Add primitive for warp load balancing

[upsj]

This adds a primitive that allows the distribution of variable-sized chunks of work across a warp for better memory coalescing and warp utilization. This can be used as a component in operations like SpGEMM, symbolic Cholesky or, combined with a segmented reduction, could also be used to implement the inner loop in an SpMV similar to MergePath. Looking for feedback and performance benchmark suggestions :)

[upsj]

If the performance is good, this could also be used as an internal component for merge path SpMV

[ginkgo-bot]

Error: The following files need to be formatted:

CMakeLists.txt
benchmark/CMakeLists.txt
benchmark/blas/blas_common.hpp
benchmark/conversion/conversion.cpp
benchmark/matrix_statistics/matrix_statistics.cpp
benchmark/preconditioner/preconditioner.cpp
benchmark/solver/distributed/solver.cpp
benchmark/solver/solver_common.hpp
benchmark/sparse_blas/sparse_blas.cpp
benchmark/spmv/spmv_common.hpp
benchmark/test/CMakeLists.txt
benchmark/utils/general.hpp
benchmark/utils/generator.hpp
benchmark/utils/stencil_matrix.hpp
cmake/CTestScript.cmake
cmake/Modules/CudaArchitectureSelector.cmake
cmake/Modules/FindHWLOC.cmake
cmake/Modules/FindMETIS.cmake
cmake/Modules/FindNUMA.cmake
cmake/Modules/FindNVTX.cmake
cmake/Modules/FindPAPI.cmake
cmake/Modules/FindROCTX.cmake
cmake/Modules/FindVTune.cmake
cmake/Modules/hwloc_helpers.cmake
cmake/autodetect_executors.cmake
cmake/autodetect_system_libs.cmake
cmake/build_helpers.cmake
cmake/build_type_helpers.cmake
cmake/compiler_features.cmake
cmake/create_test.cmake
cmake/cuda.cmake
cmake/get_info.cmake
cmake/hip.cmake
cmake/hip_helpers.cmake
cmake/information_helpers.cmake
cmake/install_helpers.cmake
cmake/rename.cmake
cmake/sycl.cmake
cmake/template_instantiation.cmake
common/cuda_hip/CMakeLists.txt
common/cuda_hip/base/device_matrix_data_kernels.cpp
common/cuda_hip/distributed/assembly_kernels.cpp
common/cuda_hip/distributed/index_map_kernels.cpp
common/cuda_hip/distributed/matrix_kernels.cpp
common/cuda_hip/distributed/partition_kernels.cpp
common/cuda_hip/distributed/vector_kernels.cpp
common/cuda_hip/matrix/ell_kernels.cpp
common/cuda_hip/matrix/sellp_kernels.cpp
common/cuda_hip/matrix/sparsity_csr_kernels.cpp
common/cuda_hip/multigrid/pgm_kernels.cpp
common/unified/CMakeLists.txt
common/unified/distributed/assembly_kernels.cpp
common/unified/matrix/coo_kernels.cpp
common/unified/matrix/csr_kernels.cpp
common/unified/matrix/dense_kernels.template.cpp
common/unified/preconditioner/jacobi_kernels.cpp
core/CMakeLists.txt
core/base/device_matrix_data.cpp
core/base/device_matrix_data_kernels.hpp
core/base/segmented_array.hpp
core/config/config_helper.cpp
core/config/config_helper.hpp
core/config/property_tree.cpp
core/config/registry.cpp
core/config/solver_config.cpp
core/config/solver_config.hpp
core/device_hooks/CMakeLists.txt
core/distributed/assembly.cpp
core/distributed/helpers.hpp
core/distributed/index_map.cpp
core/distributed/index_map_kernels.hpp
core/distributed/matrix.cpp
core/distributed/partition.cpp
core/distributed/partition_kernels.hpp
core/distributed/preconditioner/schwarz.cpp
core/distributed/vector.cpp
core/distributed/vector_cache.cpp
core/matrix/coo.cpp
core/matrix/coo_kernels.hpp
core/matrix/csr.cpp
core/matrix/csr_kernels.hpp
core/multigrid/pgm.cpp
core/multigrid/pgm_kernels.hpp
core/preconditioner/jacobi.cpp
