| /petsc/src/vec/is/sf/impls/basic/cupm/ |
| H A D | sfcupm_impl.hpp | 16 namespace kernels namespace
62 ::Petsc::device::cupm::kernels::util::grid_stride_1D(count, [&](PetscInt tid) { in d_Pack()
75 ::Petsc::device::cupm::kernels::util::grid_stride_1D(count, [&](PetscInt tid) { in d_UnpackAndOp()
88 ::Petsc::device::cupm::kernels::util::grid_stride_1D(count, [&](PetscInt tid) { in d_FetchAndOp()
101 ::Petsc::device::cupm::kernels::util::grid_stride_1D(count, [&](PetscInt tid) { in d_ScatterAndOp()
136 ::Petsc::device::cupm::kernels::util::grid_stride_1D(count, [&](PetscInt tid) { in d_FetchAndOpLocal()
285 …PetscCall(PetscCUPMLaunchKernel1D(count, 0, link->stream, kernels::d_Pack<Type, BS, EQ>, link->bs,… in Pack()
298 …if (PetscDefined(USING_NVCC) && std::is_same<Op, kernels::Insert<Type>>::value && !opt && !idx) { … in UnpackAndOp()
301 …PetscCall(PetscCUPMLaunchKernel1D(count, 0, link->stream, kernels::d_UnpackAndOp<Type, Op, BS, EQ>… in UnpackAndOp()
314 …PetscCall(PetscCUPMLaunchKernel1D(count, 0, link->stream, kernels::d_FetchAndOp<Type, Op, BS, EQ>,… in FetchAndOp()
[all …]
|
| /petsc/src/vec/vec/impls/mpi/cupm/ |
| H A D | vecmpicupm_impl.hpp | 270 namespace kernels namespace
278 …Petsc::device::cupm::kernels::util::grid_stride_1D(nnz, [=](PetscCount i) { buf[i] = vv[perm[i]]; … in pack_coo_values()
346 …PetscCall(PetscCUPMLaunchKernel1D(sendlen, 0, stream, kernels::pack_coo_values, vv, sendlen, vcu->… in SetValuesCOO()
354 …if (const auto n = x->map->n) PetscCall(PetscCUPMLaunchKernel1D(n, 0, stream, kernels::add_coo_val… in SetValuesCOO()
359 …if (const auto nnz2 = vmpi->nnz2) PetscCall(PetscCUPMLaunchKernel1D(nnz2, 0, stream, kernels::add_… in SetValuesCOO()
|
| /petsc/doc/changes/ |
| H A D | 231.md | 9 - Added supported for IBM BGL with optimzed fortran kernels
61 - Changed --with-fortran-kernels=\<0,1> option to
62 --with-fortran-kernels=\<none,generic,bgl>
|
| H A D | 39.md | 18 - Added --with-avx512-kernels to specify that hand-coded kernels
|
| H A D | 312.md | 22 - The ./configure option --with-avx512-kernels defaults to true now
23 so the manually optimized AVX-512 kernels are used by default
|
| H A D | 314.md | 54 example, one can use '--download-kokkos --download-kokkos-kernels
58 Kokkos kernels. Currently, VECKOKKOS supports all vector
|
| H A D | 322.md | 18 - Add `--with-openmp-kernels`
|
| H A D | 34.md | 157 - Added support to deal with linear systems with non-trivial kernels
|
| H A D | 321.md | 103 - Add `MATSELLHIP` and the corresponding HIP kernels for matrix-vector multiplication
|
| /petsc/src/sys/objects/device/impls/cupm/ |
| H A D | kernels.hpp | 14 namespace kernels namespace
|
| /petsc/src/vec/vec/impls/seq/cupm/ |
| H A D | vecseqcupm.hpp | 139 namespace kernels namespace
145 ::Petsc::device::cupm::kernels::util::grid_stride_1D(n, [=](PetscCount i) { in add_coo_values_impl()
|
| H A D | vecseqcupm_impl.hpp | 853 namespace kernels namespace
869 ::Petsc::device::cupm::kernels::util::grid_stride_1D(size, [&](PetscInt i) { in MAXPY_kernel()
914 kernels::MAXPY_kernel<typename detail::repeat_type<const PetscScalar *, Idx>::type...>, in MAXPY_kernel_dispatch_()
1035 namespace kernels namespace
1116 ::Petsc::device::cupm::kernels::util::grid_stride_1D(size, [&](PetscInt i) { in sum_kernel()
1126 ::Petsc::device::cupm::kernels::util::grid_stride_1D(size, [&](PetscInt i) { in sum_kernel()
1160 kernels::MDot_kernel<typename detail::repeat_type<const PetscScalar *, Idx>::type...>, in MDot_kernel_dispatch_()
1253 PetscCall(PetscCUPMLaunchKernel1D(nv, 0, stream, kernels::sum_kernel, nv, d_results)); in MDot_()
2268 …PetscCall(PetscCUPMLaunchKernel1D(n, 0, stream, kernels::add_coo_values, vv, n, vcu->jmap1_d, vcu-… in SetValuesCOO()
|
| /petsc/doc/install/ |
| H A D | external_software.md | 14 - [BLAS/LAPACK](https://www.netlib.org/lapack/lug/node11.html) Optimizes linear algebra kernels (al…
|
| H A D | install.md | 286 These packages provide some basic numeric kernels used by PETSc. `configure` will
655 Use `--with-openmp-kernels` to have some PETSc numerical routines use OpenMP to speed up their comp…
686 … most cases you need only pass the configure option `--download-kokkos` `--download-kokkos-kernels`
|
| /petsc/src/snes/tutorials/output/ |
| H A D | ex19_tut_6.out | 122 Compiled without FORTRAN kernels
|
| H A D | ex19_tut_4.out | 117 Compiled without FORTRAN kernels
|
| H A D | ex19_tut_5.out | 126 Compiled without FORTRAN kernels
|
| H A D | ex19_tut_7.out | 135 Compiled without FORTRAN kernels
|
| /petsc/doc/developers/ |
| H A D | matrices.md | 97 sequential kernels in PETSc can dramatically improve overall floating
|
| /petsc/doc/manual/ |
| H A D | getting_started.md | 622 The ./configure option `--with-openmp-kernels` causes some PETSc numerical kernels to be compiled u…
643 …open-source package that provides a slightly higher-level programming model to utilize GPU kernels.
645 …r `--with-hip` and, if they plan to use Kokkos, also `--download-kokkos --download-kokkos-kernels`.
|
| H A D | profiling.md | 680 lets you see a timeline of code performance information like kernels,
|
| H A D | vec.md | 573 … number of `*`. This allows one to write very natural Kokkos multi-dimensional parallel for kernels
|
| /petsc/doc/community/meetings/2025/ |
| H A D | index.md | 182 … automates the generation of highly optimized explicit finite-difference kernels and stencil compu…
214 …-specific language that automates the application of local computational kernels over a mesh, term…
|
| /petsc/doc/community/meetings/2023/ |
| H A D | index.md | 182 kernels libraries with PETSc on modern architectures.
|
| /petsc/doc/tutorials/physics/ |
| H A D | guide_to_stokes.md | 57 …s. The entire scheme is detailed in {cite}`knepleybrownruppsmith13`. The kernels paired with test …
|