Unigine::CPUShader Class

Header:

#include <UnigineThread.h>

This class is used to create a custom CPU shader, for example, to implement multi-threaded mesh cluster update. Your custom CPU-shader must be inherited from this class. The corresponding sample is included in the SDK(/samples/Api/Systems/CPUShader/). Engine::swap() always waits for CPU shaders, therefore they cannot be used as tasks. CPU shader is updated with the current framerate of the application.

Here is an example of implementing a custom CPU shader:

Source code (C++)

struct UpdateClustersCPUShader: public CPUShader
{
	UpdateClustersCPUShader() {}
	virtual ~UpdateClustersCPUShader() 
	{ 
		// wait until all asynchronous operations are completed
		wait(); 
	}

	Vector<AsyncCluster> clusters;

	volatile int counter{0};
	
	// overriding the process method to performa our calculations
	void process(int thread_num, int threads_count) override
	{
		UNIGINE_PROFILER_FUNCTION;

		while (true)
		{
			int num = AtomicAdd(&counter, 1);
			if (num >= clusters.size())
				break;
			clusters[num].update();
		}
	}

	void run()
	{
		for (auto &c : clusters)
			c.swap();

		counter = 0;
		
		// running shader code in asynchronous mode
		runAsync();
	}
};

Methods:

Name	Description View: Sort by:
*create*	Default CPUShader class constructor.
runSync	Runs CPU shader code synchronously.
runAsync	Runs CPU shader code asynchronously.
wait	Waits for running asynchronous shader code execution.
isRunning	Returns a value indicating if the CPU shader code is currently executed.
getNumThreads	Returns the currently used number of threads.
process	Override this method to implement calculations.

CPUShader Class

Members

static CPUShaderPtr create ( ) #

Default CPUShader class constructor.

void runSync ( int threads_count ) #

Runs CPU shader code synchronously. This method is blocking. No additional threads are created, as the Engine's thread pool is used.

Arguments

int threads_count - Number of threads to be used. The default value of -1 sets an optimum number of threads calculated for the particular PC.

void runAsync ( int threads_count ) #

Runs CPU shader code asynchronously. This method is non-blocking. No additional threads are created, as the Engine's thread pool is used.

Arguments

int threads_count - Number of threads to be used. The default value of -1 sets an optimum number of threads calculated for the particular PC.

void wait ( ) #

Waits for running asynchronous shader code execution.

int isRunning ( ) #

Returns a value indicating if the CPU shader code is currently executed.

Return value

1 if the shader code is currently executed; otherwise, 0.

int getNumThreads ( ) #

Returns the currently used number of threads.

Return value

Number of currently used threads.

virtual void process ( int thread_num, int threads_count ) #

Override this method to implement calculations.

Arguments

int thread_num - Current thread number. This number is not a thread ID, it just a virtual number.
int threads_count - Number of threads to be used.

Last update: 10.10.2022

Помогите сделать статью лучше

Была ли эта статья полезной?

Сообщить о проблеме в этой статье

(или выберите слово/фразу и нажмите Ctrl+Enter