xref: /petsc/src/mat/impls/aij/seq/seqcusparse/cusparsematimpl.h (revision 1d569b8f25bcb9f00023d5059cf903ed8ee5fb46)
1 #ifndef PETSC_CUSPARSEMATIMPL_H
2 #define PETSC_CUSPARSEMATIMPL_H
3 
4 #include <petscpkg_version.h>
5 #include <../src/vec/vec/impls/seq/cupm/vecseqcupm.hpp> /* for VecSeq_CUPM */
6 #include <../src/sys/objects/device/impls/cupm/cupmthrustutility.hpp>
7 #include <petsc/private/petsclegacycupmblas.h>
8 
9 #include <cusparse_v2.h>
10 
11 #include <algorithm>
12 #include <vector>
13 
14 #include <thrust/device_vector.h>
15 #include <thrust/device_ptr.h>
16 #include <thrust/device_malloc_allocator.h>
17 #include <thrust/transform.h>
18 #include <thrust/functional.h>
19 #include <thrust/sequence.h>
20 #include <thrust/system/system_error.h>
21 
22 #if defined(PETSC_USE_COMPLEX)
23   #if defined(PETSC_USE_REAL_SINGLE)
24 const cuComplex PETSC_CUSPARSE_ONE  = {1.0f, 0.0f};
25 const cuComplex PETSC_CUSPARSE_ZERO = {0.0f, 0.0f};
26     #define cusparseXcsrilu02_bufferSize(a, b, c, d, e, f, g, h, i)  cusparseCcsrilu02_bufferSize(a, b, c, d, (cuComplex *)e, f, g, h, i)
27     #define cusparseXcsrilu02_analysis(a, b, c, d, e, f, g, h, i, j) cusparseCcsrilu02_analysis(a, b, c, d, (cuComplex *)e, f, g, h, i, j)
28     #define cusparseXcsrilu02(a, b, c, d, e, f, g, h, i, j)          cusparseCcsrilu02(a, b, c, d, (cuComplex *)e, f, g, h, i, j)
29     #define cusparseXcsric02_bufferSize(a, b, c, d, e, f, g, h, i)   cusparseCcsric02_bufferSize(a, b, c, d, (cuComplex *)e, f, g, h, i)
30     #define cusparseXcsric02_analysis(a, b, c, d, e, f, g, h, i, j)  cusparseCcsric02_analysis(a, b, c, d, (cuComplex *)e, f, g, h, i, j)
31     #define cusparseXcsric02(a, b, c, d, e, f, g, h, i, j)           cusparseCcsric02(a, b, c, d, (cuComplex *)e, f, g, h, i, j)
32   #elif defined(PETSC_USE_REAL_DOUBLE)
33 const cuDoubleComplex PETSC_CUSPARSE_ONE  = {1.0, 0.0};
34 const cuDoubleComplex PETSC_CUSPARSE_ZERO = {0.0, 0.0};
35     #define cusparseXcsrilu02_bufferSize(a, b, c, d, e, f, g, h, i)  cusparseZcsrilu02_bufferSize(a, b, c, d, (cuDoubleComplex *)e, f, g, h, i)
36     #define cusparseXcsrilu02_analysis(a, b, c, d, e, f, g, h, i, j) cusparseZcsrilu02_analysis(a, b, c, d, (cuDoubleComplex *)e, f, g, h, i, j)
37     #define cusparseXcsrilu02(a, b, c, d, e, f, g, h, i, j)          cusparseZcsrilu02(a, b, c, d, (cuDoubleComplex *)e, f, g, h, i, j)
38     #define cusparseXcsric02_bufferSize(a, b, c, d, e, f, g, h, i)   cusparseZcsric02_bufferSize(a, b, c, d, (cuDoubleComplex *)e, f, g, h, i)
39     #define cusparseXcsric02_analysis(a, b, c, d, e, f, g, h, i, j)  cusparseZcsric02_analysis(a, b, c, d, (cuDoubleComplex *)e, f, g, h, i, j)
40     #define cusparseXcsric02(a, b, c, d, e, f, g, h, i, j)           cusparseZcsric02(a, b, c, d, (cuDoubleComplex *)e, f, g, h, i, j)
41   #endif
42 #else
43 const PetscScalar PETSC_CUSPARSE_ONE  = 1.0;
