CFD/hypre
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

code style

Browse Source
This commit is contained in:
Ruipeng Li 2022-05-27 10:21:07 -07:00
parent 8c58483b56
commit 8070877390
20 changed files with 156 additions and 63 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
src/seq_mv/csr_spgemm_device_numer1.c
View File

@ -8,19 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_numer.h>
template HYPRE_Int
hypre_spgemm_numerical_with_rownnz < 1, HYPRE_SPGEMM_NUMER_HASH_SIZE / 16,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 16, true >
hypre_spgemm_numerical_with_rownnz
< 1, HYPRE_SPGEMM_NUMER_HASH_SIZE / 16, HYPRE_SPGEMM_BASE_GROUP_SIZE / 16, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash,
HYPRE_Int exact_rownnz,
HYPRE_Int *d_ia, HYPRE_Int *d_ja, HYPRE_Complex *d_a, HYPRE_Int *d_ib, HYPRE_Int *d_jb,
HYPRE_Complex *d_b,
HYPRE_Int *d_rc, HYPRE_Int *d_ic, HYPRE_Int *d_jc, HYPRE_Complex *d_c );
template HYPRE_Int hypre_spgemm_numerical_max_num_blocks < HYPRE_SPGEMM_NUMER_HASH_SIZE / 16,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 16 >
template HYPRE_Int
hypre_spgemm_numerical_max_num_blocks
< HYPRE_SPGEMM_NUMER_HASH_SIZE / 16, HYPRE_SPGEMM_BASE_GROUP_SIZE / 16 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

6
src/seq_mv/csr_spgemm_device_numer10.c
View File

@ -13,7 +13,8 @@
#include <csr_spgemm_device_numer.h>
template HYPRE_Int
hypre_spgemm_numerical_with_rownnz< 10, HYPRE_SPGEMM_NUMER_HASH_SIZE * 32, HYPRE_SPGEMM_BASE_GROUP_SIZE * 32, true >
hypre_spgemm_numerical_with_rownnz
< 10, HYPRE_SPGEMM_NUMER_HASH_SIZE * 32, HYPRE_SPGEMM_BASE_GROUP_SIZE * 32, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash,
HYPRE_Int exact_rownnz,
HYPRE_Int *d_ia, HYPRE_Int *d_ja, HYPRE_Complex *d_a, HYPRE_Int *d_ib, HYPRE_Int *d_jb,
@ -21,7 +22,8 @@ hypre_spgemm_numerical_with_rownnz< 10, HYPRE_SPGEMM_NUMER_HASH_SIZE * 32, HYPRE
HYPRE_Int *d_rc, HYPRE_Int *d_ic, HYPRE_Int *d_jc, HYPRE_Complex *d_c );
template HYPRE_Int
hypre_spgemm_numerical_max_num_blocks<HYPRE_SPGEMM_NUMER_HASH_SIZE * 32, HYPRE_SPGEMM_BASE_GROUP_SIZE * 32>
hypre_spgemm_numerical_max_num_blocks
< HYPRE_SPGEMM_NUMER_HASH_SIZE * 32, HYPRE_SPGEMM_BASE_GROUP_SIZE * 32 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA */

11
src/seq_mv/csr_spgemm_device_numer2.c
View File

@ -8,19 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_numer.h>
template HYPRE_Int
hypre_spgemm_numerical_with_rownnz < 2, HYPRE_SPGEMM_NUMER_HASH_SIZE / 8,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 8, true >
hypre_spgemm_numerical_with_rownnz
< 2, HYPRE_SPGEMM_NUMER_HASH_SIZE / 8, HYPRE_SPGEMM_BASE_GROUP_SIZE / 8, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash,
HYPRE_Int exact_rownnz,
HYPRE_Int *d_ia, HYPRE_Int *d_ja, HYPRE_Complex *d_a, HYPRE_Int *d_ib, HYPRE_Int *d_jb,
HYPRE_Complex *d_b,
HYPRE_Int *d_rc, HYPRE_Int *d_ic, HYPRE_Int *d_jc, HYPRE_Complex *d_c );
template HYPRE_Int hypre_spgemm_numerical_max_num_blocks < HYPRE_SPGEMM_NUMER_HASH_SIZE / 8,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 8 >
template HYPRE_Int
hypre_spgemm_numerical_max_num_blocks
< HYPRE_SPGEMM_NUMER_HASH_SIZE / 8, HYPRE_SPGEMM_BASE_GROUP_SIZE / 8 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

