Google Git Google Git
Sign in
third-party-mirror/vkd3d-proton/refs/heads/v3.0/./libs/vkd3d/acceleration_structure.c
blob: e97d81c48d81357e24873eb4e62ef7fbcec8a61f [file] [edit]
/*
* Copyright 2021 Hans-Kristian Arntzen for Valve Corporation
*
* This library is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* This library is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with this library; if not, write to the Free Software
* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
*/
#define VKD3D_DBG_CHANNEL VKD3D_DBG_CHANNEL_API
#include "vkd3d_private.h"
#define RT_TRACE TRACE
static VkBuildAccelerationStructureFlagsKHR d3d12_build_flags_to_vk(
D3D12_RAYTRACING_ACCELERATION_STRUCTURE_BUILD_FLAGS flags)
{
VkBuildAccelerationStructureFlagsKHR vk_flags = 0;
if (flags & D3D12_RAYTRACING_ACCELERATION_STRUCTURE_BUILD_FLAG_ALLOW_COMPACTION)
vk_flags |= VK_BUILD_ACCELERATION_STRUCTURE_ALLOW_COMPACTION_BIT_KHR;
if (flags & D3D12_RAYTRACING_ACCELERATION_STRUCTURE_BUILD_FLAG_ALLOW_UPDATE)
vk_flags |= VK_BUILD_ACCELERATION_STRUCTURE_ALLOW_UPDATE_BIT_KHR;
if (flags & D3D12_RAYTRACING_ACCELERATION_STRUCTURE_BUILD_FLAG_MINIMIZE_MEMORY)
vk_flags |= VK_BUILD_ACCELERATION_STRUCTURE_LOW_MEMORY_BIT_KHR;
if (flags & D3D12_RAYTRACING_ACCELERATION_STRUCTURE_BUILD_FLAG_PREFER_FAST_BUILD)
vk_flags |= VK_BUILD_ACCELERATION_STRUCTURE_PREFER_FAST_BUILD_BIT_KHR;
if (flags & D3D12_RAYTRACING_ACCELERATION_STRUCTURE_BUILD_FLAG_PREFER_FAST_TRACE)
vk_flags |= VK_BUILD_ACCELERATION_STRUCTURE_PREFER_FAST_TRACE_BIT_KHR;
if (flags & D3D12_RAYTRACING_ACCELERATION_STRUCTURE_BUILD_FLAG_ALLOW_OMM_LINKAGE_UPDATE)
{
/* D3D12 spec isn't clear on what is allowed to be updated and when */
vk_flags |= VK_BUILD_ACCELERATION_STRUCTURE_ALLOW_OPACITY_MICROMAP_UPDATE_BIT_EXT |
VK_BUILD_ACCELERATION_STRUCTURE_ALLOW_OPACITY_MICROMAP_DATA_UPDATE_BIT_EXT;
}
if (flags & D3D12_RAYTRACING_ACCELERATION_STRUCTURE_BUILD_FLAG_ALLOW_DISABLE_OMMS)
vk_flags |= VK_BUILD_ACCELERATION_STRUCTURE_ALLOW_DISABLE_OPACITY_MICROMAPS_BIT_EXT;
return vk_flags;
}
static VkGeometryFlagsKHR d3d12_geometry_flags_to_vk(D3D12_RAYTRACING_GEOMETRY_FLAGS flags)
{
VkGeometryFlagsKHR vk_flags = 0;
if (flags & D3D12_RAYTRACING_GEOMETRY_FLAG_OPAQUE)
vk_flags |= VK_GEOMETRY_OPAQUE_BIT_KHR;
if (flags & D3D12_RAYTRACING_GEOMETRY_FLAG_NO_DUPLICATE_ANYHIT_INVOCATION)
vk_flags |= VK_GEOMETRY_NO_DUPLICATE_ANY_HIT_INVOCATION_BIT_KHR;
return vk_flags;
}
uint32_t vkd3d_acceleration_structure_get_geometry_count(
const D3D12_BUILD_RAYTRACING_ACCELERATION_STRUCTURE_INPUTS *desc)
{
if (desc->Type != D3D12_RAYTRACING_ACCELERATION_STRUCTURE_TYPE_BOTTOM_LEVEL)
return 1;
else
return desc->NumDescs;
}
static void vkd3d_acceleration_structure_convert_triangles(const struct d3d12_device *device,
const D3D12_RAYTRACING_GEOMETRY_TRIANGLES_DESC *desc,
VkAccelerationStructureGeometryKHR *geometry_info,
uint32_t *primitive_count)
{
VkAccelerationStructureGeometryTrianglesDataKHR *triangles;
geometry_info->geometryType = VK_GEOMETRY_TYPE_TRIANGLES_KHR;
triangles = &geometry_info->geometry.triangles;
triangles->sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_GEOMETRY_TRIANGLES_DATA_KHR;
triangles->indexData.deviceAddress = desc->IndexBuffer;
if (desc->IndexFormat != DXGI_FORMAT_UNKNOWN)
{
if (!desc->IndexBuffer)
WARN("Application is using IndexBuffer = 0 and IndexFormat != UNKNOWN. Likely application bug.\n");
triangles->indexType =
desc->IndexFormat == DXGI_FORMAT_R16_UINT ?
