OpenCL框架与示例
下面的图简单说明了OpenCL的编程框架，图是用的GPU，其他类似；
名词的概念：
Platform (平台)：主机加上OpenCL框架管理下的若干设备构成了这个平台，通过这个平台，应用程序可以与设备共享资源并在设备上执行kernel。实际使用中基本上一个厂商对应一个Platform，比如Intel, AMD都是这样。
Device（设备）：官方的解释是计算单元（Compute Units）的集合。举例来说，GPU是典型的device。Intel和AMD的多核CPU也提供OpenCL接口，所以也可以作为Device。
Context（上下文）：OpenCL的Platform上共享和使用资源的环境，包括kernel、device、memory objects、command queue等。使用中一般一个Platform对应一个Context。
Program：OpenCL程序，由kernel函数、其他函数和声明等组成。
Kernel（核函数）：可以从主机端调用，运行在设备端的函数。
Memory Object（内存对象）：在主机和设备之间传递数据的对象，一般映射到OpenCL程序中的global memory。有两种具体的类型：Buffer Object（缓存对象）和Image Object（图像对象）。
Command Queue（指令队列）：在指定设备上管理多个指令（Command）。队列里指令执行可以顺序也可以乱序。一个设备可以对应多个指令队列。
NDRange：主机端运行设备端kernel函数的主要接口。实际上还有其他的，NDRange是非常常见的，用于分组运算，以后具体用到的时候就知道区别了。
Host端来看，OpenCL的组要执行流程是这样的：
内存模型
最后写一下Opencl的内存模型，看下面的示意图：
用核函数中的内存变量来简单地解释：用clCreateBuffer 创建、用clSetKernelArg 传递的数据在global memory 和constant memory中；核函数中的寄存器变量在private memory 中；核函数的内部变量、缓存等，在local memory 中。图例中可以看到Device 并不直接访问global memory，而是通过Cache 来访问。可以想象当同时运行的work-item，使用的内存都在同一块cache 中，则内存吞吐的效率最高。对应到work group 中，就是在程序设计上尽量使同一个work group 中的work item 操作连续的内存，以提高访存效率。
TVM OpenCL示例
/*
* Licensed to the Apache Software Foundation (ASF) under one
* or more contributor license agreements. See the NOTICE file
* distributed with this work for additional information
* regarding copyright ownership. The ASF licenses this file
* to you under the Apache License, Version 2.0 (the
* "License"); you may not use this file except in compliance
* with the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing,
* software distributed under the License is distributed on an
* "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
* KIND, either express or implied. See the License for the
* specific language governing permissions and limitations
* under the License.
*/
/*!
* \file opencl_module.cc
*/
#include "opencl_module.h"
#include <dmlc/memory_io.h>
#include <tvm/runtime/registry.h>
#include <string>
#include <unordered_map>
#include <vector>
#include "../source_utils.h"
#include "opencl_common.h"
namespace tvm {
namespace runtime {
class OpenCLWrappedFunc {
public:
// initialize the OpenCL function.
void Init(OpenCLModuleNode* m, ObjectPtr<Object> sptr, OpenCLModuleNode::KTRefEntry entry,
std::string func_name, std::vector<size_t> arg_size,
const std::vector<std::string>& launch_param_tags) {
w_ = m->GetGlobalWorkspace();
m_ = m;
sptr_ = sptr;
entry_ = entry;
func_name_ = func_name;
arg_size_ = arg_size;
launch_param_config_.Init(arg_size.size(), launch_param_tags);
}
// invoke the function with void arguments
void operator()(TVMArgs args, TVMRetValue* rv, void** void_args) const {
ICHECK(w_->context != nullptr) << "No OpenCL device";
cl::OpenCLThreadEntry* t = w_->GetThreadEntry();
// get the kernel from thread local kernel table.
if (entry_.kernel_id >= t->kernel_table.size()) {
t->kernel_table.resize(entry_.kernel_id + 1);
}
const auto& e = t->kernel_table[entry_.kernel_id];
cl_kernel kernel = e.kernel;
if (kernel == nullptr || e.version != entry_.version) {
kernel = m_->InstallKernel(w_, t, func_name_, entry_);
}
// setup arguments.
