Skip to content

Commit

Permalink
Merge remote-tracking branch 'upstream/develop' into BufferAssetCCON
Browse files Browse the repository at this point in the history
  • Loading branch information
@shrinktofit
shrinktofit committed Aug 1, 2023
2 parents bc880c1 + 1af2ab3 commit 7e39685
Show file tree
Hide file tree
Showing 4 changed files with 235 additions and 16 deletions.
4 changes: 2 additions & 2 deletions .github/workflows/run_test_cases.yml
View file
Original file line number Diff line number Diff line change
Expand Up @@ -301,7 +301,7 @@ jobs:
- Test Platform: PR-Test
- Editor Version: ${process.env.EDITOR_VERSION}
- Task Result: ${process.env.TASK_STATUS}
- Task URL: http://cctest.cocos.org/#/reportview/PR-TEST/${process.env.JOB_ID}/-1
- Task URL: https://cctest.cocos.org/#/reportview/PR-TEST/${process.env.JOB_ID}/-1
- GitHub Action: https://github.com/cocos/cocos-engine/actions/runs/${process.env.GITHUB_RUN_ID}
## Task Details
${process.env.TASK_REPORT}`);
Expand All @@ -317,4 +317,4 @@ jobs:
name: comment.txt
path: |
comment.txt
6 changes: 3 additions & 3 deletions cocos/scene-graph/node.ts
View file
Original file line number Diff line number Diff line change
Expand Up @@ -1138,7 +1138,7 @@ export class Node extends CCObject implements ISchedulable, CustomSerializable {
* node.on(NodeEventType.TOUCH_END, callback, this);
* ```
*/
public on (type: string | NodeEventType, callback: AnyFunction, target?: unknown, useCapture: any = false): void {
public on (type: string | NodeEventType, callback: AnyFunction, target?: unknown, useCapture: boolean = false): void {
switch (type) {
case NodeEventType.TRANSFORM_CHANGED:
this._eventMask |= TRANSFORM_ON;
Expand All @@ -1165,7 +1165,7 @@ export class Node extends CCObject implements ISchedulable, CustomSerializable {
* node.off(NodeEventType.TOUCH_START, callback, this.node);
* ```
*/
public off (type: string, callback?: AnyFunction, target?: unknown, useCapture: any = false): void {
public off (type: string, callback?: AnyFunction, target?: unknown, useCapture: boolean = false): void {
this._eventProcessor.off(type as NodeEventType, callback, target, useCapture);

const hasListeners = this._eventProcessor.hasEventListener(type);
Expand Down Expand Up @@ -1193,7 +1193,7 @@ export class Node extends CCObject implements ISchedulable, CustomSerializable {
* The callback is ignored if it is a duplicate (the callbacks are unique).
* @param target - The target (this object) to invoke the callback, can be null
*/
public once (type: string, callback: AnyFunction, target?: unknown, useCapture?: any): void {
public once (type: string, callback: AnyFunction, target?: unknown, useCapture?: boolean): void {
this._eventProcessor.once(type as NodeEventType, callback, target, useCapture);
}

Expand Down
240 changes: 229 additions & 11 deletions native/cocos/renderer/pipeline/custom/NativePipeline.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -29,6 +29,7 @@
#include "cocos/renderer/pipeline/custom/NativeBuiltinUtils.h"
#include "cocos/renderer/pipeline/custom/NativePipelineTypes.h"
#include "cocos/renderer/pipeline/custom/NativeRenderGraphUtils.h"
#include "cocos/renderer/pipeline/custom/NativeUtils.h"
#include "cocos/renderer/pipeline/custom/RenderGraphGraphs.h"
#include "cocos/renderer/pipeline/custom/RenderingModule.h"
#include "cocos/renderer/pipeline/custom/details/GslUtils.h"
Expand Down Expand Up @@ -205,7 +206,8 @@ uint32_t NativePipeline::addDepthStencil(const ccstd::string &name, gfx::Format
resourceGraph);
}

uint32_t NativePipeline::addResource(const ccstd::string& name, ResourceDimension dimension,
uint32_t NativePipeline::addResource(
const ccstd::string &name, ResourceDimension dimension,
gfx::Format format,
uint32_t width, uint32_t height, uint32_t depth, uint32_t arraySize, uint32_t mipLevels,
gfx::SampleCount sampleCount, ResourceFlags flags, ResourceResidency residency) {
Expand All @@ -232,10 +234,14 @@ uint32_t NativePipeline::addResource(const ccstd::string& name, ResourceDimensio
resourceGraph);
}