11
src/seq_mv/csr_spgemm_device_numer3.c
View File

@ -8,19 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_numer.h>
template HYPRE_Int
hypre_spgemm_numerical_with_rownnz < 3, HYPRE_SPGEMM_NUMER_HASH_SIZE / 4,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 4, true >
hypre_spgemm_numerical_with_rownnz
< 3, HYPRE_SPGEMM_NUMER_HASH_SIZE / 4, HYPRE_SPGEMM_BASE_GROUP_SIZE / 4, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash,
HYPRE_Int exact_rownnz,
HYPRE_Int *d_ia, HYPRE_Int *d_ja, HYPRE_Complex *d_a, HYPRE_Int *d_ib, HYPRE_Int *d_jb,
HYPRE_Complex *d_b,
HYPRE_Int *d_rc, HYPRE_Int *d_ic, HYPRE_Int *d_jc, HYPRE_Complex *d_c );
template HYPRE_Int hypre_spgemm_numerical_max_num_blocks < HYPRE_SPGEMM_NUMER_HASH_SIZE / 4,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 4 >
template HYPRE_Int
hypre_spgemm_numerical_max_num_blocks
< HYPRE_SPGEMM_NUMER_HASH_SIZE / 4, HYPRE_SPGEMM_BASE_GROUP_SIZE / 4 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

11
src/seq_mv/csr_spgemm_device_numer4.c
View File

@ -8,19 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_numer.h>
template HYPRE_Int
hypre_spgemm_numerical_with_rownnz < 4, HYPRE_SPGEMM_NUMER_HASH_SIZE / 2,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 2, true >
hypre_spgemm_numerical_with_rownnz
< 4, HYPRE_SPGEMM_NUMER_HASH_SIZE / 2, HYPRE_SPGEMM_BASE_GROUP_SIZE / 2, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash,
HYPRE_Int exact_rownnz,
HYPRE_Int *d_ia, HYPRE_Int *d_ja, HYPRE_Complex *d_a, HYPRE_Int *d_ib, HYPRE_Int *d_jb,
HYPRE_Complex *d_b,
HYPRE_Int *d_rc, HYPRE_Int *d_ic, HYPRE_Int *d_jc, HYPRE_Complex *d_c );
template HYPRE_Int hypre_spgemm_numerical_max_num_blocks < HYPRE_SPGEMM_NUMER_HASH_SIZE / 2,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 2 >
template HYPRE_Int
hypre_spgemm_numerical_max_num_blocks
< HYPRE_SPGEMM_NUMER_HASH_SIZE / 2, HYPRE_SPGEMM_BASE_GROUP_SIZE / 2 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

11
src/seq_mv/csr_spgemm_device_numer5.c
View File

@ -8,6 +8,9 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_numer.h>
template HYPRE_Int
@ -29,13 +32,15 @@ hypre_spgemm_numerical_with_rownnz
HYPRE_Int *d_rc, HYPRE_Int *d_ic, HYPRE_Int *d_jc, HYPRE_Complex *d_c );
template HYPRE_Int
hypre_spgemm_numerical_max_num_blocks<HYPRE_SPGEMM_NUMER_HASH_SIZE, HYPRE_SPGEMM_BASE_GROUP_SIZE>
hypre_spgemm_numerical_max_num_blocks
< HYPRE_SPGEMM_NUMER_HASH_SIZE, HYPRE_SPGEMM_BASE_GROUP_SIZE >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
template HYPRE_Int
hypreDevice_CSRSpGemmNumerPostCopy<HYPRE_SPGEMM_BASE_GROUP_SIZE>
hypreDevice_CSRSpGemmNumerPostCopy
< HYPRE_SPGEMM_BASE_GROUP_SIZE >
( HYPRE_Int m, HYPRE_Int *d_rc, HYPRE_Int *nnzC, HYPRE_Int **d_ic, HYPRE_Int **d_jc,
HYPRE_Complex **d_c);
HYPRE_Complex **d_c );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