44 const PetscScalar PETSC_CUSPARSE_ZERO = 0.0;
45   #if defined(PETSC_USE_REAL_SINGLE)
46     #define cusparseXcsrilu02_bufferSize cusparseScsrilu02_bufferSize
47     #define cusparseXcsrilu02_analysis   cusparseScsrilu02_analysis
48     #define cusparseXcsrilu02            cusparseScsrilu02
49     #define cusparseXcsric02_bufferSize  cusparseScsric02_bufferSize
50     #define cusparseXcsric02_analysis    cusparseScsric02_analysis
51     #define cusparseXcsric02             cusparseScsric02
52   #elif defined(PETSC_USE_REAL_DOUBLE)
53     #define cusparseXcsrilu02_bufferSize cusparseDcsrilu02_bufferSize
54     #define cusparseXcsrilu02_analysis   cusparseDcsrilu02_analysis
55     #define cusparseXcsrilu02            cusparseDcsrilu02
56     #define cusparseXcsric02_bufferSize  cusparseDcsric02_bufferSize
57     #define cusparseXcsric02_analysis    cusparseDcsric02_analysis
58     #define cusparseXcsric02             cusparseDcsric02
59   #endif
60 #endif
61 
62 #if PETSC_PKG_CUDA_VERSION_GE(9, 0, 0)
63   #define csrsvInfo_t              csrsv2Info_t
64   #define cusparseCreateCsrsvInfo  cusparseCreateCsrsv2Info
65   #define cusparseDestroyCsrsvInfo cusparseDestroyCsrsv2Info
66   #if defined(PETSC_USE_COMPLEX)
67     #if defined(PETSC_USE_REAL_SINGLE)
68       #define cusparseXcsrsv_buffsize(a, b, c, d, e, f, g, h, i, j)          cusparseCcsrsv2_bufferSize(a, b, c, d, e, (cuComplex *)(f), g, h, i, j)
69       #define cusparseXcsrsv_analysis(a, b, c, d, e, f, g, h, i, j, k)       cusparseCcsrsv2_analysis(a, b, c, d, e, (const cuComplex *)(f), g, h, i, j, k)
70       #define cusparseXcsrsv_solve(a, b, c, d, e, f, g, h, i, j, k, l, m, n) cusparseCcsrsv2_solve(a, b, c, d, (const cuComplex *)(e), f, (const cuComplex *)(g), h, i, j, (const cuComplex *)(k), (cuComplex *)(l), m, n)
71     #elif defined(PETSC_USE_REAL_DOUBLE)
72       #define cusparseXcsrsv_buffsize(a, b, c, d, e, f, g, h, i, j)          cusparseZcsrsv2_bufferSize(a, b, c, d, e, (cuDoubleComplex *)(f), g, h, i, j)
73       #define cusparseXcsrsv_analysis(a, b, c, d, e, f, g, h, i, j, k)       cusparseZcsrsv2_analysis(a, b, c, d, e, (const cuDoubleComplex *)(f), g, h, i, j, k)
74       #define cusparseXcsrsv_solve(a, b, c, d, e, f, g, h, i, j, k, l, m, n) cusparseZcsrsv2_solve(a, b, c, d, (const cuDoubleComplex *)(e), f, (const cuDoubleComplex *)(g), h, i, j, (const cuDoubleComplex *)(k), (cuDoubleComplex *)(l), m, n)
75     #endif
76   #else /* not complex */
77     #if defined(PETSC_USE_REAL_SINGLE)
78       #define cusparseXcsrsv_buffsize cusparseScsrsv2_bufferSize
79       #define cusparseXcsrsv_analysis cusparseScsrsv2_analysis
80       #define cusparseXcsrsv_solve    cusparseScsrsv2_solve
81     #elif defined(PETSC_USE_REAL_DOUBLE)
82       #define cusparseXcsrsv_buffsize cusparseDcsrsv2_bufferSize
83       #define cusparseXcsrsv_analysis cusparseDcsrsv2_analysis
