Scaled permutation and improved permutation interface #1415

upsj · 2023-10-06T19:35:14Z

This creates a new permutation interface to enable more functionality without overloading the Csr and Dense interfaces (which was a valid concern in #1338) and adds scaled permutation functionality merging scaling and permutation.

This is still WIP, but the interface itself is already polished enough (and the PR is big enough) that this should already be reviewed.

TODO:

SYCL kernels
Switch from S * P to P * S, i.e. first scaling, then permuting.
Match Permutation and ScaledPermutation interface (removing size parameter and permute mask member)
Add combination functionality (Scaled)Permutation * (Scaled)Permutation
~~Add Permuted and ScaledPermuted interface~~ (probably better aimed at another PR)

upsj · 2023-10-12T20:55:46Z

format!

sonarqubecloud · 2023-10-13T06:20:39Z

Kudos, SonarCloud Quality Gate passed!

0 Bugs
0 Vulnerabilities
6 Security Hotspots
160 Code Smells

80.4% Coverage
7.5% Duplication

The version of Java (11.0.3) you have used to run this analysis is deprecated and we will stop accepting it soon. Please update to at least Java 17.
Read more here

MarcelKoch

Looks good. I mostly have some issues regarding the documentation and then just some minor nits. Also, core tests for scaled permutation are missing.

include/ginkgo/core/base/exception.hpp

include/ginkgo/core/matrix/permutation.hpp

include/ginkgo/core/matrix/scaled_permutation.hpp

common/unified/matrix/csr_kernels.cpp

reference/test/matrix/dense_kernels.cpp

reference/test/matrix/csr_kernels.cpp

test/matrix/permutation_kernels.cpp

test/matrix/scaled_permutation_kernels.cpp

pratikvn

Some comments from a first look. A few minor things, but mostly looks good.

core/matrix/csr.cpp

include/ginkgo/core/matrix/scaled_permutation.hpp

omp/matrix/csr_kernels.cpp

include/ginkgo/core/matrix/permutation.hpp

reference/test/matrix/csr_kernels.cpp

common/unified/matrix/csr_kernels.cpp

common/cuda_hip/matrix/csr_kernels.hpp.inc

common/unified/matrix/scaled_permutation_kernels.cpp

yhmtsai · 2023-10-17T09:03:59Z

core/test/matrix/permutation.cpp

-}
-
-
-TYPED_TEST(Permutation, FactorySetsCorrectPermuteMask)


the deleted test

yes, permute_mask is now fixed and cannot be changed.

yhmtsai · 2023-10-17T09:05:10Z

core/test/matrix/permutation.cpp


    ASSERT_EQ(m->get_const_permutation(), data);
 }


 TYPED_TEST(Permutation, PermutationCanBeConstructedFromExistingConstData)
 {
-    using i_type = typename TestFixture::i_type;


several test are be deleted

dpcpp/matrix/csr_kernels.dp.cpp

yhmtsai · 2023-10-17T09:12:54Z

include/ginkgo/core/base/lin_op.hpp

+     * From the linear algebra perspective, with $P_{ij} = \delta_{i
+     * \pi(i)}$, this represents the operation $P^{-1} A P^{-T}$.


are \delta and \pi commonly used? Maybe I miss that, but I only see many greek letter are used from wiki

In (linear) algebra, lowercase pi is often used to denote permutations, and delta means the Kronecker delta. But I agree that this should be clearer, at least we can refer to the permutation by variable.

I think when it's rendered in latex it makes more sense.

omp/matrix/csr_kernels.cpp

reference/matrix/csr_kernels.cpp

reference/matrix/scaled_permutation_kernels.cpp

yhmtsai · 2023-10-17T11:18:10Z

reference/test/matrix/permutation.cpp

-}
-
-
-TYPED_TEST(Permutation, AppliesRowAndColPermutationToDenseWithOneArray)


missing deleted function

The functionality is no longer available through Permutation

include/ginkgo/core/matrix/csr.hpp

include/ginkgo/core/matrix/scaled_permutation.hpp

reference/test/matrix/scaled_permutation.cpp

test/matrix/csr_kernels2.cpp

reference/test/matrix/csr_kernels.cpp

core/matrix/permutation.cpp

yhmtsai

second part of my review. Some conversations are marked resolved but actually not. I assume you do not push the changes yet. I prefer marking resolved after pushing or will push soon. Otherwise, I am not sure whether it is resolved or missed. Is it rebased after #1423 is merged?

yhmtsai · 2023-10-24T07:36:56Z

common/unified/matrix/csr_kernels.cpp

+                const auto out_col = permutation[in_col_idxs[tid]];
+                out_col_idxs[tid] = out_col;
+                out_vals[tid] = in_vals[tid] / scale[out_col];


inverse_column: did permutation first not the scaling first here, because we only have $(P \times S)^{-1} = S^{-1} \times P^{-1} $ not $P^{-1} \{P^{-T}S^{-1}P^{-1}\}$

the second formula $P^{-1} (P S^{-1} P^T)$ is only useful for inverting a scaled permutation into the $PS$ form again, we don't need it for the scaling. Is there a question on this kernel?