for (cl_uint i = 0; i < arg_size_.size(); ++i) {
void* arg = nullptr;
if (args.type_codes[i] == DLDataTypeCode::kDLOpaqueHandle) {
arg = static_cast<cl::BufferDescriptor*>(void_args[i])->buffer;
} else {
arg = void_args[i];
}
OPENCL_CALL(clSetKernelArg(kernel, i, arg_size_[i], arg));
}
cl_command_queue queue = w_->GetQueue(t->device);
ThreadWorkLoad wl = launch_param_config_.Extract(args);
cl_uint work_dim = static_cast<cl_uint>(launch_param_config_.work_dim());
for (cl_uint i = 0; i < work_dim; ++i) {
wl.work_size[i] *= wl.work_size[i + 3];
}
// launch kernel
OPENCL_CALL(clEnqueueNDRangeKernel(queue, kernel, work_dim, nullptr, wl.work_size,
wl.work_size + 3, 0, nullptr, nullptr));
}
private:
// global workspace.
cl::OpenCLWorkspace* w_;
// The module
OpenCLModuleNode* m_;
// resource handle
ObjectPtr<Object> sptr_;
// global kernel id in the kernel table.
OpenCLModuleNode::KTRefEntry entry_;
// The name of the function.
std::string func_name_;
// convert code for void argument
std::vector<size_t> arg_size_;
// launch parameters config
LaunchParamConfig launch_param_config_;
};
OpenCLModuleNode::~OpenCLModuleNode() {
{
// free the kernel ids in global table.
std::lock_guard<std::mutex> lock(workspace_->mu);
for (auto& kv : kid_map_) {
workspace_->free_kernel_ids.push_back(kv.second.kernel_id);
}
}
// free the kernels
for (cl_kernel k : kernels_) {
OPENCL_CALL(clReleaseKernel(k));
}
// free the programs
for (auto& kv : programs_) {
for (auto& program : kv.second) {
if (program) {
OPENCL_CALL(clReleaseProgram(program));
}
}
}
}
cl::OpenCLWorkspace* OpenCLModuleNode::GetGlobalWorkspace() {
return cl::OpenCLWorkspace::Global();
}
PackedFunc OpenCLModuleNode::GetFunction(const std::string& name,
const ObjectPtr<Object>& sptr_to_self) {
ICHECK_EQ(sptr_to_self.get(), this);
ICHECK_NE(name, symbol::tvm_module_main) << "Device function do not have main";
auto it = fmap_.find(name);
if (it == fmap_.end()) return PackedFunc();
const FunctionInfo& info = it->second;
OpenCLWrappedFunc f;
std::vector<size_t> arg_size(info.arg_types.size());
for (size_t i = 0; i < info.arg_types.size(); ++i) {
DLDataType t = info.arg_types[i];
ICHECK_EQ(t.lanes, 1U);
if (t.code == kTVMOpaqueHandle) {
// specially store pointer type size in OpenCL driver
arg_size[i] = sizeof(void*);
} else {
uint32_t bits = t.bits;
ICHECK_EQ(bits % 8, 0U);
arg_size[i] = bits / 8;
}
}
// initialize the wrapped func.
f.Init(this, sptr_to_self, kid_map_.at(name), name, arg_size, info.launch_param_tags);
return PackFuncVoidAddr(f, info.arg_types);
}
void OpenCLModuleNode::SaveToFile(const std::string& file_name, const std::string& format) {
std::string fmt = GetFileFormat(file_name, format);
ICHECK_EQ(fmt, fmt_) << "Can only save to format=" << fmt_;
std::string meta_file = GetMetaFilePath(file_name);
SaveMetaDataToFile(meta_file, fmap_);
SaveBinaryToFile(file_name, data_);
}
void OpenCLModuleNode::SaveToBinary(dmlc::Stream* stream) {
stream->Write(fmt_);
stream->Write(fmap_);
stream->Write(data_);
}
std::string OpenCLModuleNode::GetSource(const std::string& format) {
if (format == fmt_) return data_;
if (fmt_ == "cl") {
return data_;
} else {
return source_;
}
}
void OpenCLModuleNode::Init() {
workspace_ = GetGlobalWorkspace();
workspace_->Init();
// initialize the kernel id, need to lock global table.