84       #define cusparseXcsrsv_solve    cusparseDcsrsv2_solve
85     #endif
86   #endif
87 #else /* PETSC_PKG_CUDA_VERSION_GE(9, 0, 0) */
88   #define csrsvInfo_t              cusparseSolveAnalysisInfo_t
89   #define cusparseCreateCsrsvInfo  cusparseCreateSolveAnalysisInfo
90   #define cusparseDestroyCsrsvInfo cusparseDestroySolveAnalysisInfo
91   #if defined(PETSC_USE_COMPLEX)
92     #if defined(PETSC_USE_REAL_SINGLE)
93       #define cusparseXcsrsv_solve(a, b, c, d_IGNORED, e, f, g, h, i, j, k, l, m_IGNORED, n_IGNORED) cusparseCcsrsv_solve((a), (b), (c), (cuComplex *)(e), (f), (cuComplex *)(g), (h), (i), (j), (cuComplex *)(k), (cuComplex *)(l))
94       #define cusparseXcsrsv_analysis(a, b, c, d, e, f, g, h, i, j_IGNORED, k_IGNORED)               cusparseCcsrsv_analysis((a), (b), (c), (d), (e), (cuComplex *)(f), (g), (h), (i))
95     #elif defined(PETSC_USE_REAL_DOUBLE)
96       #define cusparseXcsrsv_solve(a, b, c, d_IGNORED, e, f, g, h, i, j, k, l, m_IGNORED, n_IGNORED) \
97         cusparseZcsrsv_solve((a), (b), (c), (cuDoubleComplex *)(e), (f), (cuDoubleComplex *)(g), (h), (i), (j), (cuDoubleComplex *)(k), (cuDoubleComplex *)(l))
98       #define cusparseXcsrsv_analysis(a, b, c, d, e, f, g, h, i, j_IGNORED, k_IGNORED) cusparseZcsrsv_analysis((a), (b), (c), (d), (e), (cuDoubleComplex *)(f), (g), (h), (i))
99     #endif
100   #else /* not complex */
101     #if defined(PETSC_USE_REAL_SINGLE)
102       #define cusparseXcsrsv_solve                                     cusparseScsrsv_solve
103       #define cusparseXcsrsv_analysis(a, b, c, d, e, f, g, h, i, j, k) cusparseScsrsv_analysis(a, b, c, d, e, f, g, h, i)
104     #elif defined(PETSC_USE_REAL_DOUBLE)
105       #define cusparseXcsrsv_solve                                     cusparseDcsrsv_solve
106       #define cusparseXcsrsv_analysis(a, b, c, d, e, f, g, h, i, j, k) cusparseDcsrsv_analysis(a, b, c, d, e, f, g, h, i)
107     #endif
108   #endif
109 #endif /* PETSC_PKG_CUDA_VERSION_GE(9, 0, 0) */
110 
111 #if PETSC_PKG_CUDA_VERSION_GE(11, 0, 0)
112   #define cusparse_csr2csc cusparseCsr2cscEx2
113   #if defined(PETSC_USE_COMPLEX)
114     #if defined(PETSC_USE_REAL_SINGLE)
115       #define cusparse_scalartype                                                             CUDA_C_32F
116       #define cusparse_csr_spgeam(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t) cusparseCcsrgeam2(a, b, c, (cuComplex *)d, e, f, (cuComplex *)g, h, i, (cuComplex *)j, k, l, (cuComplex *)m, n, o, p, (cuComplex *)q, r, s, t)
117       #define cusparse_csr_spgeam_bufferSize(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t) \
118         cusparseCcsrgeam2_bufferSizeExt(a, b, c, (cuComplex *)d, e, f, (cuComplex *)g, h, i, (cuComplex *)j, k, l, (cuComplex *)m, n, o, p, (cuComplex *)q, r, s, t)
119     #elif defined(PETSC_USE_REAL_DOUBLE)
120       #define cusparse_scalartype CUDA_C_64F