VK_INDEX_TYPE_UINT16 : VK_INDEX_TYPE_UINT32;
*primitive_count = desc->IndexCount / 3;
RT_TRACE(" Indexed : Index count = %u (%u bits)\n",
desc->IndexCount,
triangles->indexType == VK_INDEX_TYPE_UINT16 ? 16 : 32);
RT_TRACE(" Vertex count: %u\n", desc->VertexCount);
RT_TRACE(" IBO VA: %"PRIx64".\n", desc->IndexBuffer);
}
else
{
*primitive_count = desc->VertexCount / 3;
triangles->indexType = VK_INDEX_TYPE_NONE_KHR;
RT_TRACE(" Triangle list : Vertex count: %u\n", desc->VertexCount);
}
triangles->maxVertex = max(1, desc->VertexCount) - 1;
triangles->vertexStride = desc->VertexBuffer.StrideInBytes;
triangles->vertexFormat = vkd3d_internal_get_vk_format(device, desc->VertexFormat);
triangles->vertexData.deviceAddress = desc->VertexBuffer.StartAddress;
triangles->transformData.deviceAddress = desc->Transform3x4;
RT_TRACE(" Transform3x4: %s\n", desc->Transform3x4 ? "on" : "off");
RT_TRACE(" Vertex format: %s\n", debug_dxgi_format(desc->VertexFormat));
RT_TRACE(" VBO VA: %"PRIx64"\n", desc->VertexBuffer.StartAddress);
RT_TRACE(" Vertex stride: %"PRIu64" bytes\n", desc->VertexBuffer.StrideInBytes);
}
bool vkd3d_acceleration_structure_convert_inputs(struct d3d12_device *device,
const D3D12_BUILD_RAYTRACING_ACCELERATION_STRUCTURE_INPUTS *desc,
VkAccelerationStructureBuildGeometryInfoKHR *build_info,
VkAccelerationStructureGeometryKHR *geometry_infos,
VkAccelerationStructureTrianglesOpacityMicromapEXT *omm_infos,
VkAccelerationStructureBuildRangeInfoKHR *range_infos,
uint32_t *primitive_counts)
{
VkAccelerationStructureTrianglesOpacityMicromapEXT *omm;
VkAccelerationStructureGeometryAabbsDataKHR *aabbs;
const D3D12_RAYTRACING_GEOMETRY_DESC *geom_desc;
bool have_triangles, have_aabbs;
uint32_t primitive_count;
unsigned int i;
RT_TRACE("Converting inputs.\n");
RT_TRACE("=====================\n");
memset(build_info, 0, sizeof(*build_info));
build_info->sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_BUILD_GEOMETRY_INFO_KHR;
if (desc->Type == D3D12_RAYTRACING_ACCELERATION_STRUCTURE_TYPE_TOP_LEVEL)
{
build_info->type = VK_ACCELERATION_STRUCTURE_TYPE_TOP_LEVEL_KHR;
RT_TRACE("Top level build.\n");
}
else
{
build_info->type = VK_ACCELERATION_STRUCTURE_TYPE_BOTTOM_LEVEL_KHR;
RT_TRACE("Bottom level build.\n");
}
build_info->flags = d3d12_build_flags_to_vk(desc->Flags);
if (desc->Flags & D3D12_RAYTRACING_ACCELERATION_STRUCTURE_BUILD_FLAG_PERFORM_UPDATE)
{
RT_TRACE("BUILD_FLAG_PERFORM_UPDATE.\n");
build_info->mode = VK_BUILD_ACCELERATION_STRUCTURE_MODE_UPDATE_KHR;
}
else
build_info->mode = VK_BUILD_ACCELERATION_STRUCTURE_MODE_BUILD_KHR;
if (desc->Type == D3D12_RAYTRACING_ACCELERATION_STRUCTURE_TYPE_TOP_LEVEL)
{
memset(geometry_infos, 0, sizeof(*geometry_infos));
geometry_infos[0].sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_GEOMETRY_KHR;
geometry_infos[0].geometryType = VK_GEOMETRY_TYPE_INSTANCES_KHR;
geometry_infos[0].geometry.instances.sType =
VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_GEOMETRY_INSTANCES_DATA_KHR;
geometry_infos[0].geometry.instances.arrayOfPointers =
desc->DescsLayout == D3D12_ELEMENTS_LAYOUT_ARRAY_OF_POINTERS ? VK_TRUE : VK_FALSE;
geometry_infos[0].geometry.instances.data.deviceAddress = desc->InstanceDescs;
if (primitive_counts)
primitive_counts[0] = desc->NumDescs;
