History log of /petsc/src/mat/impls/aij/seq/seqcusparse/aijcusparse.cu (Results 226 – 250 of 685)
Revision Date Author Comments
# 940e4fdf 12-Jan-2022 Satish Balay <balay@mcs.anl.gov>

Merge branch 'jacobf/2021-12-09/attribute-format-mat' into 'main'

Feature: Attribute Format - Mat

See merge request petsc/petsc!4634


# c0aa6a63 09-Dec-2021 Jacob Faibussowitsch <jacob.fai@gmail.com>

applied attribute format fixes to mat headers


# 29882ab9 04-Jan-2022 Satish Balay <balay@mcs.anl.gov>

Merge branch 'jed/thrust-nosync' into 'main'

thrust: upgrade to 1.16 prerelease for thrust::cuda::par_nosync in vector operations

See merge request petsc/petsc!4676


# a2cee5fe 26-Dec-2021 Jed Brown <jed@jedbrown.org>

thrust: upgrade to 1.16 prerelease for async on streams

Headers in this version are less promiscuous so we need more includes.


# 2158cd20 31-Dec-2021 Satish Balay <balay@mcs.anl.gov>

Merge branch 'stefanozampini/petsc4py' into 'main'

Some work on petsc4py

See merge request petsc/petsc!4644


# 38d8fb69 30-Dec-2021 Satish Balay <balay@mcs.anl.gov>

Merge remote-tracking branch 'origin/release'


# f1093c19 27-Dec-2021 Satish Balay <balay@mcs.anl.gov>

Merge branch 'jczhang/revise-a-comment' into 'release'

Revise a comment

See merge request petsc/petsc!4675


# ae37ee31 26-Dec-2021 Junchao Zhang <jczhang@mcs.anl.gov>

Revise a comment


# 9a14fc28 21-Oct-2021 Stefano Zampini <stefano.zampini@gmail.com>

Mat: add missing general fallbacks for matrices flagged as symmetric

Fix MatMatSolve_Basic using MatDense API to prevent from unneeded device2host copies


# b9968e31 18-Nov-2021 Satish Balay <balay@mcs.anl.gov>

Merge branch 'adams/landau-cuda-opt' into 'main'

Landau optimizations + GPU assembly bug fix

See merge request petsc/petsc!4358


# 365b711f 25-Sep-2021 Mark Adams <ma2325@columbia.edu>

add hook for CPU cusparse solves


# 3ed23c6a 17-Nov-2021 Satish Balay <balay@mcs.anl.gov>

Merge branch 'jczhang/feature-kokkos-spgemm' into 'main'

Add initial MatProduct support for MATAIJKOKKOS

See merge request petsc/petsc!4303


# 67a45760 12-Nov-2021 Junchao Zhang <jczhang@mcs.anl.gov>

MATSEQAIJCUSPARSE: implement MatSeqAIJ{Get,Restore}Array{Read,Write}


# 6cb41cfa 01-Nov-2021 Satish Balay <balay@mcs.anl.gov>

Merge branch 'rmills/feature-dm-bind-below' into 'main'

Rmills/feature dm bind below

See merge request petsc/petsc!3455


# ea500dcf 25-Mar-2021 Richard Tran Mills <rmills@rmills.org>

Prevent unsafe setting of inode.use to true in MatBindToCPU() for cuSPARSE and ViennaCL matrices.

This fixes an issue caused by Stefano's commit c58ef05e, "MatInode: set it to
false when using devic

Prevent unsafe setting of inode.use to true in MatBindToCPU() for cuSPARSE and ViennaCL matrices.

This fixes an issue caused by Stefano's commit c58ef05e, "MatInode: set it to
false when using device matrices", which introduced flipping the inode.use flag
inside MatBindToCPU_SeqAIJCUSPARSE() and MatBindToCPU_SeqAIJViennaCL(). I have
found cases in which this would cause "Corrupted Petsc object, Missing Inode
Structure" errors. Only flipping inode.use to true if inode.size is nonzero
eliminates these errors.

show more ...