void NativePipeline::updateResource(const ccstd::string& name, gfx::Format format,
namespace {

void updateResourceImpl(
ResourceGraph &resourceGraph,
const ccstd::pmr::string &name, gfx::Format format,
uint32_t width, uint32_t height, uint32_t depth, uint32_t arraySize, uint32_t mipLevels, // NOLINT(bugprone-easily-swappable-parameters)
gfx::SampleCount sampleCount) {
auto resID = findVertex(ccstd::pmr::string(name, get_allocator()), resourceGraph);
auto resID = findVertex(name, resourceGraph);
if (resID == ResourceGraph::null_vertex()) {
return;
}
Expand Down Expand Up @@ -266,6 +272,17 @@ void NativePipeline::updateResource(const ccstd::string& name, gfx::Format forma
[](const auto & /*res*/) {});
}

} // namespace

void NativePipeline::updateResource(const ccstd::string &name, gfx::Format format,
uint32_t width, uint32_t height, uint32_t depth, uint32_t arraySize, uint32_t mipLevels, // NOLINT(bugprone-easily-swappable-parameters)
gfx::SampleCount sampleCount) {
updateResourceImpl(
resourceGraph,
ccstd::pmr::string(name, &unsyncPool),
format, width, height, depth, arraySize, mipLevels, sampleCount);
}

// NOLINTNEXTLINE
uint32_t NativePipeline::addStorageTexture(const ccstd::string &name, gfx::Format format, uint32_t width, uint32_t height, ResourceResidency residency) {
ResourceDesc desc{};
Expand All @@ -277,7 +294,7 @@ uint32_t NativePipeline::addStorageTexture(const ccstd::string &name, gfx::Forma
desc.format = format;
desc.sampleCount = gfx::SampleCount::X1;
desc.textureFlags = gfx::TextureFlagBit::NONE;
desc.flags = ResourceFlags::STORAGE | ResourceFlags::SAMPLED | ResourceFlags::TRANSFER_SRC | ResourceFlags::TRANSFER_DST;;
desc.flags = ResourceFlags::STORAGE | ResourceFlags::SAMPLED | ResourceFlags::TRANSFER_SRC | ResourceFlags::TRANSFER_DST;

CC_EXPECTS(residency == ResourceResidency::MANAGED || residency == ResourceResidency::MEMORYLESS);

Expand Down Expand Up @@ -652,7 +669,7 @@ void NativePipeline::addMovePass(const ccstd::vector<MovePair> &movePairs) {
namespace {

void setupGpuDrivenResources(
NativePipeline& ppl, uint32_t cullingID, ResourceGraph& resg, const std::string &hzbName) {
NativePipeline &ppl, uint32_t cullingID, ResourceGraph &resg, const std::string &hzbName) {
ccstd::pmr::string name(resg.get_allocator());
{ // init resource
name = "_GpuInit";
Expand Down Expand Up @@ -729,7 +746,6 @@ void setupGpuDrivenResources(
}
}
if (!hzbName.empty()) {

}
}

Expand Down Expand Up @@ -764,7 +780,8 @@ void NativePipeline::addBuiltinGpuCullingPass(
copyPass.copyPairs.emplace_back(std::move(copyPair));
}

auto copyID = addVertex2(CopyTag{},
auto copyID = addVertex2(
CopyTag{},
std::forward_as_tuple("CopyInitialIndirectBuffer"),
std::forward_as_tuple(),
std::forward_as_tuple(),
Expand All @@ -781,7 +798,7 @@ void NativePipeline::addBuiltinGpuCullingPass(
std::piecewise_construct,
std::forward_as_tuple(drawIndirectBuffer),
std::forward_as_tuple());
auto& view = res.first->second.emplace_back();
auto &view = res.first->second.emplace_back();
view.name = "CCDrawIndirectBuffer";
view.accessType = AccessType::WRITE;
view.shaderStageFlags = gfx::ShaderStageFlagBit::COMPUTE;
Expand All @@ -791,7 +808,7 @@ void NativePipeline::addBuiltinGpuCullingPass(
std::piecewise_construct,
std::forward_as_tuple(drawInstanceBuffer),
std::forward_as_tuple());
auto& view = res.first->second.emplace_back();
auto &view = res.first->second.emplace_back();
view.name = "CCDrawInstanceBuffer";
view.accessType = AccessType::WRITE;
view.shaderStageFlags = gfx::ShaderStageFlagBit::COMPUTE;
Expand All @@ -801,13 +818,14 @@ void NativePipeline::addBuiltinGpuCullingPass(
std::piecewise_construct,
std::forward_as_tuple(visibilityBuffer),
std::forward_as_tuple());
auto& view = res.first->second.emplace_back();
auto &view = res.first->second.emplace_back();
view.name = "CCVisibilityBuffer";
view.accessType = AccessType::WRITE;
view.shaderStageFlags = gfx::ShaderStageFlagBit::COMPUTE;
}