if (range_infos)
{
range_infos[0].primitiveCount = desc->NumDescs;
range_infos[0].firstVertex = 0;
range_infos[0].primitiveOffset = 0;
range_infos[0].transformOffset = 0;
}
build_info->geometryCount = 1;
RT_TRACE(" ArrayOfPointers: %u.\n",
desc->DescsLayout == D3D12_ELEMENTS_LAYOUT_ARRAY_OF_POINTERS ? 1 : 0);
RT_TRACE(" NumDescs: %u.\n", desc->NumDescs);
}
else
{
have_triangles = false;
have_aabbs = false;
memset(geometry_infos, 0, sizeof(*geometry_infos) * desc->NumDescs);
if (omm_infos)
memset(omm_infos, 0, sizeof(*omm_infos) * desc->NumDescs);
if (primitive_counts)
memset(primitive_counts, 0, sizeof(*primitive_counts) * desc->NumDescs);
build_info->geometryCount = desc->NumDescs;
for (i = 0; i < desc->NumDescs; i++)
{
geometry_infos[i].sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_GEOMETRY_KHR;
RT_TRACE(" Geom %u:\n", i);
if (desc->DescsLayout == D3D12_ELEMENTS_LAYOUT_ARRAY_OF_POINTERS)
{
geom_desc = desc->ppGeometryDescs[i];
RT_TRACE(" ArrayOfPointers\n");
}
else
{
geom_desc = &desc->pGeometryDescs[i];
RT_TRACE(" PointerToArray\n");
}
geometry_infos[i].flags = d3d12_geometry_flags_to_vk(geom_desc->Flags);
RT_TRACE(" Flags = #%x\n", geom_desc->Flags);
switch (geom_desc->Type)
{
case D3D12_RAYTRACING_GEOMETRY_TYPE_TRIANGLES:
/* Runtime validates this. */
if (have_aabbs)
{
ERR("Cannot mix and match geometry types in a BLAS.\n");
return false;
}
have_triangles = true;
vkd3d_acceleration_structure_convert_triangles(device,
&geom_desc->Triangles, &geometry_infos[i], &primitive_count);
break;
case D3D12_RAYTRACING_GEOMETRY_TYPE_PROCEDURAL_PRIMITIVE_AABBS:
/* Runtime validates this. */
if (have_triangles)
{
ERR("Cannot mix and match geometry types in a BLAS.\n");
return false;
}
have_aabbs = true;
geometry_infos[i].geometryType = VK_GEOMETRY_TYPE_AABBS_KHR;
aabbs = &geometry_infos[i].geometry.aabbs;
aabbs->sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_GEOMETRY_AABBS_DATA_KHR;
aabbs->stride = geom_desc->AABBs.AABBs.StrideInBytes;
aabbs->data.deviceAddress = geom_desc->AABBs.AABBs.StartAddress;
primitive_count = geom_desc->AABBs.AABBCount;
RT_TRACE(" AABB stride: %"PRIu64" bytes\n", geom_desc->AABBs.AABBs.StrideInBytes);
break;
case D3D12_RAYTRACING_GEOMETRY_TYPE_OMM_TRIANGLES:
/* Does the runtime validate this as well? */
if (have_aabbs)
{
ERR("Cannot mix and match geometry types in a BLAS.\n");
return false;
}
have_triangles = true;
if (!d3d12_device_supports_ray_tracing_tier_1_2(device))
{
ERR("Opacity micromap is not supported.\n");
return false;
}
vkd3d_acceleration_structure_convert_triangles(device,
geom_desc->OmmTriangles.pTriangles, &geometry_infos[i], &primitive_count);
geometry_infos[i].geometry.triangles.pNext = omm = &omm_infos[i];
omm->sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_TRIANGLES_OPACITY_MICROMAP_EXT;
omm->pNext = NULL;
switch (geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapIndexFormat)
{
case DXGI_FORMAT_UNKNOWN:
omm->indexType = VK_INDEX_TYPE_NONE_KHR;
break;
case DXGI_FORMAT_R32_UINT:
omm->indexType = VK_INDEX_TYPE_UINT32;
break;
case DXGI_FORMAT_R16_UINT:
omm->indexType = VK_INDEX_TYPE_UINT16;
break;
case DXGI_FORMAT_R8_UINT:
FIXME_ONCE("Using UINT8 as OMM index format is technically out of spec.\n");
omm->indexType = VK_INDEX_TYPE_UINT8;
