qmcplusplus::CUDA Namespace Reference

interface to cuBLAS_inhouse calls for different data types S/C/D/Z More...

Functions
template<typename T >
cudaError_t	copyAinvRow_saveGL_batched (cudaStream_t hstream, const int rowchanged, const int n, const T *const Ainv[], const int lda, T *const temp[], T *const rcopy[], const T *const phi_vgl_in[], const size_t phi_vgl_stride, T *const dphi_out[], T *const d2phi_out[], const int batch_count)
	helper function for SM-1 Fahy update subtract one in temp copy Ainv changed row to rcopy save phi G and L as accept. More...
template<typename T >
cudaError_t	calcGradients_batched (cudaStream_t hstream, const int n, const T *const Ainvrow[], const T *const dpsiMrow[], T *const grads_now, const int batch_count)
	calculate gradients More...
template<typename T >
cudaError_t	add_delay_list_save_sigma_VGL_batched (cudaStream_t hstream, int *const delay_list[], const int rowchanged, const int delay_count, T *const binv[], const int binv_lda, const T *const ratio_inv, const T *const phi_vgl_in[], const size_t phi_vgl_stride, T *const phi_out[], T *const dphi_out[], T *const d2phi_out[], const int norb, const int n_accepted, const int batch_count)
template<typename T >
cudaError_t	applyW_batched (cudaStream_t hstream, const int *const delay_list[], const int delay_count, T *const tempMat[], const int lda, const int batch_count)
cudaError_t	print_delay_list_batched (cudaStream_t hstream, int *const delay_list[], const int delay_count, const int batch_count)

Detailed Description

interface to cuBLAS_inhouse calls for different data types S/C/D/Z

Function Documentation

add_delay_list_save_sigma_VGL_batched()

cudaError_t qmcplusplus::CUDA::add_delay_list_save_sigma_VGL_batched	(	cudaStream_t	hstream,
		int *const	delay_list[],
		const int	rowchanged,
		const int	delay_count,
		T *const	binv[],
		const int	binv_lda,
		const T *const	ratio_inv,
		const T *const	phi_vgl_in[],
		const size_t	phi_vgl_stride,
		T *const	phi_out[],
		T *const	dphi_out[],
		T *const	d2phi_out[],
		const int	norb,
		const int	n_accepted,
		const int	batch_count
	)

Referenced by qmcplusplus::compute::add_delay_list_save_sigma_VGL_batched().

applyW_batched()

cudaError_t qmcplusplus::CUDA::applyW_batched	(	cudaStream_t	hstream,
		const int *const	delay_list[],
		const int	delay_count,
		T *const	tempMat[],
		const int	lda,
		const int	batch_count
	)

Referenced by qmcplusplus::compute::applyW_batched().

calcGradients_batched()

cudaError_t qmcplusplus::CUDA::calcGradients_batched	(	cudaStream_t	hstream,
		const int	n,
		const T *const	Ainvrow[],
		const T *const	dpsiMrow[],
		T *const	grads_now,
		const int	batch_count
	)

calculate gradients

Referenced by qmcplusplus::compute::calcGradients_batched().

copyAinvRow_saveGL_batched()

cudaError_t qmcplusplus::CUDA::copyAinvRow_saveGL_batched	(	cudaStream_t	hstream,
		const int	rowchanged,
		const int	n,
		const T *const	Ainv[],
		const int	lda,
		T *const	temp[],
		T *const	rcopy[],
		const T *const	phi_vgl_in[],
		const size_t	phi_vgl_stride,
		T *const	dphi_out[],
		T *const	d2phi_out[],
		const int	batch_count
	)

helper function for SM-1 Fahy update subtract one in temp copy Ainv changed row to rcopy save phi G and L as accept.

Referenced by qmcplusplus::compute::copyAinvRow_saveGL_batched().

print_delay_list_batched()

cudaError_t qmcplusplus::CUDA::print_delay_list_batched	(	cudaStream_t	hstream,
		int *const	delay_list[],
		const int	delay_count,
		const int	batch_count
	)