It is only to clear my understanding is correct or not.
in this kernel, you did the permutation first and then the scaling directly not the other direction because you do not have and do not need to generate the second formula

common/unified/matrix/dense_kernels.template.cpp

include/ginkgo/core/matrix/csr.hpp

reference/matrix/dense_kernels.cpp

yhmtsai · 2023-10-24T08:04:38Z

core/matrix/dense.cpp

+    if ((mode & permute_mode::symmetric) == permute_mode::none) {
+        output->copy_from(this);
+        return;
+    }


I assume symmetric is the full bit set of all possible settings but it is not obvious here I think.
Could we just check permute_mode::none without the mask? the user input more than the bitset will also throw the error in the end.

no, symmetric is rows | columns, so this checks that we are permuting both at the same time

core/matrix/dense.cpp

core/matrix/csr.cpp

MarcelKoch

LGTM, only few minor comments.

core/matrix/csr.cpp

reference/test/matrix/permutation.cpp

reference/test/matrix/scaled_permutation.cpp

pratikvn

Minor things, but otherwise LGTM.

core/factorization/factorization.cpp

pratikvn · 2023-10-25T12:52:50Z

core/matrix/permutation.cpp

+    run<const Dense<double>*, const Dense<float>*,
+        const Dense<complex<double>>*, const Dense<complex<float>>*>(op, fn);


Can we have this in a compile time list and maybe substitute with a macro instead ? I think if half and other value types are added, updating these lists becomes tedious.

I think this is a bigger topic that I would like to tackle more uniformly elsewhere. Right now we don't have that many use cases yet.

dpcpp/matrix/csr_kernels.dp.cpp

include/ginkgo/core/matrix/csr.hpp

core/matrix/permutation.cpp

include/ginkgo/core/matrix/scaled_permutation.hpp

yhmtsai

I still have two files left for review. It should be done later today
I am thinking using prepend rather than combine

core/matrix/permutation.hpp

yhmtsai · 2023-10-25T12:40:10Z

include/ginkgo/core/matrix/permutation.hpp

-          col_size_(size[1]),
-          enabled_permute_(enabled_permute)
-    {}
+    Permutation(std::shared_ptr<const Executor> exec, size_type = 0);


I thought it is not necessary in core.cpp

yhmtsai · 2023-10-25T12:42:23Z

include/ginkgo/core/matrix/permutation.hpp

-            }
-        }
-        out->move_from(tmp);
+        GKO_ASSERT_EQ(enabled_permute, row_permute);


It was supported column_permute or combined one

they didn't make sense as LinOps and only supported square matrices, the interface was not suited for what it represents.

yhmtsai · 2023-10-25T12:44:21Z

include/ginkgo/core/matrix/permutation.hpp

+    std::unique_ptr<Permutation> combine(
+        ptr_param<const Permutation> other) const;


maybe prepend? combine may be confused with the combination of ginkgo (c1 op1 + c2 op2)

I think prepend is more suitable for collections, this is an operation that produces a single object from two objects. I do agree that this is confusing though, I will change it to compose, which pretty accurately represents things, albeit a bit reversed (op1->compose(op2) = $op2 \circ op1$)

or composed if it helps the order?

I think the name should describe an action, composed sounds more like a check.

include/ginkgo/core/matrix/scaled_permutation.hpp

reference/matrix/scaled_permutation_kernels.cpp

core/matrix/dense.cpp

yhmtsai · 2023-10-25T13:48:58Z

core/matrix/permutation.cpp

+{
+    switch (mode) {
+    case permute_mode::none:
+        return stream << "none";


with some prefix like perm::none?