break;
default:
ERR("Unsupported OMM index format #%x.\n", geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapIndexFormat);
return false;
}
omm->indexBuffer.deviceAddress = geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapIndexBuffer.StartAddress;
omm->indexStride = geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapIndexBuffer.StrideInBytes;
omm->baseTriangle = geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapBaseLocation;
if (geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapArray)
{
omm->micromap = vkd3d_va_map_place_opacity_micromap(
&device->memory_allocator.va_map, device,
geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapArray);
if (omm->micromap == VK_NULL_HANDLE)
ERR("Failed to place OMM at VA 0x%"PRIx64".\n", geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapArray);
}
RT_TRACE(" OMM Index type: %s\n", debug_dxgi_format(geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapIndexFormat));
RT_TRACE(" OMM IBO VA: %"PRIx64"\n", geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapIndexBuffer.StartAddress);
RT_TRACE(" OMM Index stride: %"PRIu64" bytes\n", geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapIndexBuffer.StrideInBytes);
RT_TRACE(" OMM Base: %u\n", geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapBaseLocation);
RT_TRACE(" OMM Micromap VA: %"PRIx64"\n", geom_desc->OmmTriangles.pOmmLinkage->OpacityMicromapArray);
break;
default:
FIXME("Unsupported geometry type %u.\n", geom_desc->Type);
return false;
}
if (primitive_counts)
primitive_counts[i] = primitive_count;
if (range_infos)
{
range_infos[i].primitiveCount = primitive_count;
range_infos[i].firstVertex = 0;
range_infos[i].primitiveOffset = 0;
range_infos[i].transformOffset = 0;
}
RT_TRACE(" Primitive count %u.\n", primitive_count);
}
}
RT_TRACE("=====================\n");
return true;
}
static void vkd3d_acceleration_structure_end_barrier(struct d3d12_command_list *list)
{
/* We resolve the query in TRANSFER, but DXR expects UNORDERED_ACCESS. */
const struct vkd3d_vk_device_procs *vk_procs = &list->device->vk_procs;
VkDependencyInfo dep_info;
VkMemoryBarrier2 barrier;
memset(&barrier, 0, sizeof(barrier));
barrier.sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER_2;
barrier.srcStageMask = VK_PIPELINE_STAGE_2_COPY_BIT;
barrier.srcAccessMask = VK_ACCESS_2_TRANSFER_WRITE_BIT;
barrier.dstStageMask = VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT;
memset(&dep_info, 0, sizeof(dep_info));
dep_info.sType = VK_STRUCTURE_TYPE_DEPENDENCY_INFO;
dep_info.memoryBarrierCount = 1;
dep_info.pMemoryBarriers = &barrier;
VK_CALL(vkCmdPipelineBarrier2(list->cmd.vk_command_buffer, &dep_info));
}
static void vkd3d_acceleration_structure_write_postbuild_info(
struct d3d12_command_list *list,
const D3D12_RAYTRACING_ACCELERATION_STRUCTURE_POSTBUILD_INFO_DESC *desc,
VkDeviceSize desc_offset,
VkAccelerationStructureKHR vk_acceleration_structure)
{
const struct vkd3d_vk_device_procs *vk_procs = &list->device->vk_procs;
const struct vkd3d_unique_resource *resource;
VkQueryPool vk_query_pool;
VkQueryType vk_query_type;
uint32_t vk_query_index;
VkDeviceSize stride;
uint32_t type_index;
VkBuffer vk_buffer;
uint32_t offset;
resource = vkd3d_va_map_deref(&list->device->memory_allocator.va_map, desc->DestBuffer);