# 9c1083e7 25-Mar-2021 Richard Tran Mills <rmills@rmills.org>

MatGetFactor_seqaijcusparse_cusparse() now sets ops to AIJ versions if matrix is bound to CPU.


# 08e3941c 25-Oct-2021 Satish Balay <balay@mcs.anl.gov>

Merge remote-tracking branch 'origin/release'


# f4e315f7 25-Oct-2021 Satish Balay <balay@mcs.anl.gov>

Merge branch 'jczhang/fix-mat-gpu-bugs' into 'release'

Fix several matrix GPU related bugs

See merge request petsc/petsc!4302


# fa046f9f 12-Sep-2021 Junchao Zhang <jczhang@mcs.anl.gov>

MatProduct: need to check symmetry between symbolic and numeric

Suppose we have code:
MatTransposeMatMult(A,B,MAT_INITIAL_MATRIX,fill,C);
MatTransposeMatMult(E,B,MAT_REUSE_MATRIX,fill,C);

If A

MatProduct: need to check symmetry between symbolic and numeric

Suppose we have code:
MatTransposeMatMult(A,B,MAT_INITIAL_MATRIX,fill,C);
MatTransposeMatMult(E,B,MAT_REUSE_MATRIX,fill,C);

If A is symmetric and E is not, but E has the same nonzero pattern as A, then above code is legitimate.
If we change the MatProductType in MAT_INITIAL_MATRIX from MATPRODUCT_AtB to MATPRODUCT_AB, then
in MAT_REUSE_MATRIX, we need to redo symbolic otherwise C=E^B would be wronlgy computed as C=EB.

show more ...


# d60bce21 12-Sep-2021 Junchao Zhang <jczhang@mcs.anl.gov>

MAT: access spptr after MatSeqAIJCUSPARSECopyToGPU, but not before


# 3606e59f 12-Sep-2021 Junchao Zhang <jczhang@mcs.anl.gov>

MAT: form explicit transpose whenever MatSeqAIJCUSPARSEFormExplicitTransposeForMult is called


# 0baf1215 20-Oct-2021 Satish Balay <balay@mcs.anl.gov>

Merge branch 'jacobf/2021-09-04/petscdevice-cupminitialization' into 'main'

Replace current CUPM initialization with unified PetscDevice version

See merge request petsc/petsc!4310


# a4af0cee 19-Oct-2021 Jacob Faibussowitsch <jacob.fai@gmail.com>

- Replace the current CUPM initialization infrastructure with PetscDevice.
- Remove -petsc_use_default_null_stream and -[cuda|hip]_synchronize options
- Remove PetscCUDASynchronize and PetscHIPSynchr

- Replace the current CUPM initialization infrastructure with PetscDevice.
- Remove -petsc_use_default_null_stream and -[cuda|hip]_synchronize options
- Remove PetscCUDASynchronize and PetscHIPSynchronize
- Remove PetscCUDAInitialize(), PetscCUDAInitializeCheck(), PetscHIPInitialize(), and
PetscHIPInitializeCheck()
- Remove PetscCUBLASInitializeHandle(), PetscCUSOLVERDnInitializeHandle(),
PetscHIPBLASInitializeHandle(), and PetscHIPSOLVERInitializeHandle().
- Remove petsc_gputimer_begin and petsc_gputimer_begin
- Add -device_enable, -device_select and -device_view options
- Replace -[cuda|hip]_device with split options -device_enable_[cuda|hip] and -device_select_[cuda|hip]
- Replace -[cuda|hip]_view with -device_view_[cuda|hip]
- Add PetscDeviceInitType
- Add PetscDeviceInitialize() and PetscDeviceInitialized()
- Add PetscDeviceView()

show more ...


# 5ce27ff0 11-Sep-2021 Satish Balay <balay@mcs.anl.gov>

Merge branch 'stefanozampini/hypre-gpu' into 'main'

Interface HYPRE GPU solvers

See merge request petsc/petsc!4248


# 5b7e41fe 11-Sep-2021 Stefano Zampini <stefano.zampini@gmail.com>

MATSEQAIJCUSPARSE: add missing man pages


12345678910>>...28