I would argue that is the caller's responsibility, as they already know they are printing a permutation_mode. This is only used in debugging though, so it's not very important.

dpcpp/matrix/csr_kernels.dp.cpp

reference/matrix/permutation_kernels.cpp

reference/test/matrix/permutation.cpp

common/cuda_hip/matrix/csr_kernels.hpp.inc

reference/test/matrix/dense_kernels.cpp

reference/test/matrix/csr_kernels.cpp

yhmtsai

LGTM. some nit and some name suggestion.
also reminder: need to move the latest version of check_diag and add_scaled_identity

- remove permute_mask - deprecate permute_mask and dim<2> parameters

Co-authored-by: Tobias Ribizel <[email protected]>

- fix and test csr column permutation for hypersparse matrices - fix inverse nonsymm scaled permutation for dense - extract permutation dimension validation - simplify permutation dispatch - re-add templated constructors to preserve interface - move more code to source - add documentation for permutation to permute_mode instead of matrix classes - improve documentation - make scaled_permutation final - remove unnecessary InvalidValueError - improve error message on dispatch_helper run(...) - simplify rectangular permutation tests - fix dense tests not calling scaled permutation - test apply in (scaled) permutation on the device - throw DimensionMismatch instead of ValueMismatch - add more round-trip and inverted permutation tests - add dpcpp subgroup size Co-authored-by: Pratik Nayak <[email protected]> Co-authored-by: Yuhsiang M. Tsai <[email protected]> Co-authored-by: Marcel Koch <[email protected]>

also adds a few more tests to be sure.

- const-correctness - rename combine to compose - rename invert to compute_inverse - improve documentation - add more tests - simplify DPC++ kernel launch Co-authored-by: Marcel Koch <[email protected]> Co-authored-by: Yuhsiang M. Tsai <[email protected]> Co-authored-by: Pratik Nayak <[email protected]>