if (!resource)
{
ERR("Invalid resource.\n");
return;
}
vk_buffer = resource->vk_buffer;
offset = desc->DestBuffer - resource->va;
offset += desc_offset;
if (desc->InfoType == D3D12_RAYTRACING_ACCELERATION_STRUCTURE_POSTBUILD_INFO_COMPACTED_SIZE)
{
vk_query_type = VK_QUERY_TYPE_ACCELERATION_STRUCTURE_COMPACTED_SIZE_KHR;
type_index = VKD3D_QUERY_TYPE_INDEX_RT_COMPACTED_SIZE;
stride = sizeof(uint64_t);
}
else if (desc->InfoType == D3D12_RAYTRACING_ACCELERATION_STRUCTURE_POSTBUILD_INFO_CURRENT_SIZE &&
list->device->device_info.ray_tracing_maintenance1_features.rayTracingMaintenance1)
{
vk_query_type = VK_QUERY_TYPE_ACCELERATION_STRUCTURE_SIZE_KHR;
type_index = VKD3D_QUERY_TYPE_INDEX_RT_CURRENT_SIZE;
stride = sizeof(uint64_t);
}
else if (desc->InfoType == D3D12_RAYTRACING_ACCELERATION_STRUCTURE_POSTBUILD_INFO_SERIALIZATION)
{
vk_query_type = VK_QUERY_TYPE_ACCELERATION_STRUCTURE_SERIALIZATION_SIZE_KHR;
type_index = VKD3D_QUERY_TYPE_INDEX_RT_SERIALIZE_SIZE;
stride = sizeof(uint64_t);
}
else
{
FIXME("Unsupported InfoType %u.\n", desc->InfoType);
/* TODO: CURRENT_SIZE is something we cannot query in Vulkan, so
* we'll need to keep around a buffer to handle this.
* For now, just clear to 0. */
VK_CALL(vkCmdFillBuffer(list->cmd.vk_command_buffer, vk_buffer, offset,
sizeof(uint64_t), 0));
return;
}
if (!d3d12_command_allocator_allocate_query_from_type_index(list->allocator,
type_index, &vk_query_pool, &vk_query_index))
{
ERR("Failed to allocate query.\n");
return;
}
d3d12_command_list_reset_query(list, vk_query_pool, vk_query_index);
VK_CALL(vkCmdWriteAccelerationStructuresPropertiesKHR(list->cmd.vk_command_buffer,
1, &vk_acceleration_structure, vk_query_type, vk_query_pool, vk_query_index));
VK_CALL(vkCmdCopyQueryPoolResults(list->cmd.vk_command_buffer,
vk_query_pool, vk_query_index, 1,
vk_buffer, offset, stride,
VK_QUERY_RESULT_64_BIT | VK_QUERY_RESULT_WAIT_BIT));
if (desc->InfoType == D3D12_RAYTRACING_ACCELERATION_STRUCTURE_POSTBUILD_INFO_SERIALIZATION)
{
if (list->device->device_info.ray_tracing_maintenance1_features.rayTracingMaintenance1)
{
type_index = VKD3D_QUERY_TYPE_INDEX_RT_SERIALIZE_SIZE_BOTTOM_LEVEL_POINTERS;
if (!d3d12_command_allocator_allocate_query_from_type_index(list->allocator,
type_index, &vk_query_pool, &vk_query_index))
{
ERR("Failed to allocate query.\n");
return;
}
d3d12_command_list_reset_query(list, vk_query_pool, vk_query_index);
VK_CALL(vkCmdWriteAccelerationStructuresPropertiesKHR(list->cmd.vk_command_buffer,
1, &vk_acceleration_structure, VK_QUERY_TYPE_ACCELERATION_STRUCTURE_SERIALIZATION_BOTTOM_LEVEL_POINTERS_KHR,
vk_query_pool, vk_query_index));
VK_CALL(vkCmdCopyQueryPoolResults(list->cmd.vk_command_buffer,
vk_query_pool, vk_query_index, 1,
vk_buffer, offset + sizeof(uint64_t), stride,
VK_QUERY_RESULT_64_BIT | VK_QUERY_RESULT_WAIT_BIT));
}
else
{
FIXME("NumBottomLevelPointers will always return 0.\n");
VK_CALL(vkCmdFillBuffer(list->cmd.vk_command_buffer, vk_buffer, offset + sizeof(uint64_t),
sizeof(uint64_t), 0));
}
}
}
void vkd3d_acceleration_structure_emit_postbuild_info(
struct d3d12_command_list *list,
const D3D12_RAYTRACING_ACCELERATION_STRUCTURE_POSTBUILD_INFO_DESC *desc,
uint32_t count,
const D3D12_GPU_VIRTUAL_ADDRESS *addresses)
{