8
src/seq_mv/csr_spgemm_device_numer6.c
View File

@ -8,10 +8,13 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_numer.h>
template HYPRE_Int
hypre_spgemm_numerical_with_rownnz< 6, HYPRE_SPGEMM_NUMER_HASH_SIZE * 2, HYPRE_SPGEMM_BASE_GROUP_SIZE * 2, true >
hypre_spgemm_numerical_with_rownnz
< 6, HYPRE_SPGEMM_NUMER_HASH_SIZE * 2, HYPRE_SPGEMM_BASE_GROUP_SIZE * 2, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash,
HYPRE_Int exact_rownnz,
HYPRE_Int *d_ia, HYPRE_Int *d_ja, HYPRE_Complex *d_a, HYPRE_Int *d_ib, HYPRE_Int *d_jb,
@ -19,7 +22,8 @@ hypre_spgemm_numerical_with_rownnz< 6, HYPRE_SPGEMM_NUMER_HASH_SIZE * 2, HYPRE_S
HYPRE_Int *d_rc, HYPRE_Int *d_ic, HYPRE_Int *d_jc, HYPRE_Complex *d_c );
template HYPRE_Int
hypre_spgemm_numerical_max_num_blocks<HYPRE_SPGEMM_NUMER_HASH_SIZE * 2, HYPRE_SPGEMM_BASE_GROUP_SIZE * 2>
hypre_spgemm_numerical_max_num_blocks
< HYPRE_SPGEMM_NUMER_HASH_SIZE * 2, HYPRE_SPGEMM_BASE_GROUP_SIZE * 2 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

8
src/seq_mv/csr_spgemm_device_numer7.c
View File

@ -8,10 +8,13 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_numer.h>
template HYPRE_Int
hypre_spgemm_numerical_with_rownnz< 7, HYPRE_SPGEMM_NUMER_HASH_SIZE * 4, HYPRE_SPGEMM_BASE_GROUP_SIZE * 4, true >
hypre_spgemm_numerical_with_rownnz
< 7, HYPRE_SPGEMM_NUMER_HASH_SIZE * 4, HYPRE_SPGEMM_BASE_GROUP_SIZE * 4, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash,
HYPRE_Int exact_rownnz,
HYPRE_Int *d_ia, HYPRE_Int *d_ja, HYPRE_Complex *d_a, HYPRE_Int *d_ib, HYPRE_Int *d_jb,
@ -19,7 +22,8 @@ hypre_spgemm_numerical_with_rownnz< 7, HYPRE_SPGEMM_NUMER_HASH_SIZE * 4, HYPRE_S
HYPRE_Int *d_rc, HYPRE_Int *d_ic, HYPRE_Int *d_jc, HYPRE_Complex *d_c );
template HYPRE_Int
hypre_spgemm_numerical_max_num_blocks<HYPRE_SPGEMM_NUMER_HASH_SIZE * 4, HYPRE_SPGEMM_BASE_GROUP_SIZE * 4>
hypre_spgemm_numerical_max_num_blocks
< HYPRE_SPGEMM_NUMER_HASH_SIZE * 4, HYPRE_SPGEMM_BASE_GROUP_SIZE * 4 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

8
src/seq_mv/csr_spgemm_device_numer8.c
View File

@ -8,10 +8,13 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_numer.h>
template HYPRE_Int
hypre_spgemm_numerical_with_rownnz< 8, HYPRE_SPGEMM_NUMER_HASH_SIZE * 8, HYPRE_SPGEMM_BASE_GROUP_SIZE * 8, true >
hypre_spgemm_numerical_with_rownnz
< 8, HYPRE_SPGEMM_NUMER_HASH_SIZE * 8, HYPRE_SPGEMM_BASE_GROUP_SIZE * 8, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash,
HYPRE_Int exact_rownnz,
HYPRE_Int *d_ia, HYPRE_Int *d_ja, HYPRE_Complex *d_a, HYPRE_Int *d_ib, HYPRE_Int *d_jb,
@ -19,7 +22,8 @@ hypre_spgemm_numerical_with_rownnz< 8, HYPRE_SPGEMM_NUMER_HASH_SIZE * 8, HYPRE_S
HYPRE_Int *d_rc, HYPRE_Int *d_ic, HYPRE_Int *d_jc, HYPRE_Complex *d_c );
template HYPRE_Int
hypre_spgemm_numerical_max_num_blocks<HYPRE_SPGEMM_NUMER_HASH_SIZE * 8, HYPRE_SPGEMM_BASE_GROUP_SIZE * 8>
hypre_spgemm_numerical_max_num_blocks
< HYPRE_SPGEMM_NUMER_HASH_SIZE * 8, HYPRE_SPGEMM_BASE_GROUP_SIZE * 8 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