auto computePassID = addVertex2(ComputeTag{},
auto computePassID = addVertex2(
ComputeTag{},
std::forward_as_tuple("Scene"),
std::forward_as_tuple(),
std::forward_as_tuple(),
Expand All @@ -818,9 +836,209 @@ void NativePipeline::addBuiltinGpuCullingPass(
}
}

namespace {

constexpr uint32_t getHalfSize(uint32_t x) noexcept {
x /= 2;
return x ? x : 1;
}

constexpr uint32_t getGroupCount(uint32_t x, uint32_t batch) noexcept { // NOLINT(bugprone-easily-swappable-parameters)
return (x + batch - 1) / batch;
}

} // namespace

void NativePipeline::addBuiltinHzbGenerationPass(
// NOLINTNEXTLINE(bugprone-easily-swappable-parameters)
const std::string &sourceDepthStencilName, const std::string &targetHzbName) {
const auto &lg = programLibrary->layoutGraph;
const auto sourceID = findVertex(ccstd::pmr::string{sourceDepthStencilName, &unsyncPool}, resourceGraph);
if (sourceID == ResourceGraph::null_vertex()) {
return;
}
const auto targetID = findVertex(ccstd::pmr::string{targetHzbName, &unsyncPool}, resourceGraph);
if (targetID == ResourceGraph::null_vertex()) {
return;
}

const auto &targetDesc = get(ResourceGraph::DescTag{}, resourceGraph, targetID);

boost::container::pmr::memory_resource *scratch = &unsyncPool;
ccstd::pmr::string currMipName(scratch);
currMipName.reserve(targetHzbName.size() + 6);
currMipName.append(targetHzbName);
// register all mips
{
CC_EXPECTS(targetDesc.mipLevels && targetDesc.mipLevels != std::numeric_limits<decltype(targetDesc.mipLevels)>::max());
auto desc = targetDesc;
desc.mipLevels = 1;
CC_EXPECTS(desc.depthOrArraySize == 1);
CC_EXPECTS(desc.sampleCount == gfx::SampleCount::X1);
for (uint32_t k = 0; k != targetDesc.mipLevels; ++k) {
currMipName.resize(targetHzbName.size());
CC_ENSURES(currMipName == std::string_view{targetHzbName});
currMipName.append("_Mip");
currMipName.append(std::to_string(k));
auto resID = findVertex(currMipName, resourceGraph);
if (resID == ResourceGraph::null_vertex()) {
resID = addVertex(
ManagedTextureTag{},
std::forward_as_tuple(name.c_str()),
std::forward_as_tuple(),
std::forward_as_tuple(ResourceTraits{ResourceResidency::MANAGED}),
std::forward_as_tuple(),
std::forward_as_tuple(),
std::forward_as_tuple(),
resourceGraph);
} else {
CC_EXPECTS(holds<ManagedTextureTag>(resID, resourceGraph));
updateResourceImpl(
resourceGraph,
currMipName,
gfx::Format::UNKNOWN,
desc.width,
desc.height,
1, 1, 1, gfx::SampleCount::X1);
}
desc.width = getHalfSize(desc.width);
desc.height = getHalfSize(desc.height);
}
}

// add passes
ccstd::pmr::string prevMipName(scratch);
prevMipName.reserve(targetHzbName.size() + 6);
prevMipName.append(std::string_view{targetHzbName});
uint32_t width = targetDesc.width;
uint32_t height = targetDesc.height;
for (uint32_t k = 0; k != targetDesc.mipLevels; ++k) {
// add HZ ComputePass
auto passID = addVertex2(
ComputeTag{},
std::forward_as_tuple("HZPass"),
std::forward_as_tuple("hiz-cs"),
std::forward_as_tuple(),
std::forward_as_tuple(),
std::forward_as_tuple(),
renderGraph);
auto &pass = get(ComputeTag{}, passID, renderGraph);
// source
if (k == 0) {
auto res = pass.computeViews.emplace(
std::piecewise_construct,
std::forward_as_tuple(std::string_view{sourceDepthStencilName}), // source depth stencil
std::forward_as_tuple());
CC_ENSURES(res.second);