const struct vkd3d_vk_device_procs *vk_procs = &list->device->vk_procs;
VkAccelerationStructureKHR vk_acceleration_structure;
VkMicromapEXT vk_opacity_micromap;
VkDependencyInfo dep_info;
VkMemoryBarrier2 barrier;
VkDeviceSize stride;
uint32_t i;
/* We resolve the query in TRANSFER, but DXR expects UNORDERED_ACCESS. */
memset(&barrier, 0, sizeof(barrier));
barrier.sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER_2;
barrier.srcStageMask = VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT;
barrier.dstStageMask = VK_PIPELINE_STAGE_2_COPY_BIT;
barrier.dstAccessMask = VK_ACCESS_2_TRANSFER_WRITE_BIT;
memset(&dep_info, 0, sizeof(dep_info));
dep_info.sType = VK_STRUCTURE_TYPE_DEPENDENCY_INFO;
dep_info.memoryBarrierCount = 1;
dep_info.pMemoryBarriers = &barrier;
VK_CALL(vkCmdPipelineBarrier2(list->cmd.vk_command_buffer, &dep_info));
stride = desc->InfoType == D3D12_RAYTRACING_ACCELERATION_STRUCTURE_POSTBUILD_INFO_SERIALIZATION ?
2 * sizeof(uint64_t) : sizeof(uint64_t);
for (i = 0; i < count; i++)
{
if (d3d12_device_supports_ray_tracing_tier_1_2(list->device))
{
vkd3d_va_map_try_read_rtas(&list->device->memory_allocator.va_map,
list->device, addresses[i], &vk_acceleration_structure, &vk_opacity_micromap);
if (vk_acceleration_structure != VK_NULL_HANDLE)
{
vkd3d_acceleration_structure_write_postbuild_info(list, desc, i * stride, vk_acceleration_structure);
continue;
}
else if (vk_opacity_micromap != VK_NULL_HANDLE)
{
vkd3d_opacity_micromap_write_postbuild_info(list, desc, i * stride, vk_opacity_micromap);
continue;
}
else
FIXME("Failed to query existing RTAS for VA 0x%"PRIx64", falling back to placement.\n", addresses[i]);
}
vk_acceleration_structure = vkd3d_va_map_place_acceleration_structure(
&list->device->memory_allocator.va_map, list->device, addresses[i]);
if (vk_acceleration_structure != VK_NULL_HANDLE)
vkd3d_acceleration_structure_write_postbuild_info(list, desc, i * stride, vk_acceleration_structure);
else
ERR("Failed to query acceleration structure for VA 0x%"PRIx64".\n", addresses[i]);
}
vkd3d_acceleration_structure_end_barrier(list);
}
void vkd3d_acceleration_structure_emit_immediate_postbuild_info(
struct d3d12_command_list *list, uint32_t count,
const D3D12_RAYTRACING_ACCELERATION_STRUCTURE_POSTBUILD_INFO_DESC *desc,
VkAccelerationStructureKHR vk_acceleration_structure)
{
/* In D3D12 we are supposed to be able to emit without an explicit barrier,
* but we need to emit them for Vulkan. */
const struct vkd3d_vk_device_procs *vk_procs = &list->device->vk_procs;
VkDependencyInfo dep_info;
VkMemoryBarrier2 barrier;
uint32_t i;
memset(&barrier, 0, sizeof(barrier));
barrier.sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER_2;
barrier.srcStageMask = VK_PIPELINE_STAGE_2_ALL_COMMANDS_BIT;
barrier.srcAccessMask = VK_ACCESS_2_ACCELERATION_STRUCTURE_WRITE_BIT_KHR;
/* The query accesses STRUCTURE_READ_BIT in BUILD_BIT stage. */
barrier.dstStageMask = VK_PIPELINE_STAGE_2_ACCELERATION_STRUCTURE_BUILD_BIT_KHR | VK_PIPELINE_STAGE_2_COPY_BIT;
barrier.dstAccessMask = VK_ACCESS_2_ACCELERATION_STRUCTURE_READ_BIT_KHR | VK_ACCESS_2_TRANSFER_WRITE_BIT;
/* Writing to the result buffer is supposed to happen in UNORDERED_ACCESS on DXR for
* some bizarre reason, so we have to satisfy a transfer barrier.