121       #define cusparse_csr_spgeam(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t) \
122         cusparseZcsrgeam2(a, b, c, (cuDoubleComplex *)d, e, f, (cuDoubleComplex *)g, h, i, (cuDoubleComplex *)j, k, l, (cuDoubleComplex *)m, n, o, p, (cuDoubleComplex *)q, r, s, t)
123       #define cusparse_csr_spgeam_bufferSize(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t) \
124         cusparseZcsrgeam2_bufferSizeExt(a, b, c, (cuDoubleComplex *)d, e, f, (cuDoubleComplex *)g, h, i, (cuDoubleComplex *)j, k, l, (cuDoubleComplex *)m, n, o, p, (cuDoubleComplex *)q, r, s, t)
125     #endif
126   #else /* not complex */
127     #if defined(PETSC_USE_REAL_SINGLE)
128       #define cusparse_scalartype            CUDA_R_32F
129       #define cusparse_csr_spgeam            cusparseScsrgeam2
130       #define cusparse_csr_spgeam_bufferSize cusparseScsrgeam2_bufferSizeExt
131     #elif defined(PETSC_USE_REAL_DOUBLE)
132       #define cusparse_scalartype            CUDA_R_64F
133       #define cusparse_csr_spgeam            cusparseDcsrgeam2
134       #define cusparse_csr_spgeam_bufferSize cusparseDcsrgeam2_bufferSizeExt
135     #endif
136   #endif
137 #else /* PETSC_PKG_CUDA_VERSION_GE(11, 0, 0) */
138   #if defined(PETSC_USE_COMPLEX)
139     #if defined(PETSC_USE_REAL_SINGLE)
140       #define cusparse_csr_spmv(a, b, c, d, e, f, g, h, i, j, k, l, m)                        cusparseCcsrmv((a), (b), (c), (d), (e), (cuComplex *)(f), (g), (cuComplex *)(h), (i), (j), (cuComplex *)(k), (cuComplex *)(l), (cuComplex *)(m))
141       #define cusparse_csr_spmm(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p)               cusparseCcsrmm((a), (b), (c), (d), (e), (f), (cuComplex *)(g), (h), (cuComplex *)(i), (j), (k), (cuComplex *)(l), (m), (cuComplex *)(n), (cuComplex *)(o), (p))
142       #define cusparse_csr2csc(a, b, c, d, e, f, g, h, i, j, k, l)                            cusparseCcsr2csc((a), (b), (c), (d), (cuComplex *)(e), (f), (g), (cuComplex *)(h), (i), (j), (k), (l))
143       #define cusparse_hyb_spmv(a, b, c, d, e, f, g, h)                                       cusparseChybmv((a), (b), (cuComplex *)(c), (d), (e), (cuComplex *)(f), (cuComplex *)(g), (cuComplex *)(h))
144       #define cusparse_csr2hyb(a, b, c, d, e, f, g, h, i, j)                                  cusparseCcsr2hyb((a), (b), (c), (d), (cuComplex *)(e), (f), (g), (h), (i), (j))
145       #define cusparse_hyb2csr(a, b, c, d, e, f)                                              cusparseChyb2csr((a), (b), (c), (cuComplex *)(d), (e), (f))
146       #define cusparse_csr_spgemm(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t) cusparseCcsrgemm(a, b, c, d, e, f, g, h, (cuComplex *)i, j, k, l, m, (cuComplex *)n, o, p, q, (cuComplex *)r, s, t)
147       #define cusparse_csr_spgeam(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s)    cusparseCcsrgeam(a, b, c, (cuComplex *)d, e, f, (cuComplex *)g, h, i, (cuComplex *)j, k, l, (cuComplex *)m, n, o, p, (cuComplex *)q, r, s)