auto &view = res.first->second.emplace_back();
view.name = "sourceImage";
view.accessType = AccessType::READ;
view.shaderStageFlags = gfx::ShaderStageFlagBit::COMPUTE;
} else {
prevMipName.resize(targetHzbName.size());
CC_ENSURES(prevMipName == std::string_view{targetHzbName});
prevMipName.append("_Mip");
CC_EXPECTS(k > 0);
prevMipName.append(std::to_string(k - 1)); // previous mip, k - 1

auto res = pass.computeViews.emplace(
std::piecewise_construct,
std::forward_as_tuple(prevMipName), // previous mip
std::forward_as_tuple());
CC_ENSURES(res.second);

auto &view = res.first->second.emplace_back();
view.name = "sourceImage";
view.accessType = AccessType::READ;
view.shaderStageFlags = gfx::ShaderStageFlagBit::COMPUTE;
}
// target
currMipName.resize(targetHzbName.size());
CC_ENSURES(currMipName == std::string_view{targetHzbName});
currMipName.append("_Mip");
currMipName.append(std::to_string(k));
auto res = pass.computeViews.emplace(
std::piecewise_construct,
std::forward_as_tuple(std::string_view{currMipName}),
std::forward_as_tuple());
CC_ENSURES(res.second);

auto &view = res.first->second.emplace_back();
view.name = "targetImage";
view.accessType = AccessType::WRITE;
view.shaderStageFlags = gfx::ShaderStageFlagBit::COMPUTE;

// uniforms
auto &data = get(RenderGraph::DataTag{}, renderGraph, passID);
setVec2Impl(
data, lg, "imageSize",
Vec2{
static_cast<float>(width),
static_cast<float>(height),
});

// add HZ Dispatch
CC_ENSURES(passID != RenderGraph::null_vertex());
auto queueID = addVertex2(
QueueTag{},
std::forward_as_tuple("HZQueue"),
std::forward_as_tuple(),
std::forward_as_tuple(),
std::forward_as_tuple(),
std::forward_as_tuple(),
renderGraph,
passID);

CC_ENSURES(queueID != RenderGraph::null_vertex());
auto dispatchID = addVertex2(
DispatchTag{},
std::forward_as_tuple("HZDispatch"),
std::forward_as_tuple(),
std::forward_as_tuple(),
std::forward_as_tuple(),
std::forward_as_tuple(
IntrusivePtr<Material>{},
0,
getGroupCount(width, 32),
getGroupCount(height, 32),
1),
renderGraph,
queueID);
CC_ENSURES(dispatchID != RenderGraph::null_vertex());

width = getHalfSize(width);
height = getHalfSize(height);
}

// add MovePass
{
MovePass move(renderGraph.get_allocator());
move.movePairs.reserve(targetDesc.mipLevels);
for (uint32_t k = 0; k != targetDesc.mipLevels; ++k) {
currMipName.resize(targetHzbName.size());
CC_ENSURES(currMipName == std::string_view{targetHzbName});
currMipName.append("_Mip");
currMipName.append(std::to_string(k));
MovePair pair(move.get_allocator());
pair.source = currMipName;
pair.target = targetHzbName;
pair.mipLevels = 1;
pair.numSlices = 1;
pair.targetMostDetailedMip = k;
move.movePairs.emplace_back(std::move(pair));
}

auto moveID = addVertex2(
MoveTag{},
std::forward_as_tuple("HZMove"),
std::forward_as_tuple(),
std::forward_as_tuple(),
std::forward_as_tuple(),
std::forward_as_tuple(),
renderGraph);
CC_ENSURES(moveID != RenderGraph::null_vertex());
}
}

void NativePipeline::addCopyPass(const ccstd::vector<CopyPair> &copyPairs) {
Expand Down
1 change: 1 addition & 0 deletions native/cocos/renderer/pipeline/custom/NativeSceneCulling.cpp
View file
Original file line number Diff line number Diff line change
Expand Up @@ -445,6 +445,7 @@ void SceneCulling::clear() noexcept {
sceneQueryIndex.clear();
numCullingQueries = 0;
numRenderQueues = 0;
gpuCullingPassID = 0xFFFFFFFF;
}

} // namespace render
Expand Down

0 comments on commit 7e39685

Please sign in to comment.