* Have to basically do a full stall to make this work ... */
memset(&dep_info, 0, sizeof(dep_info));
dep_info.sType = VK_STRUCTURE_TYPE_DEPENDENCY_INFO;
dep_info.memoryBarrierCount = 1;
dep_info.pMemoryBarriers = &barrier;
VK_CALL(vkCmdPipelineBarrier2(list->cmd.vk_command_buffer, &dep_info));
/* Could optimize a bit by batching more aggressively, but no idea if it's going to help in practice. */
for (i = 0; i < count; i++)
vkd3d_acceleration_structure_write_postbuild_info(list, &desc[i], 0, vk_acceleration_structure);
vkd3d_acceleration_structure_end_barrier(list);
}
static bool convert_copy_mode(
D3D12_RAYTRACING_ACCELERATION_STRUCTURE_COPY_MODE mode,
VkCopyAccelerationStructureModeKHR *vk_mode)
{
switch (mode)
{
case D3D12_RAYTRACING_ACCELERATION_STRUCTURE_COPY_MODE_CLONE:
*vk_mode = VK_COPY_ACCELERATION_STRUCTURE_MODE_CLONE_KHR;
return true;
case D3D12_RAYTRACING_ACCELERATION_STRUCTURE_COPY_MODE_COMPACT:
*vk_mode = VK_COPY_ACCELERATION_STRUCTURE_MODE_COMPACT_KHR;
return true;
default:
FIXME("Unsupported RTAS copy mode #%x.\n", mode);
return false;
}
}
void vkd3d_acceleration_structure_copy(
struct d3d12_command_list *list,
D3D12_GPU_VIRTUAL_ADDRESS dst, VkAccelerationStructureKHR src_as,
D3D12_RAYTRACING_ACCELERATION_STRUCTURE_COPY_MODE mode)
{
const struct vkd3d_vk_device_procs *vk_procs = &list->device->vk_procs;
VkCopyAccelerationStructureInfoKHR info;
VkAccelerationStructureKHR dst_as;
dst_as = vkd3d_va_map_place_acceleration_structure(&list->device->memory_allocator.va_map, list->device, dst);
if (dst_as == VK_NULL_HANDLE)
{
ERR("Invalid dst address #%"PRIx64" for RTAS copy.\n", dst);
return;
}
info.sType = VK_STRUCTURE_TYPE_COPY_ACCELERATION_STRUCTURE_INFO_KHR;
info.pNext = NULL;
info.dst = dst_as;
info.src = src_as;
if (convert_copy_mode(mode, &info.mode))
VK_CALL(vkCmdCopyAccelerationStructureKHR(list->cmd.vk_command_buffer, &info));
}
struct vkd3d_empty_rtas_build_info
{
VkAccelerationStructureBuildGeometryInfoKHR build_info;
VkAccelerationStructureBuildSizesInfoKHR size_info;
VkAccelerationStructureGeometryKHR geom;
};
static void vkd3d_setup_empty_rtas_build(
struct d3d12_device *device,
struct vkd3d_empty_rtas_build_info *info)
{
const struct vkd3d_vk_device_procs *vk_procs = &device->vk_procs;
uint32_t count = 0;
/* Build an empty RTAS. */
memset(info, 0, sizeof(*info));
info->build_info.sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_BUILD_GEOMETRY_INFO_KHR;
info->size_info.sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_BUILD_SIZES_INFO_KHR;
info->geom.sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_GEOMETRY_KHR;
info->geom.geometry.instances.sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_GEOMETRY_INSTANCES_DATA_KHR;
info->build_info.mode = VK_BUILD_ACCELERATION_STRUCTURE_MODE_BUILD_KHR;
info->build_info.type = VK_ACCELERATION_STRUCTURE_TYPE_TOP_LEVEL_KHR;
info->build_info.geometryCount = 1;
info->build_info.pGeometries = &info->geom;
info->geom.geometryType = VK_GEOMETRY_TYPE_INSTANCES_KHR;
VK_CALL(vkGetAccelerationStructureBuildSizesKHR(device->vk_device,
VK_ACCELERATION_STRUCTURE_BUILD_TYPE_DEVICE_KHR,
&info->build_info, &count, &info->size_info));
}
void vkd3d_build_null_rtas_va(struct d3d12_device *device, VkCommandBuffer vk_cmd_buffer)
{
const struct vkd3d_vk_device_procs *vk_procs = &device->vk_procs;
const VkAccelerationStructureBuildRangeInfoKHR *ptr_range;
VkAccelerationStructureBuildRangeInfoKHR range;
struct vkd3d_empty_rtas_build_info info;