148     #elif defined(PETSC_USE_REAL_DOUBLE)
149       #define cusparse_csr_spmv(a, b, c, d, e, f, g, h, i, j, k, l, m) cusparseZcsrmv((a), (b), (c), (d), (e), (cuDoubleComplex *)(f), (g), (cuDoubleComplex *)(h), (i), (j), (cuDoubleComplex *)(k), (cuDoubleComplex *)(l), (cuDoubleComplex *)(m))
150       #define cusparse_csr_spmm(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p) \
151         cusparseZcsrmm((a), (b), (c), (d), (e), (f), (cuDoubleComplex *)(g), (h), (cuDoubleComplex *)(i), (j), (k), (cuDoubleComplex *)(l), (m), (cuDoubleComplex *)(n), (cuDoubleComplex *)(o), (p))
152       #define cusparse_csr2csc(a, b, c, d, e, f, g, h, i, j, k, l)                            cusparseZcsr2csc((a), (b), (c), (d), (cuDoubleComplex *)(e), (f), (g), (cuDoubleComplex *)(h), (i), (j), (k), (l))
153       #define cusparse_hyb_spmv(a, b, c, d, e, f, g, h)                                       cusparseZhybmv((a), (b), (cuDoubleComplex *)(c), (d), (e), (cuDoubleComplex *)(f), (cuDoubleComplex *)(g), (cuDoubleComplex *)(h))
154       #define cusparse_csr2hyb(a, b, c, d, e, f, g, h, i, j)                                  cusparseZcsr2hyb((a), (b), (c), (d), (cuDoubleComplex *)(e), (f), (g), (h), (i), (j))
155       #define cusparse_hyb2csr(a, b, c, d, e, f)                                              cusparseZhyb2csr((a), (b), (c), (cuDoubleComplex *)(d), (e), (f))
156       #define cusparse_csr_spgemm(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s, t) cusparseZcsrgemm(a, b, c, d, e, f, g, h, (cuDoubleComplex *)i, j, k, l, m, (cuDoubleComplex *)n, o, p, q, (cuDoubleComplex *)r, s, t)
157       #define cusparse_csr_spgeam(a, b, c, d, e, f, g, h, i, j, k, l, m, n, o, p, q, r, s) \
158         cusparseZcsrgeam(a, b, c, (cuDoubleComplex *)d, e, f, (cuDoubleComplex *)g, h, i, (cuDoubleComplex *)j, k, l, (cuDoubleComplex *)m, n, o, p, (cuDoubleComplex *)q, r, s)
159     #endif
160   #else
161     #if defined(PETSC_USE_REAL_SINGLE)
162       #define cusparse_csr_spmv   cusparseScsrmv
163       #define cusparse_csr_spmm   cusparseScsrmm
164       #define cusparse_csr2csc    cusparseScsr2csc
165       #define cusparse_hyb_spmv   cusparseShybmv
166       #define cusparse_csr2hyb    cusparseScsr2hyb
167       #define cusparse_hyb2csr    cusparseShyb2csr
168       #define cusparse_csr_spgemm cusparseScsrgemm
169       #define cusparse_csr_spgeam cusparseScsrgeam
170     #elif defined(PETSC_USE_REAL_DOUBLE)
171       #define cusparse_csr_spmv   cusparseDcsrmv
172       #define cusparse_csr_spmm   cusparseDcsrmm
173       #define cusparse_csr2csc    cusparseDcsr2csc
174       #define cusparse_hyb_spmv   cusparseDhybmv
175       #define cusparse_csr2hyb    cusparseDcsr2hyb
176       #define cusparse_hyb2csr    cusparseDhyb2csr
177       #define cusparse_csr_spgemm cusparseDcsrgemm