8
src/seq_mv/csr_spgemm_device_numer9.c
View File

@ -8,10 +8,13 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_numer.h>
template HYPRE_Int
hypre_spgemm_numerical_with_rownnz< 9, HYPRE_SPGEMM_NUMER_HASH_SIZE * 16, HYPRE_SPGEMM_BASE_GROUP_SIZE * 16, true >
hypre_spgemm_numerical_with_rownnz
< 9, HYPRE_SPGEMM_NUMER_HASH_SIZE * 16, HYPRE_SPGEMM_BASE_GROUP_SIZE * 16, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash,
HYPRE_Int exact_rownnz,
HYPRE_Int *d_ia, HYPRE_Int *d_ja, HYPRE_Complex *d_a, HYPRE_Int *d_ib, HYPRE_Int *d_jb,
@ -19,7 +22,8 @@ hypre_spgemm_numerical_with_rownnz< 9, HYPRE_SPGEMM_NUMER_HASH_SIZE * 16, HYPRE_
HYPRE_Int *d_rc, HYPRE_Int *d_ic, HYPRE_Int *d_jc, HYPRE_Complex *d_c );
template HYPRE_Int
hypre_spgemm_numerical_max_num_blocks<HYPRE_SPGEMM_NUMER_HASH_SIZE * 16, HYPRE_SPGEMM_BASE_GROUP_SIZE * 16>
hypre_spgemm_numerical_max_num_blocks
< HYPRE_SPGEMM_NUMER_HASH_SIZE * 16, HYPRE_SPGEMM_BASE_GROUP_SIZE * 16 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

13
src/seq_mv/csr_spgemm_device_symbl1.c
View File

@ -8,17 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_symbl.h>
template HYPRE_Int
hypre_spgemm_symbolic_rownnz < 1, HYPRE_SPGEMM_SYMBL_HASH_SIZE / 16,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 16, true >
hypre_spgemm_symbolic_rownnz
< 1, HYPRE_SPGEMM_SYMBL_HASH_SIZE / 16, HYPRE_SPGEMM_BASE_GROUP_SIZE / 16, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int hypre_spgemm_symbolic_max_num_blocks < HYPRE_SPGEMM_SYMBL_HASH_SIZE / 16,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 16 >
template HYPRE_Int
hypre_spgemm_symbolic_max_num_blocks
< HYPRE_SPGEMM_SYMBL_HASH_SIZE / 16, HYPRE_SPGEMM_BASE_GROUP_SIZE / 16 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

9
src/seq_mv/csr_spgemm_device_symbl10.c
View File

@ -15,12 +15,15 @@
#include <csr_spgemm_device_symbl.h>
template HYPRE_Int
hypre_spgemm_symbolic_rownnz<10, HYPRE_SPGEMM_SYMBL_HASH_SIZE * 32, HYPRE_SPGEMM_BASE_GROUP_SIZE * 32, true>
hypre_spgemm_symbolic_rownnz
< 10, HYPRE_SPGEMM_SYMBL_HASH_SIZE * 32, HYPRE_SPGEMM_BASE_GROUP_SIZE * 32, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja, HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int
hypre_spgemm_symbolic_max_num_blocks<HYPRE_SPGEMM_SYMBL_HASH_SIZE * 32, HYPRE_SPGEMM_BASE_GROUP_SIZE * 32>
hypre_spgemm_symbolic_max_num_blocks
< HYPRE_SPGEMM_SYMBL_HASH_SIZE * 32, HYPRE_SPGEMM_BASE_GROUP_SIZE * 32 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