core/preconditioner/jacobi_kernels.hpp
core/solver/gmres.cpp
core/solver/ir.cpp
core/solver/multigrid.cpp
core/test/accessor/CMakeLists.txt
core/test/config/config.cpp
core/test/config/preconditioner.cpp
core/test/config/property_tree.cpp
core/test/config/solver.cpp
core/test/gtest/CMakeLists.txt
core/test/log/CMakeLists.txt
core/test/matrix/csr.cpp
core/test/mpi/base/bindings.cpp
core/test/mpi/distributed/matrix.cpp
core/test/mpi/distributed/preconditioner/schwarz.cpp
core/test/preconditioner/jacobi.cpp
core/test/utils.hpp
core/test/utils/assertions_test.cpp
cuda/CMakeLists.txt
devices/cuda/CMakeLists.txt
devices/dpcpp/CMakeLists.txt
devices/hip/CMakeLists.txt
devices/omp/CMakeLists.txt
devices/reference/CMakeLists.txt
doc/CMakeLists.txt
doc/examples/CMakeLists.txt
dpcpp/CMakeLists.txt
dpcpp/base/device_matrix_data_kernels.dp.cpp
dpcpp/distributed/assembly_kernels.dp.cpp
dpcpp/distributed/index_map_kernels.dp.cpp
dpcpp/distributed/matrix_kernels.dp.cpp
dpcpp/distributed/partition_kernels.dp.cpp
dpcpp/distributed/vector_kernels.dp.cpp
dpcpp/matrix/csr_kernels.dp.cpp
dpcpp/matrix/dense_kernels.dp.cpp
dpcpp/matrix/ell_kernels.dp.cpp
dpcpp/matrix/sellp_kernels.dp.cpp
dpcpp/matrix/sparsity_csr_kernels.dp.cpp
dpcpp/multigrid/pgm_kernels.dp.cpp
dpcpp/test/base/CMakeLists.txt
examples/CMakeLists.txt
examples/adaptiveprecision-blockjacobi/CMakeLists.txt
examples/batched-solver/CMakeLists.txt
examples/cb-gmres/CMakeLists.txt
examples/custom-logger/CMakeLists.txt
examples/custom-matrix-format/CMakeLists.txt
examples/custom-stopping-criterion/CMakeLists.txt
examples/distributed-solver/distributed-solver.cpp
examples/external-lib-interfacing/CMakeLists.txt
examples/file-config-solver/CMakeLists.txt
examples/ginkgo-overhead/CMakeLists.txt
examples/ginkgo-ranges/CMakeLists.txt
examples/heat-equation/CMakeLists.txt
examples/ilu-preconditioned-solver/CMakeLists.txt
examples/inverse-iteration/CMakeLists.txt
examples/ir-ilu-preconditioned-solver/CMakeLists.txt
examples/iterative-refinement/CMakeLists.txt
examples/minimal-cuda-solver/CMakeLists.txt
examples/mixed-multigrid-preconditioned-solver/CMakeLists.txt
examples/mixed-multigrid-solver/CMakeLists.txt
examples/mixed-precision-ir/CMakeLists.txt
examples/mixed-spmv/CMakeLists.txt
examples/multigrid-preconditioned-solver-customized/CMakeLists.txt
examples/multigrid-preconditioned-solver/CMakeLists.txt
examples/nine-pt-stencil-solver/CMakeLists.txt
examples/papi-logging/CMakeLists.txt
examples/par-ilu-convergence/CMakeLists.txt
examples/performance-debugging/CMakeLists.txt
examples/poisson-solver/CMakeLists.txt
examples/preconditioned-solver/CMakeLists.txt
examples/preconditioner-export/CMakeLists.txt
examples/reordered-preconditioned-solver/CMakeLists.txt
examples/schroedinger-splitting/CMakeLists.txt
examples/simple-solver-logging/CMakeLists.txt
examples/simple-solver/CMakeLists.txt
examples/three-pt-stencil-solver/CMakeLists.txt
extensions/CMakeLists.txt
extensions/test/CMakeLists.txt
extensions/test/config/CMakeLists.txt
extensions/test/kokkos/CMakeLists.txt
hip/CMakeLists.txt
hip/test/matrix/CMakeLists.txt
include/CMakeLists.txt
include/ginkgo/core/base/mpi.hpp
include/ginkgo/core/base/precision_dispatch.hpp
include/ginkgo/core/base/std_extensions.hpp
include/ginkgo/core/base/types.hpp
include/ginkgo/core/config/config.hpp
include/ginkgo/core/config/property_tree.hpp
