EVENGINE_API_WORLD public API. 更多...

#include <Sequence.h>

类
struct	Impl

Public 成员函数
	Sequence ()
	Sequence.

	~Sequence ()
	Sequence.

	Sequence (const Sequence &)=delete

Sequence &	operator= (const Sequence &)=delete

bool	isAvailable () const
	True when available.

void	begin ()
	Begins begin.

void	recordUpload (GpuBuffer dst, const void src, uint64_t nbytes, uint64_t dstOffset=0)
	Record upload.

void	recordDownload (GpuBuffer src, GpuBuffer staging, uint64_t nbytes, uint64_t srcOffset=0)
	Record download.

void	recordDispatch (ComputeShader *shader, int groupsX, int groupsY=1, int groupsZ=1)
	Record dispatch.

void	submit ()
	Submit.

SequenceStatus	submitAsync ()
	Submit without waiting for GPU completion. @lifetime Buffers and shaders referenced by recorded work must remain alive until poll() or wait() returns Complete/Failed. begin() rejects pending work.

SequenceStatus	poll ()
	Non-blocking completion query; also retires completed backend resources.

SequenceStatus	wait ()
	Wait for pending work and retire its backend resources.

SequenceStatus	getStatus () const
	Current lifecycle state without driving backend event processing.

std::string	getStatusName () const
	Stable script/debug name for getStatus().

详细描述

EVENGINE_API_WORLD public API.

Kompute-style GPU command sequence.

Records storage-buffer transfers and compute dispatches into a single command buffer, then submits everything once. This turns an inference graph (placeholder uploads -> N fused kernel dispatches -> output download) from N+2 separate record/submit/wait round trips into one GPU submission.

Typical usage:

seq->begin();
seq->recordUpload(inputBuffer, inputData, nbytes);
seq->recordDispatch(shaderA, groupsA);   // shader bindings/push
seq->recordDispatch(shaderB, groupsB);   // constants already set
seq->recordDownload(outputBuffer, staging, nbytes);
seq->submit();                            // one submit, waits once
// or submitAsync() + poll()/wait() to overlap unrelated CPU work
staging->downloadBytes(dst, nbytes);      // host-visible memcpy

A Sequence is reusable after synchronous submit(), or after an asynchronous submission reaches Complete/Failed. Shaders keep their bindings between records; pending work retains no C++ owners, so callers must keep referenced shaders and buffers alive until completion.

在文件 Sequence.h 第 41 行定义.

构造及析构函数说明

◆ Sequence() [1/2]

eve::gpgpu::Sequence::Sequence ( )

Sequence.

在文件 Sequence.cpp 第 60 行定义.

◆ ~Sequence()

eve::gpgpu::Sequence::~Sequence ( )

Sequence.

在文件 Sequence.cpp 第 61 行定义.

引用了 eve::gpgpu::Sequence::Impl::v , 以及 eve::gpgpu::vulkanSequenceDestroy().

◆ Sequence() [2/2]

eve::gpgpu::Sequence::Sequence ( const Sequence & )

delete

成员函数说明

◆ begin()

void eve::gpgpu::Sequence::begin ( )

Begins begin.

Start recording. Safe to call again after submit() to reuse.

在文件 Sequence.cpp 第 75 行定义.

引用了 eve::gpgpu::Sequence::Impl::v , 以及 eve::gpgpu::vulkanSequenceBegin().

被这些函数引用 eve::tensor::GpuProgram::run().

◆ getStatus()

SequenceStatus eve::gpgpu::Sequence::getStatus ( ) const

Current lifecycle state without driving backend event processing.

在文件 Sequence.cpp 第 97 行定义.

引用了 eve::gpgpu::Sequence::Impl::v , 以及 eve::gpgpu::vulkanSequenceStatus().

被这些函数引用 getStatusName().

◆ getStatusName()

std::string eve::gpgpu::Sequence::getStatusName ( ) const

Stable script/debug name for getStatus().

在文件 Sequence.cpp 第 105 行定义.

引用了 eve::gpgpu::Complete, eve::gpgpu::Failed, getStatus(), eve::gpgpu::Idle, eve::gpgpu::Recording , 以及 eve::gpgpu::Submitted.

◆ isAvailable()

bool eve::gpgpu::Sequence::isAvailable ( ) const

True when available.