178       #define cusparse_csr_spgeam cusparseDcsrgeam
179     #endif
180   #endif
181 #endif /* PETSC_PKG_CUDA_VERSION_GE(11, 0, 0) */
182 
183 #define THRUSTINTARRAY32 thrust::device_vector<int>
184 #define THRUSTINTARRAY   thrust::device_vector<PetscInt>
185 #define THRUSTARRAY      thrust::device_vector<PetscScalar>
186 
187 /* A CSR matrix structure */
188 struct CsrMatrix {
189   PetscInt          num_rows;
190   PetscInt          num_cols;
191   PetscInt          num_entries;
192   THRUSTINTARRAY32 *row_offsets;
193   THRUSTINTARRAY32 *column_indices;
194   THRUSTARRAY      *values;
195 };
196 
197 /* This is struct holding the relevant data needed to a MatSolve */
198 struct Mat_SeqAIJCUSPARSETriFactorStruct {
199   /* Data needed for triangular solve */
200   cusparseMatDescr_t  descr;
201   cusparseOperation_t solveOp;
202   CsrMatrix          *csrMat;
203 #if PETSC_PKG_CUDA_VERSION_LT(11, 4, 0)
204   csrsvInfo_t           solveInfo;
205   cusparseSolvePolicy_t solvePolicy; /* whether level information is generated and used */
206 #endif
207   int          solveBufferSize;
208   void        *solveBuffer;
209   size_t       csr2cscBufferSize; /* to transpose the triangular factor (only used for CUDA >= 11.0) */
210   void        *csr2cscBuffer;
211   PetscScalar *AA_h; /* managed host buffer for moving values to the GPU */
212 };
213 
214 /* This is a larger struct holding all the triangular factors for a solve, transpose solve, and any indices used in a reordering */
215 struct Mat_SeqAIJCUSPARSETriFactors {
216 #if PETSC_PKG_CUDA_VERSION_LT(11, 4, 0)
217   Mat_SeqAIJCUSPARSETriFactorStruct *loTriFactorPtr;          /* pointer for lower triangular (factored matrix) on GPU */
218   Mat_SeqAIJCUSPARSETriFactorStruct *upTriFactorPtr;          /* pointer for upper triangular (factored matrix) on GPU */
219   Mat_SeqAIJCUSPARSETriFactorStruct *loTriFactorPtrTranspose; /* pointer for lower triangular (factored matrix) on GPU for the transpose (useful for BiCG) */
220   Mat_SeqAIJCUSPARSETriFactorStruct *upTriFactorPtrTranspose; /* pointer for upper triangular (factored matrix) on GPU for the transpose (useful for BiCG)*/
221 #endif
222 
223   THRUSTINTARRAY  *rpermIndices; /* indices used for any reordering */
224   THRUSTINTARRAY  *cpermIndices; /* indices used for any reordering */
225   THRUSTARRAY     *workVector;
226   cusparseHandle_t handle; /* a handle to the cusparse library */
227   PetscInt         nnz;    /* number of nonzeros ... need this for accurate logging between ICC and ILU */
228   cudaDeviceProp   dev_prop;
229   PetscBool        init_dev_prop;
230 
231   PetscBool factorizeOnDevice; /* Do factorization on device or not */
232 #if PETSC_PKG_CUDA_VERSION_GE(11, 4, 0)
233   /* csrilu0/csric0 appeared in cusparse-8.0, but we use it along with cusparseSpSV,
234      which first appeared in cusparse-11.5 with cuda-11.3.