include/ginkgo/core/distributed/index_map.hpp
include/ginkgo/core/distributed/matrix.hpp
include/ginkgo/core/distributed/partition.hpp
include/ginkgo/core/distributed/preconditioner/schwarz.hpp
include/ginkgo/core/distributed/vector.hpp
include/ginkgo/core/distributed/vector_cache.hpp
include/ginkgo/core/matrix/coo.hpp
include/ginkgo/core/matrix/csr.hpp
include/ginkgo/core/matrix/dense.hpp
include/ginkgo/core/multigrid/pgm.hpp
include/ginkgo/core/preconditioner/jacobi.hpp
include/ginkgo/ginkgo.hpp
matrices/CMakeLists.txt
omp/CMakeLists.txt
omp/base/device_matrix_data_kernels.cpp
omp/distributed/assembly_kernels.cpp
omp/distributed/index_map_kernels.cpp
omp/distributed/matrix_kernels.cpp
omp/distributed/partition_kernels.cpp
omp/distributed/vector_kernels.cpp
omp/matrix/csr_kernels.cpp
omp/matrix/dense_kernels.cpp
omp/matrix/ell_kernels.cpp
omp/matrix/fbcsr_kernels.cpp
omp/matrix/sellp_kernels.cpp
omp/matrix/sparsity_csr_kernels.cpp
omp/multigrid/pgm_kernels.cpp
reference/CMakeLists.txt
reference/base/device_matrix_data_kernels.cpp
reference/distributed/assembly_kernels.cpp
reference/distributed/index_map_kernels.cpp
reference/distributed/matrix_kernels.cpp
reference/distributed/partition_helpers.hpp
reference/distributed/partition_kernels.cpp
reference/distributed/vector_kernels.cpp
reference/matrix/coo_kernels.cpp
reference/matrix/csr_kernels.cpp
reference/matrix/dense_kernels.cpp
reference/matrix/ell_kernels.cpp
reference/matrix/fbcsr_kernels.cpp
reference/matrix/sellp_kernels.cpp
reference/matrix/sparsity_csr_kernels.cpp
reference/multigrid/pgm_kernels.cpp
reference/preconditioner/jacobi_kernels.cpp
reference/test/distributed/assembly_kernels.cpp
reference/test/distributed/index_map_kernels.cpp
reference/test/distributed/matrix_kernels.cpp
reference/test/distributed/partition_kernels.cpp
reference/test/distributed/vector_kernels.cpp
reference/test/log/CMakeLists.txt
reference/test/matrix/coo_kernels.cpp
reference/test/matrix/csr_kernels.cpp
reference/test/matrix/dense_kernels.cpp
reference/test/matrix/ell_kernels.cpp
reference/test/matrix/sellp_kernels.cpp
reference/test/matrix/sparsity_csr_kernels.cpp
reference/test/preconditioner/jacobi_kernels.cpp
reference/test/solver/ir_kernels.cpp
test/distributed/assembly_kernels.cpp
test/distributed/index_map_kernels.cpp
test/distributed/matrix_kernels.cpp
test/distributed/partition_kernels.cpp
test/distributed/vector_kernels.cpp
test/factorization/ic_kernels.cpp
test/factorization/ilu_kernels.cpp
test/matrix/CMakeLists.txt
test/matrix/coo_kernels.cpp
test/matrix/csr_kernels.cpp
test/matrix/csr_kernels2.cpp
test/matrix/matrix.cpp
test/mpi/assembly.cpp
test/mpi/matrix.cpp
test/mpi/multigrid/pgm.cpp
test/mpi/preconditioner/schwarz.cpp
test/mpi/solver/solver.cpp
test/preconditioner/jacobi_kernels.cpp
test/reorder/CMakeLists.txt
test/solver/CMakeLists.txt
test/solver/solver.cpp
test/test_exportbuild/CMakeLists.txt
test/test_install/CMakeLists.txt
test/test_install/test_install.cpp
test/test_pkgconfig/CMakeLists.txt
test/test_subdir/CMakeLists.txt
third_party/CMakeLists.txt
third_party/dummy-hook/CMakeLists.txt
third_party/gflags/CMakeLists.txt
third_party/gtest/CMakeLists.txt
third_party/identify_stream_usage/CMakeLists.txt
third_party/nlohmann_json/CMakeLists.txt

You can find a formatting patch under Artifacts here or run format! if you have write access to Ginkgo