True when the active Vulkan or WebGPU backend supports command recording.

在文件 Sequence.cpp 第 66 行定义.

引用了 eve::gpgpu::VulkanSequence::ensureReady(), eve::gpgpu::VulkanSequence::ready() , 以及 eve::gpgpu::Sequence::Impl::v.

被这些函数引用 eve::physics::ClothGPU::ClothGPU().

◆ operator=()

Sequence & eve::gpgpu::Sequence::operator= ( const Sequence & )

delete

◆ poll()

SequenceStatus eve::gpgpu::Sequence::poll ( )

Non-blocking completion query; also retires completed backend resources.

在文件 Sequence.cpp 第 95 行定义.

引用了 eve::gpgpu::Sequence::Impl::v , 以及 eve::gpgpu::vulkanSequencePoll().

◆ recordDispatch()

void eve::gpgpu::Sequence::recordDispatch	(	ComputeShader *	shader,
		int	groupsX,
		int	groupsY = `1`,
		int	groupsZ = `1`
	)

Record dispatch.

Record a compute dispatch using the shader's current bindings/push constants.

在文件 Sequence.cpp 第 87 行定义.

引用了 shader, eve::gpgpu::Sequence::Impl::v , 以及 eve::gpgpu::vulkanSequenceRecordDispatch().

被这些函数引用 eve::tensor::GpuProgram::Impl::recordGroup().

◆ recordDownload()

void eve::gpgpu::Sequence::recordDownload	(	GpuBuffer *	src,
		GpuBuffer *	staging,
		uint64_t	nbytes,
		uint64_t	srcOffset = `0`
	)

Record download.

Record a device->host copy into staging (a "staging" usage buffer). Read the staging buffer after submit().

在文件 Sequence.cpp 第 82 行定义.

引用了 eve::gpgpu::Sequence::Impl::v , 以及 eve::gpgpu::vulkanSequenceRecordDownload().

被这些函数引用 eve::tensor::GpuProgram::run().

◆ recordUpload()

void eve::gpgpu::Sequence::recordUpload	(	GpuBuffer *	dst,
		const void *	src,
		uint64_t	nbytes,
		uint64_t	dstOffset = `0`
	)

Record upload.

Record a host->device copy. src is memcpy'd into an internal host-visible staging buffer immediately, so it only needs to stay valid until this call returns.

在文件 Sequence.cpp 第 77 行定义.

引用了 eve::gpgpu::Sequence::Impl::v , 以及 eve::gpgpu::vulkanSequenceRecordUpload().

被这些函数引用 eve::tensor::GpuProgram::run().

◆ submit()

void eve::gpgpu::Sequence::submit ( )

Submit.

End recording, submit the whole sequence once, and wait for completion.

在文件 Sequence.cpp 第 92 行定义.

引用了 eve::gpgpu::Sequence::Impl::v , 以及 eve::gpgpu::vulkanSequenceSubmit().

被这些函数引用 eve::tensor::GpuProgram::run().

◆ submitAsync()

SequenceStatus eve::gpgpu::Sequence::submitAsync ( )

Submit without waiting for GPU completion. @lifetime Buffers and shaders referenced by recorded work must remain alive until poll() or wait() returns Complete/Failed. begin() rejects pending work.

在文件 Sequence.cpp 第 94 行定义.

引用了 eve::gpgpu::Sequence::Impl::v , 以及 eve::gpgpu::vulkanSequenceSubmitAsync().

◆ wait()

SequenceStatus eve::gpgpu::Sequence::wait ( )

Wait for pending work and retire its backend resources.

在文件 Sequence.cpp 第 96 行定义.

引用了 eve::gpgpu::Sequence::Impl::v , 以及 eve::gpgpu::vulkanSequenceWait().

该类的文档由以下文件生成:

src/modules/gpgpu/Sequence.h
src/modules/gpgpu/Sequence.cpp

类

Public 成员函数

详细描述

构造及析构函数说明

◆ Sequence() [1/2]

◆ ~Sequence()

◆ Sequence() [2/2]

成员函数说明

◆ begin()

◆ getStatus()

◆ getStatusName()

◆ isAvailable()

◆ operator=()

◆ poll()

◆ recordDispatch()

◆ recordDownload()

◆ recordUpload()

◆ submit()

◆ submitAsync()

◆ wait()