/home/brandes/workspace/LAMA/src/lama/cuda/CUDAContext.hpp

Go to the documentation of this file.

#ifndef LAMA_CUDA_CONTEXT_HPP_
#define LAMA_CUDA_CONTEXT_HPP_

// for dll_import
#include <lama/config.hpp>

// base classes
#include <lama/Context.hpp>
#include <boost/enable_shared_from_this.hpp>

// others
#include <lama/task/Thread.hpp>

#include <cuda.h>
#include <cuda_runtime.h>
#include <cusparse.h>

// boost
#include <boost/weak_ptr.hpp>
#include <boost/version.hpp>

#include <string>

namespace lama
{

class LAMA_DLL_IMPORTEXPORT CUDAStreamSyncToken;

#define LAMA_CUDA_ENV_FOR_DEVICE "LAMA_DEVICE"

class LAMA_DLL_IMPORTEXPORT CUDAContext : public Context, public boost::enable_shared_from_this<CUDAContext>
{

friend class CUDAContextManager;

public:

    virtual ~CUDAContext();

    int getDeviceNr() const { return mDeviceNr; }

    virtual bool canUseData(const Context& other) const;

    virtual void writeAt( std::ostream& stream ) const;

    virtual void* allocate( const size_t size ) const;

    virtual void allocate( ContextData& contextData, const size_t size ) const;

    virtual void free( void* pointer, const size_t size ) const;

    virtual void free( ContextData& contextData ) const;

    virtual void memcpy( void* dst, const void* src, const size_t size ) const;

    virtual std::auto_ptr<SyncToken> memcpyAsync( void* dst, const void* src, const size_t size ) const;

    virtual bool cancpy( const ContextData& dst, const ContextData& src ) const;

    virtual void memcpy( ContextData& dst, const ContextData& src, const size_t size ) const;

    virtual std::auto_ptr<SyncToken> memcpyAsync( ContextData& dst, const ContextData& src, const size_t size ) const;

    virtual void enable( const char* filename, int line ) const;

    virtual void disable( const char* filename, int line ) const;

    std::auto_ptr<CUDAStreamSyncToken> getComputeSyncToken() const;
    std::auto_ptr<CUDAStreamSyncToken> getTransferSyncToken() const;

    virtual std::auto_ptr<SyncToken> getSyncToken() const;

protected:

    CUDAContext( int device );

private:

    void memcpyFromHost( void* dst, const void* src, const size_t size ) const;
    void memcpyToHost( void* dst, const void* src, const size_t size ) const;
    void memcpyFromCUDAHost( void* dst, const void* src, const size_t size ) const;
    void memcpyToCUDAHost( void* dst, const void* src, const size_t size ) const;

    std::auto_ptr<SyncToken> memcpyAsyncFromHost( void* dst, const void* src, const size_t size ) const;
    std::auto_ptr<SyncToken> memcpyAsyncToHost( void* dst, const void* src, const size_t size ) const;
    std::auto_ptr<SyncToken> memcpyAsyncFromCUDAHost( void* dst, const void* src, const size_t size ) const;
    std::auto_ptr<SyncToken> memcpyAsyncToCUDAHost( void* dst, const void* src, const size_t size ) const;

    int mDeviceNr;  

    CUdevice  mCUdevice;   

    CUcontext mCUcontext;  

    CUstream mTransferStream;  
    CUstream mComputeStream;   

    cusparseHandle_t mCusparseHandle;  

    std::string mDeviceName;  

    Thread::Id mOwnerThread;

    mutable int mNumberOfAllocates;                 
    mutable long long mNumberOfAllocatedBytes;      
    mutable long long mMaxNumberOfAllocatedBytes;   

    static int currentDeviceNr;  

    static int numUsedDevices;   
    
    static size_t minPinnedSize;

    LAMA_LOG_DECL_STATIC_LOGGER( logger );
};

} //namespace lama

#endif // LAMA_CUDA_CONTEXT_HPP_