SpikeGPU  1.0.0
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros
Functions
spike::device::var Namespace Reference

spike::device::var contains all CUDA kernels for the variable bandwidth preconditioner. More...

Functions

template<typename T >
__global__ void bandLU (T *dA, int *ks, int *offsets, int partition_size, int rest_num)
 
template<typename T >
__global__ void bandLU_safe (T *dA, int *ks, int *offsets, int partition_size, int rest_num)
 
template<typename T >
__global__ void bandLU_g32 (T *dA, int *ks, int *offsets, int partition_size, int rest_num)
 
template<typename T >
__global__ void bandLU_g32_safe (T *dA, int *ks, int *offsets, int partition_size, int rest_num)
 
template<typename T >
__global__ void bandLU_critical_div_general (T *dA, int start_row, int *ks, int *offsets, int partition_size, int rest_num)
 
template<typename T >
__global__ void bandLU_critical_div_safe_general (T *dA, int start_row, int *ks, int *offsets, int partition_size, int rest_num)
 
template<typename T >
__global__ void bandLU_critical_sub_general (T *dA, int start_row, int *ks, int *offsets, int partition_size, int rest_num, int last)
 
template<typename T >
__global__ void bandLU_post_divide_per_partition (T *dA, int k, int offset, int partSize)
 
template<typename T >
__global__ void bandLU_post_divide_per_partition_general (T *dA, int k, int offset, int partSize)
 
template<typename T >
__global__ void fullLU_div (T *dA, int *ks, int *offsets, int cur_row)
 
template<typename T >
__global__ void fullLU_div_safe (T *dA, int *ks, int *offsets, int cur_row)
 
template<typename T >
__global__ void fullLU_div_general (T *dA, int *ks, int *offsets, int cur_row)
 
template<typename T >
__global__ void fullLU_div_safe_general (T *dA, int *ks, int *offsets, int cur_row)
 
template<typename T >
__global__ void fullLU_sub (T *dA, int *ks, int *offsets, int cur_row)
 
template<typename T >
__global__ void fullLU_sub_general (T *dA, int *ks, int *offsets, int cur_row)
 
template<typename T >
__global__ void fullLU_sub_spec (T *dA, int *ks, int *offsets)
 
template<typename T >
__global__ void fullLU_sub_spec_general (T *dA, int *ks, int *offsets)
 
template<typename T >
__global__ void fullLU_post_divide (T *dA, int *ks, int *offsets)
 
template<typename T >
__global__ void fullLU_post_divide_general (T *dA, int *ks, int *offsets)
 
template<typename T >
__global__ void boostLastPivot (T *dA, int start_row, int *ks, int *offsets, int partition_size, int rest_num)
 
template<typename T >
__global__ void fwdElim_full_narrow (int N, int *ks, int *offsets, T *dA, T *dB, int b_partition_size, int b_rest_num)
 
template<typename T >
__global__ void fwdElim_full (int N, int *ks, int *offsets, T *dA, T *dB, int b_partition_size, int b_rest_num)
 
template<typename T >
__global__ void preBck_full_divide_narrow (int N, int *ks, int *offsets, T *dA, T *dB, int b_partition_size, int b_rest_num)
 
template<typename T >
__global__ void preBck_full_divide (int N, int *ks, int *offsets, T *dA, T *dB, int b_partition_size, int b_rest_num)
 
template<typename T >
__global__ void bckElim_full_narrow (int N, int *ks, int *offsets, T *dA, T *dB, int b_partition_size, int b_rest_num)
 
template<typename T >
__global__ void bckElim_full (int N, int *ks, int *offsets, T *dA, T *dB, int b_partition_size, int b_rest_num)
 
template<typename T >
__global__ void fwdElim_sol (int N, int *ks, int *offsets, T *dA, T *dB, int partition_size, int rest_num)
 
template<typename T >
__global__ void fwdElim_sol_medium (int N, int *ks, int *offsets, T *dA, T *dB, int partition_size, int rest_num)
 
template<typename T >
__global__ void fwdElim_sol_narrow (int N, int *ks, int *offsets, T *dA, T *dB, int partition_size, int rest_num)
 
template<typename T >
__global__ void bckElim_sol (int N, int *ks, int *offsets, T *dA, T *dB, int partition_size, int rest_num)
 
template<typename T >
__global__ void bckElim_sol_medium (int N, int *ks, int *offsets, T *dA, T *dB, int partition_size, int rest_num)
 
template<typename T >
__global__ void bckElim_sol_narrow (int N, int *ks, int *offsets, T *dA, T *dB, int partition_size, int rest_num)
 
