp : 32) >> 3); uint32_t fragConstants[4]; fragConstants[0] = *(uint32_t*)&w; fragConstants[1] = *(uint32_t*)&bppfloat; fragConstants[2] = size; fragConstants[3] = pRegions[c].bufferOffset; rpi_vkCmdPushConstants(commandBuffer, blitPipelineLayout, VK_SHADER_STAGE_FRAGMENT_BIT, 0, sizeof(fragConstants), &fragConstants); rpi_vkCmdDraw(commandBuffer, 6, 1, 0, 0); rpi_vkCmdEndRenderPass(commandBuffer); //free up resources rpi_vkDestroyPipelineLayout(device, blitPipelineLayout, 0); rpi_vkDestroyPipeline(device, blitPipeline, 0); rpi_vkFreeDescriptorSets(device, device->emulDescriptorPool, 1, &blitDescriptorSet); rpi_vkDestroyImageView(device, textureView, 0); rpi_vkDestroyRenderPass(device, offscreenRenderPass, 0); rpi_vkDestroyFramebuffer(device, offscreenFramebuffer, 0); } img->layout = dstImageLayout; } VKAPI_ATTR void VKAPI_CALL rpi_vkCmdBlitImage( VkCommandBuffer commandBuffer, VkImage srcImage, VkImageLayout srcImageLayout, VkImage dstImage, VkImageLayout dstImageLayout, uint32_t regionCount, const VkImageBlit* pRegions, VkFilter filter) { _commandBuffer* cmdBuf = commandBuffer; _device* device = cmdBuf->dev; _image* srcImg = srcImage; _image* dstImg = dstImage; for(uint32_t c = 0; c < regionCount; ++c) { uint32_t srcWidth = pRegions[c].srcOffsets[1].x - pRegions[c].srcOffsets[0].x; uint32_t srcHeight = pRegions[c].srcOffsets[1].y - pRegions[c].srcOffsets[0].y; uint32_t dstWidth = pRegions[c].dstOffsets[1].x - pRegions[c].dstOffsets[0].x; uint32_t dstHeight = pRegions[c].dstOffsets[1].y - pRegions[c].dstOffsets[0].y; uint32_t srcMipLevel = pRegions[c].srcSubresource.mipLevel; uint32_t dstMipLevel = pRegions[c].dstSubresource.mipLevel; uint32_t srcPixelBpp = getFormatBpp(srcImg->format); uint32_t dstPixelBpp = getFormatBpp(dstImg->format); VkDescriptorSet blitDescriptorSet; VkImageView srcTextureView; VkImageView dstTextureView; VkRenderPass offscreenRenderPass; VkFramebuffer offscreenFramebuffer; VkPipeline blitPipeline; VkPipelineLayout blitPipelineLayout; VkSampler mipSampler; VkSamplerCreateInfo samplerCI = {}; samplerCI.sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO; samplerCI.magFilter = filter == VK_FILTER_LINEAR ? VK_FILTER_LINEAR : VK_FILTER_NEAREST; samplerCI.minFilter = filter == VK_FILTER_LINEAR ? VK_FILTER_LINEAR : VK_FILTER_NEAREST; samplerCI.mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST; samplerCI.addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE; samplerCI.addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE; samplerCI.addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE; samplerCI.mipLodBias = 1.0f; //disable auto lod samplerCI.compareOp = VK_COMPARE_OP_NEVER; samplerCI.borderColor = VK_BORDER_COLOR_FLOAT_TRANSPARENT_BLACK; rpi_vkCreateSampler(device, &samplerCI, 0, &mipSampler); _sampler* s = mipSampler; VkImageViewCreateInfo view = {}; view.sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO; view.viewType = VK_IMAGE_VIEW_TYPE_2D; view.format = srcImg->format; view.subresourceRange.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT; view.subresourceRange.baseMipLevel = srcMipLevel; view.subresourceRange.baseArrayLayer = 0; view.subresourceRange.layerCount = 1; view.subresourceRange.levelCount = srcImg->miplevels; view.image = srcImage; rpi_vkCreateImageView(device, &view, 0, &srcTextureView); //create blit descriptor set VkDescriptorSetAllocateInfo allocInfo = {}; allocInfo.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_ALLOCATE_INFO; allocInfo.descriptorPool = device->emulDescriptorPool; allocInfo.descriptorSetCount = 1; allocInfo.pSetLayouts = &device->emulTextureDsl; rpi_vkAllocateDescriptorSets(device, &allocInfo, &blitDescriptorSet); VkDescriptorImageInfo imageInfo; imageInfo.imageView = srcTextureView; imageInfo.imageLayout = VK_IMAGE_LAYOUT_SHADER_READ_ONLY_OPTIMAL; imageInfo.sampler = mipSampler; VkWriteDescriptorSet writeDescriptorSet = {}; writeDescriptorSet.sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET; writeDescriptorSet.dstSet = blitDescriptorSet; writeDescriptorSet.dstBinding = 