13
src/seq_mv/csr_spgemm_device_symbl2.c
View File

@ -8,17 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_symbl.h>
template HYPRE_Int
hypre_spgemm_symbolic_rownnz < 2, HYPRE_SPGEMM_SYMBL_HASH_SIZE / 8,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 8, true >
hypre_spgemm_symbolic_rownnz
< 2, HYPRE_SPGEMM_SYMBL_HASH_SIZE / 8, HYPRE_SPGEMM_BASE_GROUP_SIZE / 8, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int hypre_spgemm_symbolic_max_num_blocks < HYPRE_SPGEMM_SYMBL_HASH_SIZE / 8,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 8 >
template HYPRE_Int
hypre_spgemm_symbolic_max_num_blocks
< HYPRE_SPGEMM_SYMBL_HASH_SIZE / 8, HYPRE_SPGEMM_BASE_GROUP_SIZE / 8 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

13
src/seq_mv/csr_spgemm_device_symbl3.c
View File

@ -8,17 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_symbl.h>
template HYPRE_Int
hypre_spgemm_symbolic_rownnz < 3, HYPRE_SPGEMM_SYMBL_HASH_SIZE / 4,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 4, true >
hypre_spgemm_symbolic_rownnz
< 3, HYPRE_SPGEMM_SYMBL_HASH_SIZE / 4, HYPRE_SPGEMM_BASE_GROUP_SIZE / 4, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int hypre_spgemm_symbolic_max_num_blocks < HYPRE_SPGEMM_SYMBL_HASH_SIZE / 4,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 4 >
template HYPRE_Int
hypre_spgemm_symbolic_max_num_blocks
< HYPRE_SPGEMM_SYMBL_HASH_SIZE / 4, HYPRE_SPGEMM_BASE_GROUP_SIZE / 4 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

13
src/seq_mv/csr_spgemm_device_symbl4.c
View File

@ -8,17 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_symbl.h>
template HYPRE_Int
hypre_spgemm_symbolic_rownnz < 4, HYPRE_SPGEMM_SYMBL_HASH_SIZE / 2,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 2, true >
hypre_spgemm_symbolic_rownnz
< 4, HYPRE_SPGEMM_SYMBL_HASH_SIZE / 2, HYPRE_SPGEMM_BASE_GROUP_SIZE / 2, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int hypre_spgemm_symbolic_max_num_blocks < HYPRE_SPGEMM_SYMBL_HASH_SIZE / 2,
HYPRE_SPGEMM_BASE_GROUP_SIZE / 2 >
template HYPRE_Int
hypre_spgemm_symbolic_max_num_blocks
< HYPRE_SPGEMM_SYMBL_HASH_SIZE / 2, HYPRE_SPGEMM_BASE_GROUP_SIZE / 2 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

13
src/seq_mv/csr_spgemm_device_symbl5.c
View File

@ -8,22 +8,29 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_symbl.h>
template HYPRE_Int
hypre_spgemm_symbolic_rownnz<5, HYPRE_SPGEMM_SYMBL_HASH_SIZE, HYPRE_SPGEMM_BASE_GROUP_SIZE, true>
hypre_spgemm_symbolic_rownnz
< 5, HYPRE_SPGEMM_SYMBL_HASH_SIZE, HYPRE_SPGEMM_BASE_GROUP_SIZE, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int
hypre_spgemm_symbolic_rownnz<5, HYPRE_SPGEMM_SYMBL_HASH_SIZE, HYPRE_SPGEMM_BASE_GROUP_SIZE, false>
hypre_spgemm_symbolic_rownnz
< 5, HYPRE_SPGEMM_SYMBL_HASH_SIZE, HYPRE_SPGEMM_BASE_GROUP_SIZE, false >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int
hypre_spgemm_symbolic_max_num_blocks<HYPRE_SPGEMM_SYMBL_HASH_SIZE, HYPRE_SPGEMM_BASE_GROUP_SIZE>
hypre_spgemm_symbolic_max_num_blocks
< HYPRE_SPGEMM_SYMBL_HASH_SIZE, HYPRE_SPGEMM_BASE_GROUP_SIZE >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

13
src/seq_mv/csr_spgemm_device_symbl6.c
View File

@ -8,15 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_symbl.h>
template HYPRE_Int
hypre_spgemm_symbolic_rownnz<6, HYPRE_SPGEMM_SYMBL_HASH_SIZE * 2, HYPRE_SPGEMM_BASE_GROUP_SIZE * 2, true>
hypre_spgemm_symbolic_rownnz
< 6, HYPRE_SPGEMM_SYMBL_HASH_SIZE * 2, HYPRE_SPGEMM_BASE_GROUP_SIZE * 2, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja, HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int
hypre_spgemm_symbolic_max_num_blocks<HYPRE_SPGEMM_SYMBL_HASH_SIZE * 2, HYPRE_SPGEMM_BASE_GROUP_SIZE * 2>
hypre_spgemm_symbolic_max_num_blocks
< HYPRE_SPGEMM_SYMBL_HASH_SIZE * 2, HYPRE_SPGEMM_BASE_GROUP_SIZE * 2 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