template<typename T >
__global__ void preBck_sol_divide (int N, int *ks, int *offsets, T *dA, T *dB, int partition_size, int rest_num)
 
template<typename T >
__device__ void fwdElim_offDiag_large_tiled (T *dA, T *dB, int idx, int k, int g_k, int r, int first_row, int last_row, int offset, T *a_elements)
 
template<typename T >
__device__ void bckElim_offDiag_large_tiled (T *dA, T *dB, int idx, int k, int g_k, int r, int first_row, int last_row, int offset, T *a_elements)
 
template<typename T >
__global__ void fwdElim_spike (int N, int *ks, int g_k, int rightWidth, int *offsets, T *dA, T *dB, int partition_size, int rest_num, int *left_spike_widths, int *right_spike_widths, int *first_rows)
 
template<typename T >
__global__ void bckElim_spike (int N, int *ks, int g_k, int rightWidth, int *offsets, T *dA, T *dB, int partition_size, int rest_num, int *left_spike_widths, int *right_spike_widths, int *first_rows)
 
template<typename T >
__global__ void fwdElim_rightSpike_per_partition (int N, int k, int pivotIdx, T *dA, T *dB, int first_row, int last_row)
 
template<typename T >
__global__ void preBck_rightSpike_divide_per_partition (int N, int k, int pivotIdx, T *dA, T *dB, int first_row, int last_row)
 
template<typename T >
__global__ void preBck_offDiag_divide_per_partition (int g_k, int k, int pivotIdx, T *dA, T *dB, int first_row, int last_row)
 
template<typename T >
__global__ void preBck_offDiag_divide (int N, int g_k, int *ks, int *offsets, T *dA, T *dB, int partSize, int remainder)
 
template<typename T >
__global__ void bckElim_rightSpike_per_partition (int N, int k, int pivotIdx, T *dA, T *dB, int first_row, int last_row)
 
template<typename T >
__global__ void fwdElim_leftSpike_per_partition (int N, int k, int bid_delta, int pivotIdx, T *dA, T *dB, int first_row, int last_row)
 
template<typename T >
__global__ void preBck_leftSpike_divide_per_partition (int N, int k, int bid_delta, int pivotIdx, T *dA, T *dB, int first_row, int last_row)
 
template<typename T >
__global__ void bckElim_leftSpike_per_partition (int N, int k, int bid_delta, int pivotIdx, T *dA, T *dB, int first_row, int last_row)
 

Detailed Description

spike::device::var contains all CUDA kernels for the variable bandwidth preconditioner.

