13 #ifndef QMCPLUSPLUS_COMPUTE_MATRIX_UPDATE_SYCL_H 14 #define QMCPLUSPLUS_COMPUTE_MATRIX_UPDATE_SYCL_H 29 const T*
const Ainv[],
33 const T*
const phi_vgl_in[],
34 const size_t phi_vgl_stride,
37 const int batch_count)
42 phi_vgl_stride, dphi_out, d2phi_out, batch_count);
44 catch (sycl::exception&
e)
46 throw std::runtime_error(std::string(
"SYCL::copyAinvRow_saveGL_batched exception: ") +
e.what());
53 const T*
const Ainvrow[],
54 const T*
const dpsiMrow[],
56 const int batch_count)
62 catch (sycl::exception&
e)
64 throw std::runtime_error(std::string(
"SYCL::calcGradients_batched exception: ") +
e.what());
70 int*
const delay_list[],
72 const int delay_count,
75 const T*
const ratio_inv,
76 const T*
const phi_vgl_in[],
77 const size_t phi_vgl_stride,
83 const int batch_count)
88 ratio_inv, phi_vgl_in, phi_vgl_stride, phi_out, dphi_out, d2phi_out,
89 norb, n_accepted, batch_count);
91 catch (sycl::exception&
e)
93 throw std::runtime_error(std::string(
"SYCL::add_delay_list_save_y_VGL_batched exception: ") +
e.what());
100 const int*
const delay_list[],
101 const int delay_count,
104 const int batch_count)
110 catch (sycl::exception&
e)
112 throw std::runtime_error(std::string(
"SYCL::applyW_batched exception: ") +
e.what());
void add_delay_list_save_sigma_VGL_batched(Queue< PlatformKind::CUDA > &queue, 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)
void applyW_batched(Queue< PlatformKind::CUDA > &queue, const int *const delay_list[], const int delay_count, T *const tempMat[], const int lda, const int batch_count)
helper functions for EinsplineSetBuilder
sycl::event calcGradients_batched(sycl::queue &aq, const int n, const T *const Ainvrow[], const T *const dpsiMrow[], T *const grads_now, const int batch_count, const std::vector< sycl::event > &dependencies)
sycl::event add_delay_list_save_sigma_VGL_batched(sycl::queue &aq, 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 int 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, const std::vector< sycl::event > &dependencies)
sycl::event applyW_batched(sycl::queue &aq, const int *const delay_list[], const int delay_count, T *const tempMat[], const int lda, const int batch_count, const std::vector< sycl::event > &dependencies)
void calcGradients_batched(Queue< PlatformKind::CUDA > &queue, const int n, const T *const Ainvrow[], const T *const dpsiMrow[], T *const grads_now, const int batch_count)
void copyAinvRow_saveGL_batched(Queue< PlatformKind::CUDA > &queue, 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)
sycl::event copyAinvRow_saveGL_batched(sycl::queue &aq, 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 int phi_vgl_stride, T *const dphi_out[], T *const d2phi_out[], const int batch_count, const std::vector< sycl::event > &dependencies)