0; writeDescriptorSet.descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER; writeDescriptorSet.pImageInfo = &imageInfo; writeDescriptorSet.descriptorCount = 1; rpi_vkUpdateDescriptorSets(device, 1, &writeDescriptorSet, 0, 0); createRendertarget(device, pRegions[c].dstSubresource.baseArrayLayer, dstMipLevel, dstWidth, dstHeight, dstImage, &dstTextureView, &offscreenRenderPass, &offscreenFramebuffer); createPipeline(device, 1, 4, 2, device->emulTextureToTextureShaderModule, device->emulTextureDsl, &blitPipelineLayout, offscreenRenderPass, &blitPipeline); //offscreen rendering VkClearValue offscreenClearValues = { .color = { 1.0f, 0.0f, 1.0f, 1.0f } }; VkRenderPassBeginInfo renderPassInfo = {}; renderPassInfo.sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO; renderPassInfo.renderArea.offset.x = 0; renderPassInfo.renderArea.offset.y = 0; renderPassInfo.renderArea.extent.width = dstWidth; renderPassInfo.renderArea.extent.height = dstHeight; renderPassInfo.framebuffer = offscreenFramebuffer; renderPassInfo.renderPass = offscreenRenderPass; renderPassInfo.clearValueCount = 1; renderPassInfo.pClearValues = &offscreenClearValues; rpi_vkCmdBeginRenderPass(commandBuffer, &renderPassInfo, VK_SUBPASS_CONTENTS_INLINE); rpi_vkCmdBindPipeline(commandBuffer, VK_PIPELINE_BIND_POINT_GRAPHICS, blitPipeline); VkViewport vp = {}; vp.x = 0.0f; vp.y = 0.0f; vp.width = (float)dstWidth; vp.height = (float)dstHeight; vp.minDepth = 0.0f; vp.maxDepth = 1.0f; rpi_vkCmdSetViewport(commandBuffer, 0, 1, &vp); VkDeviceSize offsets = 0; rpi_vkCmdBindVertexBuffers(commandBuffer, 0, 1, &device->emulFsqVertexBuffer, &offsets ); rpi_vkCmdBindDescriptorSets(commandBuffer, VK_PIPELINE_BIND_POINT_GRAPHICS, blitPipelineLayout, 0, 1, &blitDescriptorSet, 0, 0); float Wcoeff = 1.0f; //1.0f / Wc = 2.0 - Wcoeff float viewportScaleX = (float)(dstWidth) * 0.5f * 16.0f; float viewportScaleY = -1.0f * (float)(dstHeight) * 0.5f * 16.0f; float Zs = 0.5f; uint32_t vertConstants[4]; vertConstants[0] = *(uint32_t*)&Wcoeff; vertConstants[1] = *(uint32_t*)&viewportScaleX; vertConstants[2] = *(uint32_t*)&viewportScaleY; vertConstants[3] = *(uint32_t*)&Zs; rpi_vkCmdPushConstants(commandBuffer, blitPipelineLayout, VK_SHADER_STAGE_VERTEX_BIT, 0, sizeof(vertConstants), &vertConstants); float mipBias = srcMipLevel; uint32_t fragConstants[1]; fragConstants[0] = *(uint32_t*)&mipBias; rpi_vkCmdPushConstants(commandBuffer, blitPipelineLayout, VK_SHADER_STAGE_FRAGMENT_BIT, 0, sizeof(fragConstants), &fragConstants); rpi_vkCmdDraw(commandBuffer, 6, 1, 0, 0); rpi_vkCmdEndRenderPass(commandBuffer); //free up resources rpi_vkDestroySampler(device, mipSampler, 0); rpi_vkDestroyPipelineLayout(device, blitPipelineLayout, 0); rpi_vkDestroyPipeline(device, blitPipeline, 0); rpi_vkFreeDescriptorSets(device, device->emulDescriptorPool, 1, &blitDescriptorSet); rpi_vkDestroyImageView(device, srcTextureView, 0); rpi_vkDestroyImageView(device, dstTextureView, 0); rpi_vkDestroyRenderPass(device, offscreenRenderPass, 0); rpi_vkDestroyFramebuffer(device, offscreenFramebuffer, 0); } } VKAPI_ATTR void VKAPI_CALL rpi_vkCmdResolveImage( VkCommandBuffer commandBuffer, VkImage srcImage, VkImageLayout srcImageLayout, VkImage dstImage, VkImageLayout dstImageLayout, uint32_t regionCount, const VkImageResolve* pRegions) { //TODO } VKAPI_ATTR void VKAPI_CALL rpi_vkCmdCopyImageToBuffer( VkCommandBuffer commandBuffer, VkImage srcImage, VkImageLayout srcImageLayout, VkBuffer dstBuffer, uint32_t regionCount, const VkBufferImageCopy* pRegions) { //TODO } VKAPI_ATTR void VKAPI_CALL rpi_vkCmdCopyImage( VkCommandBuffer commandBuffer, VkImage srcImage, VkImageLayout srcImageLayout, VkImage dstImage, VkImageLayout dstImageLayout, uint32_t regionCount, const VkImageCopy* pRegions) { rpi_vkCmdBlitImage(commandBuffer, srcImage, srcImageLayout, dstImage, dstImageLayout, regionCount, pRegions, VK_FILTER_NEAREST); } VKAPI_ATTR void VKAPI_CALL rpi_vkCmdCopyBuffer( VkCommandBuffer commandBuffer, VkBuffer srcBuffer, VkBuffer dstBuffer, uint32_t regionCount, const VkBufferCopy* pRegions) { //TODO }