Function Documentation

template<typename T >
__global__ void spike::device::var::bandLU ( T *  dA,
int *  ks,
int *  offsets,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::bandLU_critical_div_general ( T *  dA,
int  start_row,
int *  ks,
int *  offsets,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::bandLU_critical_div_safe_general ( T *  dA,
int  start_row,
int *  ks,
int *  offsets,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::bandLU_critical_sub_general ( T *  dA,
int  start_row,
int *  ks,
int *  offsets,
int  partition_size,
int  rest_num,
int  last 
)
template<typename T >
__global__ void spike::device::var::bandLU_g32 ( T *  dA,
int *  ks,
int *  offsets,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::bandLU_g32_safe ( T *  dA,
int *  ks,
int *  offsets,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::bandLU_post_divide_per_partition ( T *  dA,
int  k,
int  offset,
int  partSize 
)
template<typename T >
__global__ void spike::device::var::bandLU_post_divide_per_partition_general ( T *  dA,
int  k,
int  offset,
int  partSize 
)
template<typename T >
__global__ void spike::device::var::bandLU_safe ( T *  dA,
int *  ks,
int *  offsets,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::bckElim_full ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  b_partition_size,
int  b_rest_num 
)
template<typename T >
__global__ void spike::device::var::bckElim_full_narrow ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  b_partition_size,
int  b_rest_num 
)
template<typename T >
__global__ void spike::device::var::bckElim_leftSpike_per_partition ( int  N,
int  k,
int  bid_delta,
int  pivotIdx,
T *  dA,
T *  dB,
int  first_row,
int  last_row 
)
template<typename T >
__device__ void spike::device::var::bckElim_offDiag_large_tiled ( T *  dA,
T *  dB,
int  idx,
int  k,
int  g_k,
int  r,
int  first_row,
int  last_row,
int  offset,
T *  a_elements 
)
template<typename T >
__global__ void spike::device::var::bckElim_rightSpike_per_partition ( int  N,
int  k,
int  pivotIdx,
T *  dA,
T *  dB,
int  first_row,
int  last_row 
)
template<typename T >
__global__ void spike::device::var::bckElim_sol ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::bckElim_sol_medium ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::bckElim_sol_narrow ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::bckElim_spike ( int  N,
int *  ks,
int  g_k,
int  rightWidth,
int *  offsets,
T *  dA,
T *  dB,
int  partition_size,
int  rest_num,
int *  left_spike_widths,
int *  right_spike_widths,
int *  first_rows 
)
template<typename T >
__global__ void spike::device::var::boostLastPivot ( T *  dA,
int  start_row,
int *  ks,
int *  offsets,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::fullLU_div ( T *  dA,
int *  ks,
int *  offsets,
int  cur_row 
)
template<typename T >
__global__ void spike::device::var::fullLU_div_general ( T *  dA,
int *  ks,
int *  offsets,
int  cur_row 
)
template<typename T >
__global__ void spike::device::var::fullLU_div_safe ( T *  dA,
int *  ks,
int *  offsets,
int  cur_row 
)
template<typename T >
__global__ void spike::device::var::fullLU_div_safe_general ( T *  dA,
int *  ks,
int *  offsets,
int  cur_row 
)
template<typename T >
__global__ void spike::device::var::fullLU_post_divide ( T *  dA,
int *  ks,
int *  offsets 
)
template<typename T >
__global__ void spike::device::var::fullLU_post_divide_general ( T *  dA,
int *  ks,
int *  offsets 
)
template<typename T >
__global__ void spike::device::var::fullLU_sub ( T *  dA,
int *  ks,
int *  offsets,
int  cur_row 
)
template<typename T >
__global__ void spike::device::var::fullLU_sub_general ( T *  dA,
int *  ks,
int *  offsets,
int  cur_row 
)
template<typename T >
__global__ void spike::device::var::fullLU_sub_spec ( T *  dA,
int *  ks,
int *  offsets 
)
template<typename T >
__global__ void spike::device::var::fullLU_sub_spec_general ( T *  dA,
int *  ks,
int *  offsets 
)
template<typename T >
__global__ void spike::device::var::fwdElim_full ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  b_partition_size,
int  b_rest_num 
)
template<typename T >
__global__ void spike::device::var::fwdElim_full_narrow ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  b_partition_size,
int  b_rest_num 
)
template<typename T >
__global__ void spike::device::var::fwdElim_leftSpike_per_partition ( int  N,
int  k,
int  bid_delta,
int  pivotIdx,
T *  dA,
T *  dB,
int  first_row,
int  last_row 
)
template<typename T >
__device__ void spike::device::var::fwdElim_offDiag_large_tiled ( T *  dA,
T *  dB,
int  idx,
int  k,
int  g_k,
int  r,
int  first_row,
int  last_row,
int  offset,
T *  a_elements 
)
template<typename T >
__global__ void spike::device::var::fwdElim_rightSpike_per_partition ( int  N,
int  k,
int  pivotIdx,
T *  dA,
T *  dB,
int  first_row,
int  last_row 
)
template<typename T >
__global__ void spike::device::var::fwdElim_sol ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::fwdElim_sol_medium ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::fwdElim_sol_narrow ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  partition_size,
int  rest_num 
)
template<typename T >
__global__ void spike::device::var::fwdElim_spike ( int  N,
int *  ks,
int  g_k,
int  rightWidth,
int *  offsets,
T *  dA,
T *  dB,
int  partition_size,
int  rest_num,
int *  left_spike_widths,
int *  right_spike_widths,
int *  first_rows 
)
template<typename T >
__global__ void spike::device::var::preBck_full_divide ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  b_partition_size,
int  b_rest_num 
)
template<typename T >
__global__ void spike::device::var::preBck_full_divide_narrow ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  b_partition_size,
int  b_rest_num 
)
template<typename T >
__global__ void spike::device::var::preBck_leftSpike_divide_per_partition ( int  N,
int  k,
int  bid_delta,
int  pivotIdx,
T *  dA,
T *  dB,
int  first_row,
int  last_row 
)
template<typename T >
__global__ void spike::device::var::preBck_offDiag_divide ( int  N,
int  g_k,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  partSize,
int  remainder 
)
template<typename T >
__global__ void spike::device::var::preBck_offDiag_divide_per_partition ( int  g_k,
int  k,
int  pivotIdx,
T *  dA,
T *  dB,
int  first_row,
int  last_row 
)
template<typename T >
__global__ void spike::device::var::preBck_rightSpike_divide_per_partition ( int  N,
int  k,
int  pivotIdx,
T *  dA,
T *  dB,
int  first_row,
int  last_row 
)
template<typename T >
__global__ void spike::device::var::preBck_sol_divide ( int  N,
int *  ks,
int *  offsets,
T *  dA,
T *  dB,
int  partition_size,
int  rest_num 
)