235   */
236   PetscScalar *csrVal, *diag;             // the diagonal D in UtDU of Cholesky
237   int         *csrRowPtr32, *csrColIdx32; // i,j of M. cusparseScsrilu02/ic02() etc require 32-bit indices
238 
239   PetscInt    *csrRowPtr, *csrColIdx; // i, j of M on device for CUDA APIs that support 64-bit indices
240   PetscScalar *csrVal_h, *diag_h;     // Since LU is done on host, we prepare a factored matrix in regular csr format on host and then copy it to device
241   PetscInt    *csrRowPtr_h;           // csrColIdx_h is temporary, so it is not here
242 
243   /* Mixed mat descriptor types? yes, different cusparse APIs use different types */
244   cusparseMatDescr_t   matDescr_M;
245   cusparseSpMatDescr_t spMatDescr_L, spMatDescr_U;
246   cusparseSpSVDescr_t  spsvDescr_L, spsvDescr_Lt, spsvDescr_U, spsvDescr_Ut;
247 
248   cusparseDnVecDescr_t dnVecDescr_X, dnVecDescr_Y;
249   PetscScalar         *X, *Y; /* data array of dnVec X and Y */
250 
251   /* Mixed size types? yes, CUDA-11.7.0 declared cusparseDcsrilu02_bufferSizeExt() that returns size_t but did not implement it! */
252   int    factBufferSize_M; /* M ~= LU or LLt */
253   size_t spsvBufferSize_L, spsvBufferSize_Lt, spsvBufferSize_U, spsvBufferSize_Ut;
254   /* cusparse needs various buffers for factorization and solve of L, U, Lt, or Ut.
255      So save memory, we share the factorization buffer with one of spsvBuffer_L/U.
256   */
257   void *factBuffer_M, *spsvBuffer_L, *spsvBuffer_U, *spsvBuffer_Lt, *spsvBuffer_Ut;
258 
259   csrilu02Info_t        ilu0Info_M;
260   csric02Info_t         ic0Info_M;
261   int                   structural_zero, numerical_zero;
262   cusparseSolvePolicy_t policy_M;
263 
264   /* In MatSolveTranspose() for ILU0, we use the two flags to do on-demand solve */
265   PetscBool createdTransposeSpSVDescr;    /* Have we created SpSV descriptors for Lt, Ut? */
266   PetscBool updatedTransposeSpSVAnalysis; /* Have we updated SpSV analysis with the latest L, U values? */
267 
268   PetscLogDouble numericFactFlops; /* Estimated FLOPs in ILU0/ICC0 numeric factorization */
269 #endif
270 };
271 
272 struct Mat_CusparseSpMV {
273   PetscBool initialized;    /* Don't rely on spmvBuffer != NULL to test if the struct is initialized, */
274   size_t    spmvBufferSize; /* since I'm not sure if smvBuffer can be NULL even after cusparseSpMV_bufferSize() */
275   void     *spmvBuffer;
276 #if PETSC_PKG_CUDA_VERSION_GE(11, 0, 0)      /* these are present from CUDA 10.1, but PETSc code makes use of them from CUDA 11 on */
277   cusparseDnVecDescr_t vecXDescr, vecYDescr; /* descriptor for the dense vectors in y=op(A)x */
278 #endif
279 };
280 
281 /* This is struct holding the relevant data needed to a MatMult */
282 struct Mat_SeqAIJCUSPARSEMultStruct {
283   void              *mat;          /* opaque pointer to a matrix. This could be either a cusparseHybMat_t or a CsrMatrix */
284   cusparseMatDescr_t descr;        /* Data needed to describe the matrix for a multiply */
285   THRUSTINTARRAY    *cprowIndices; /* compressed row indices used in the parallel SpMV */
286   PetscScalar       *alpha_one;    /* pointer to a device "scalar" storing the alpha parameter in the SpMV */
287   PetscScalar       *beta_zero;    /* pointer to a device "scalar" storing the beta parameter in the SpMV as zero*/