std::lock_guard<std::mutex> lock(workspace_->mu);
for (const auto& kv : fmap_) {
const std::string& key = kv.first;
KTRefEntry e;
if (workspace_->free_kernel_ids.size() != 0) {
e.kernel_id = workspace_->free_kernel_ids.back();
workspace_->free_kernel_ids.pop_back();
} else {
e.kernel_id = workspace_->num_registered_kernels++;
}
e.version = workspace_->timestamp++;
kid_map_[key] = e;
}
// split into source artifacts for each kernel
parsed_kernels_ = SplitKernels(GetSource("cl"));
ICHECK(!parsed_kernels_.empty()) << "The OpenCL module expects a kernel delimited "
<< "source from code generation, but no kernel "
<< "delimiter was found.";
ICHECK_EQ(fmap_.size(), parsed_kernels_.size())
<< "The number of parsed kernel sources does not match the number of kernel functions";
// zero initialize cl_program pointers for each device kernel
for (auto& kv : parsed_kernels_) {
programs_.insert({kv.first, std::vector<cl_program>(workspace_->devices.size(), nullptr)});
}
}
cl_kernel OpenCLModuleNode::InstallKernel(cl::OpenCLWorkspace* w, cl::OpenCLThreadEntry* t,
const std::string& func_name, const KTRefEntry& e) {
std::lock_guard<std::mutex> lock(build_lock_);
int device_id = t->device.device_id;
if (programs_[func_name][device_id] == nullptr) {
// create program
if (fmt_ == "cl") {
const char* s = parsed_kernels_[func_name].c_str();
size_t len = parsed_kernels_[func_name].length();
cl_int err;
programs_[func_name][device_id] = clCreateProgramWithSource(w->context, 1, &s, &len, &err);
OPENCL_CHECK_ERROR(err);
} else if (fmt_ == "xclbin" || fmt_ == "awsxclbin" || fmt_ == "aocx") {
const unsigned char* s = (const unsigned char*)data_.c_str();
size_t len = data_.length();
cl_int err;
cl_device_id dev = w->devices[device_id];
programs_[func_name][device_id] =
clCreateProgramWithBinary(w->context, 1, &dev, &len, &s, NULL, &err);
OPENCL_CHECK_ERROR(err);
} else {
LOG(FATAL) << "Unknown OpenCL format " << fmt_;
}
// build program
cl_int err;
cl_device_id dev = w->devices[device_id];
err = clBuildProgram(programs_[func_name][device_id], 1, &dev, nullptr, nullptr, nullptr);
if (err != CL_SUCCESS) {
size_t len;
std::string log;
clGetProgramBuildInfo(programs_[func_name][device_id], dev, CL_PROGRAM_BUILD_LOG, 0, nullptr,
&len);
log.resize(len);
clGetProgramBuildInfo(programs_[func_name][device_id], dev, CL_PROGRAM_BUILD_LOG, len,
&log[0], nullptr);
LOG(FATAL) << "OpenCL build error for device=" << dev << "\n" << log;
}
}
// build kernel
cl_int err;
cl_kernel kernel = clCreateKernel(programs_[func_name][device_id], func_name.c_str(), &err);
OPENCL_CHECK_ERROR(err);
t->kernel_table[e.kernel_id].kernel = kernel;
t->kernel_table[e.kernel_id].version = e.version;
kernels_.push_back(kernel);
return kernel;
}
Module OpenCLModuleCreate(std::string data, std::string fmt,
std::unordered_map<std::string, FunctionInfo> fmap, std::string source) {
auto n = make_object<OpenCLModuleNode>(data, fmt, fmap, source);
n->Init();
return Module(n);
}
// Load module from module.
Module OpenCLModuleLoadFile(const std::string& file_name, const std::string& format) {
std::string data;
std::unordered_map<std::string, FunctionInfo> fmap;
std::string fmt = GetFileFormat(file_name, format);
std::string meta_file = GetMetaFilePath(file_name);
LoadBinaryFromFile(file_name, &data);
LoadMetaDataFromFile(meta_file, &fmap);
return OpenCLModuleCreate(data, fmt, fmap, std::string());
}
Module OpenCLModuleLoadBinary(void* strm) {
dmlc::Stream* stream = static_cast<dmlc::Stream*>(strm);
std::string data;
std::unordered_map<std::string, FunctionInfo> fmap;
std::string fmt;
stream->Read(&fmt);
stream->Read(&fmap);
stream->Read(&data);
return OpenCLModuleCreate(data, fmt, fmap, std::string());
}
TVM_REGISTER_GLOBAL("runtime.module.loadfile_cl").set_body_typed(OpenCLModuleLoadFile);
TVM_REGISTER_GLOBAL("runtime.module.loadfile_clbin").set_body_typed(OpenCLModuleLoadFile);
TVM_REGISTER_GLOBAL("runtime.module.loadbinary_opencl").set_body_typed(OpenCLModuleLoadBinary);
} // namespace runtime
} // namespace tvm
参考链接：
参考链接：https://blog.csdn.net/xbinworld/article/details/46494275
https://github.com/apache/tvm/blob/main/src/runtime/opencl/opencl_module.cc