13
src/seq_mv/csr_spgemm_device_symbl7.c
View File

@ -8,15 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_symbl.h>
template HYPRE_Int
hypre_spgemm_symbolic_rownnz<7, HYPRE_SPGEMM_SYMBL_HASH_SIZE * 4, HYPRE_SPGEMM_BASE_GROUP_SIZE * 4, true>
hypre_spgemm_symbolic_rownnz
< 7, HYPRE_SPGEMM_SYMBL_HASH_SIZE * 4, HYPRE_SPGEMM_BASE_GROUP_SIZE * 4, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja, HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int
hypre_spgemm_symbolic_max_num_blocks<HYPRE_SPGEMM_SYMBL_HASH_SIZE * 4, HYPRE_SPGEMM_BASE_GROUP_SIZE * 4>
hypre_spgemm_symbolic_max_num_blocks
< HYPRE_SPGEMM_SYMBL_HASH_SIZE * 4, HYPRE_SPGEMM_BASE_GROUP_SIZE * 4 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

13
src/seq_mv/csr_spgemm_device_symbl8.c
View File

@ -8,15 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_symbl.h>
template HYPRE_Int
hypre_spgemm_symbolic_rownnz<8, HYPRE_SPGEMM_SYMBL_HASH_SIZE * 8, HYPRE_SPGEMM_BASE_GROUP_SIZE * 8, true>
hypre_spgemm_symbolic_rownnz
< 8, HYPRE_SPGEMM_SYMBL_HASH_SIZE * 8, HYPRE_SPGEMM_BASE_GROUP_SIZE * 8, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja, HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int
hypre_spgemm_symbolic_max_num_blocks<HYPRE_SPGEMM_SYMBL_HASH_SIZE * 8, HYPRE_SPGEMM_BASE_GROUP_SIZE * 8>
hypre_spgemm_symbolic_max_num_blocks
< HYPRE_SPGEMM_SYMBL_HASH_SIZE * 8, HYPRE_SPGEMM_BASE_GROUP_SIZE * 8 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */

13
src/seq_mv/csr_spgemm_device_symbl9.c
View File

@ -8,15 +8,22 @@
#if defined(HYPRE_USING_CUDA) || defined(HYPRE_USING_HIP)
#include <csr_spgemm_device_symbl.h>
template HYPRE_Int
hypre_spgemm_symbolic_rownnz<9, HYPRE_SPGEMM_SYMBL_HASH_SIZE * 16, HYPRE_SPGEMM_BASE_GROUP_SIZE * 16, true>
hypre_spgemm_symbolic_rownnz
< 9, HYPRE_SPGEMM_SYMBL_HASH_SIZE * 16, HYPRE_SPGEMM_BASE_GROUP_SIZE * 16, true >
( HYPRE_Int m, HYPRE_Int *row_ind, HYPRE_Int k, HYPRE_Int n, bool need_ghash, HYPRE_Int *d_ia,
HYPRE_Int *d_ja, HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
HYPRE_Int *d_ja,
HYPRE_Int *d_ib, HYPRE_Int *d_jb, HYPRE_Int *d_rc, bool can_fail, char *d_rf );
template HYPRE_Int
hypre_spgemm_symbolic_max_num_blocks<HYPRE_SPGEMM_SYMBL_HASH_SIZE * 16, HYPRE_SPGEMM_BASE_GROUP_SIZE * 16>
hypre_spgemm_symbolic_max_num_blocks
< HYPRE_SPGEMM_SYMBL_HASH_SIZE * 16, HYPRE_SPGEMM_BASE_GROUP_SIZE * 16 >
( HYPRE_Int multiProcessorCount, HYPRE_Int *num_blocks_ptr, HYPRE_Int *block_size_ptr );
#endif /* HYPRE_USING_CUDA || defined(HYPRE_USING_HIP) */