288   PetscScalar       *beta_one;     /* pointer to a device "scalar" storing the beta parameter in the SpMV as one */
289 #if PETSC_PKG_CUDA_VERSION_GE(11, 0, 0)
290   cusparseSpMatDescr_t matDescr;  /* descriptor for the matrix, used by SpMV and SpMM */
291   Mat_CusparseSpMV     cuSpMV[3]; /* different Mat_CusparseSpMV structs for non-transpose, transpose, conj-transpose */
292   Mat_SeqAIJCUSPARSEMultStruct() : matDescr(NULL)
293   {
294     for (int i = 0; i < 3; i++) cuSpMV[i].initialized = PETSC_FALSE;
295   }
296 #endif
297 };
298 
299 /* This is a larger struct holding all the matrices for a SpMV, and SpMV Transpose */
300 struct Mat_SeqAIJCUSPARSE {
301   Mat_SeqAIJCUSPARSEMultStruct *mat;            /* pointer to the matrix on the GPU */
302   Mat_SeqAIJCUSPARSEMultStruct *matTranspose;   /* pointer to the matrix on the GPU (for the transpose ... useful for BiCG) */
303   THRUSTARRAY                  *workVector;     /* pointer to a workvector to which we can copy the relevant indices of a vector we want to multiply */
304   THRUSTINTARRAY32             *rowoffsets_gpu; /* rowoffsets on GPU in non-compressed-row format. It is used to convert CSR to CSC */
305   PetscInt                      nrows;          /* number of rows of the matrix seen by GPU */
306   MatCUSPARSEStorageFormat      format;         /* the storage format for the matrix on the device */
307   PetscBool                     use_cpu_solve;  /* Use AIJ_Seq (I)LU solve */
308   cudaStream_t                  stream;         /* a stream for the parallel SpMV ... this is not owned and should not be deleted */
309   cusparseHandle_t              handle;         /* a handle to the cusparse library ... this may not be owned (if we're working in parallel i.e. multiGPUs) */
310   PetscObjectState              nonzerostate;   /* track nonzero state to possibly recreate the GPU matrix */
311 #if PETSC_PKG_CUDA_VERSION_GE(11, 0, 0)
312   size_t               csr2cscBufferSize; /* stuff used to compute the matTranspose above */
313   void                *csr2cscBuffer;     /* This is used as a C struct and is calloc'ed by PetscNew() */
314   cusparseCsr2CscAlg_t csr2cscAlg;        /* algorithms can be selected from command line options */
315   cusparseSpMVAlg_t    spmvAlg;
316   cusparseSpMMAlg_t    spmmAlg;
317 #endif
318   THRUSTINTARRAY *csr2csc_i;
319   THRUSTINTARRAY *coords; /* permutation array used in MatSeqAIJCUSPARSEMergeMats */
320 };
321 
322 typedef struct Mat_SeqAIJCUSPARSETriFactors *Mat_SeqAIJCUSPARSETriFactors_p;
323 
324 PETSC_INTERN PetscErrorCode MatSeqAIJCUSPARSECopyToGPU(Mat);
325 PETSC_INTERN PetscErrorCode MatSeqAIJCUSPARSEMergeMats(Mat, Mat, MatReuse, Mat *);
326 PETSC_INTERN PetscErrorCode MatSeqAIJCUSPARSETriFactors_Reset(Mat_SeqAIJCUSPARSETriFactors_p *);
327 
328 using VecSeq_CUDA = Petsc::vec::cupm::impl::VecSeq_CUPM<Petsc::device::cupm::DeviceType::CUDA>;
329 
330 static inline bool isCudaMem(const void *data)
331 {
332   using namespace Petsc::device::cupm;
333   auto mtype = PETSC_MEMTYPE_HOST;
334 
335   PetscFunctionBegin;
336   PetscCallAbort(PETSC_COMM_SELF, impl::Interface<DeviceType::CUDA>::PetscCUPMGetMemType(data, &mtype));
337   PetscFunctionReturn(PetscMemTypeDevice(mtype));
338 }
339 
340 #endif // PETSC_CUSPARSEMATIMPL_H
341