memset(&range, 0, sizeof(range));
vkd3d_setup_empty_rtas_build(device, &info);
info.build_info.scratchData.deviceAddress =
vkd3d_get_buffer_device_address(device, device->null_rtas_allocation.buffer) +
align64(info.size_info.accelerationStructureSize,
device->device_info.acceleration_structure_properties.minAccelerationStructureScratchOffsetAlignment);
info.build_info.dstAccelerationStructure = device->null_rtas_allocation.rtas;
ptr_range = &range;
memset(&range, 0, sizeof(range));
VK_CALL(vkCmdBuildAccelerationStructuresKHR(vk_cmd_buffer, 1, &info.build_info, &ptr_range));
}
D3D12_GPU_VIRTUAL_ADDRESS vkd3d_get_null_rtas_va(struct d3d12_device *device)
{
const struct vkd3d_vk_device_procs *vk_procs = &device->vk_procs;
VkAccelerationStructureDeviceAddressInfoKHR get_addr;
VkAccelerationStructureCreateInfoKHR rtas_info;
struct vkd3d_allocate_memory_info alloc_info;
struct vkd3d_empty_rtas_build_info info;
VkDeviceAddress va;
VkDeviceSize size;
va = vkd3d_atomic_uint64_load_explicit(&device->null_rtas_allocation.va, vkd3d_memory_order_acquire);
if (va || !d3d12_device_supports_ray_tracing_tier_1_0(device))
goto end;
/* Spinlock is weird here, but it avoids lots of ugly init code
* and we only expect to hit this path once. */
spinlock_acquire(&device->null_rtas_allocation.lock);
if ((va = device->null_rtas_allocation.va))
goto end_unlock;
vkd3d_setup_empty_rtas_build(device, &info);
memset(&alloc_info, 0, sizeof(alloc_info));
size = info.size_info.accelerationStructureSize;
size = align64(size, device->device_info.acceleration_structure_properties.minAccelerationStructureScratchOffsetAlignment);
size += info.size_info.buildScratchSize;
if (FAILED(vkd3d_create_buffer_explicit_usage(device,
VK_BUFFER_USAGE_ACCELERATION_STRUCTURE_STORAGE_BIT_KHR | VK_BUFFER_USAGE_STORAGE_BUFFER_BIT |
VK_BUFFER_USAGE_SHADER_DEVICE_ADDRESS_BIT,
size, "null-rtas", &device->null_rtas_allocation.buffer)))
{
ERR("Failed to create null RTAS VkBuffer.\n");
goto end_unlock;
}
if (FAILED(vkd3d_allocate_internal_buffer_memory(device, device->null_rtas_allocation.buffer,
VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT, &device->null_rtas_allocation.alloc)))
{
ERR("Failed to allocate empty RTAS memory.\n");
goto end_unlock;
}
memset(&rtas_info, 0, sizeof(rtas_info));
rtas_info.sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_CREATE_INFO_KHR;
rtas_info.buffer = device->null_rtas_allocation.buffer;
rtas_info.type = VK_ACCELERATION_STRUCTURE_TYPE_TOP_LEVEL_KHR;
rtas_info.size = info.size_info.accelerationStructureSize;
if (VK_CALL(vkCreateAccelerationStructureKHR(device->vk_device, &rtas_info, NULL, &device->null_rtas_allocation.rtas)) != VK_SUCCESS)
{
ERR("Failed to create RTAS.\n");
goto end_unlock;
}
memset(&get_addr, 0, sizeof(get_addr));
get_addr.sType = VK_STRUCTURE_TYPE_ACCELERATION_STRUCTURE_DEVICE_ADDRESS_INFO_KHR;
get_addr.accelerationStructure = device->null_rtas_allocation.rtas;
va = VK_CALL(vkGetAccelerationStructureDeviceAddressKHR(device->vk_device, &get_addr));
/* Make sure that we flush any write to the memory init system before we broadcast the VA
* since a different thread could read the VA and submit work to GPU before we do the init work. */
vkd3d_memory_transfer_queue_build_empty_rtas(&device->memory_transfers);
vkd3d_atomic_uint64_store_explicit(&device->null_rtas_allocation.va, va, vkd3d_memory_order_release);
end_unlock:
spinlock_release(&device->null_rtas_allocation.lock);
end:
return va;
}