Release 1.7.0 to master The Ginkgo team is proud to announce the new Ginkgo minor release 1.7.0. This release brings new features such as: - Complete GPU-resident sparse direct solvers feature set and interfaces, - Improved Cholesky factorization performance, - A new MC64 reordering, - Batched iterative solver support with the BiCGSTAB solver with batched Dense and ELL matrix types, - MPI support for the SYCL backend, - Improved ParILU(T)/ParIC(T) preconditioner convergence, and more! If you face an issue, please first check our [known issues page](https://github.com/ginkgo-project/ginkgo/wiki/Known-Issues) and the [open issues list](https://github.com/ginkgo-project/ginkgo/issues) and if you do not find a solution, feel free to [open a new issue](https://github.com/ginkgo-project/ginkgo/issues/new/choose) or ask a question using the [github discussions](https://github.com/ginkgo-project/ginkgo/discussions). Supported systems and requirements: + For all platforms, CMake 3.16+ + C++14 compliant compiler + Linux and macOS + GCC: 5.5+ + clang: 3.9+ + Intel compiler: 2019+ + Apple Clang: 14.0 is tested. Earlier versions might also work. + NVHPC: 22.7+ + Cray Compiler: 14.0.1+ + CUDA module: CMake 3.18+, and CUDA 10.1+ or NVHPC 22.7+ + HIP module: ROCm 4.5+ + DPC++ module: Intel oneAPI 2022.1+ with oneMKL and oneDPL. Set the CXX compiler to `dpcpp` or `icpx`. + MPI: standard version 3.1+, ideally GPU Aware, for best performance + Windows + MinGW: GCC 5.5+ + Microsoft Visual Studio: VS 2019+ + CUDA module: CUDA 10.1+, Microsoft Visual Studio + OpenMP module: MinGW. ### Version support changes + CUDA 9.2 is no longer supported and 10.0 is untested [#1382](#1382) + Ginkgo now requires CMake version 3.16 (and 3.18 for CUDA) [#1368](#1368) ### Interface changes + `const` Factory parameters can no longer be modified through `with_*` functions, as this breaks const-correctness [#1336](#1336) [#1439](#1439) ### New Deprecations + The `device_reset` parameter of CUDA and HIP executors no longer has an effect, and its `allocation_mode` parameters have been deprecated in favor of the `Allocator` interface. [#1315](#1315) + The CMake parameter `GINKGO_BUILD_DPCPP` has been deprecated in favor of `GINKGO_BUILD_SYCL`. [#1350](#1350) + The `gko::reorder::Rcm` interface has been deprecated in favor of `gko::experimental::reorder::Rcm` based on `Permutation`. [#1418](#1418) + The Permutation class' `permute_mask` functionality. [#1415](#1415) + Multiple functions with typos (`set_complex_subpsace()`, range functions such as `conj_operaton` etc). [#1348](#1348) ### Summary of previous deprecations + `gko::lend()` is not necessary anymore. + The classes `RelativeResidualNorm` and `AbsoluteResidualNorm` are deprecated in favor of `ResidualNorm`. + The class `AmgxPgm` is deprecated in favor of `Pgm`. + Default constructors for the CSR `load_balance` and `automatical` strategies + The PolymorphicObject's move-semantic `copy_from` variant + The templated `SolverBase` class. + The class `MachineTopology` is deprecated in favor of `machine_topology`. + Logger constructors and create functions with the `executor` parameter. + The virtual, protected, Dense functions `compute_norm1_impl`, `add_scaled_impl`, etc. + Logger events for solvers and criterion without the additional `implicit_tau_sq` parameter. + The global `gko::solver::default_krylov_dim`, use instead `gko::solver::gmres_default_krylov_dim`. ### Added features + Adds a batch::BatchLinOp class that forms a base class for batched linear operators such as batched matrix formats, solver and preconditioners [#1379](#1379) + Adds a batch::MultiVector class that enables operations such as dot, norm, scale on batched vectors [#1371](#1371) + Adds a batch::Dense matrix format that stores batched dense matrices and provides gemv operations for these dense matrices. [#1413](#1413) + Adds a batch::Ell matrix format that stores batched Ell matrices and provides spmv operations for these batched Ell matrices. [#1416](#1416) [#1437](#1437) + Add a batch::Bicgstab solver (class, core, and reference kernels) that enables iterative solution of batched linear systems [#1438](#1438). + Add device kernels (CUDA, HIP, and DPCPP) for batch::Bicgstab solver. [#1443](#1443). + New MC64 reordering algorithm which optimizes the diagonal product or sum of a matrix by permuting the rows, and computes additional scaling factors for equilibriation [#1120](#1120) + New interface for (non-symmetric) permutation and scaled permutation of Dense and Csr matrices [#1415](#1415) + LU and Cholesky Factorizations can now be separated into their factors [#1432](#1432) + New symbolic LU factorization algorithm that is optimized for matrices with an almost-symmetric sparsity pattern [#1445](#1445) + Sorting kernels for SparsityCsr on all backends [#1343](#1343) + Allow passing pre-generated local solver as factory parameter for the distributed Schwarz preconditioner [#1426](#1426) + Add DPCPP kernels for Partition [#1034](#1034), and CSR's `check_diagonal_entries` and `add_scaled_identity` functionality [#1436](#1436) + Adds a helper function to create a partition based on either local sizes, or local ranges [#1227](#1227) + Add function to compute arithmetic mean of dense and distributed vectors [#1275](#1275) + Adds `icpx` compiler supports [#1350](#1350) + All backends can be built simultaneously [#1333](#1333) + Emits a CMake warning in downstream projects that use different compilers than the installed Ginkgo [#1372](#1372) + Reordering algorithms in sparse_blas benchmark [#1354](#1354) + Benchmarks gained an `-allocator` parameter to specify device allocators [#1385](#1385) + Benchmarks gained an `-input_matrix` parameter that initializes the input JSON based on the filename [#1387](#1387) + Benchmark inputs can now be reordered as a preprocessing step [#1408](#1408) ### Improvements + Significantly improve Cholesky factorization performance [#1366](#1366) + Improve parallel build performance [#1378](#1378) + Allow constrained parallel test execution using CTest resources [#1373](#1373) + Use arithmetic type more inside mixed precision ELL [#1414](#1414) + Most factory parameters of factory type no longer need to be constructed explicitly via `.on(exec)` [#1336](#1336) [#1439](#1439) + Improve ParILU(T)/ParIC(T) convergence by using more appropriate atomic operations [#1434](#1434) ### Fixes + Fix an over-allocation for OpenMP reductions [#1369](#1369) + Fix DPCPP's common-kernel reduction for empty input sizes [#1362](#1362) + Fix several typos in the API and documentation [#1348](#1348) + Fix inconsistent `Threads` between generations [#1388](#1388) + Fix benchmark median condition [#1398](#1398) + Fix HIP 5.6.0 compilation [#1411](#1411) + Fix missing destruction of rand_generator from cuda/hip [#1417](#1417) + Fix PAPI logger destruction order [#1419](#1419) + Fix TAU logger compilation [#1422](#1422) + Fix relative criterion to not iterate if the residual is already zero [#1079](#1079) + Fix memory_order invocations with C++20 changes [#1402](#1402) + Fix `check_diagonal_entries_exist` report correctly when only missing diagonal value in the last rows. [#1440](#1440) + Fix checking OpenMPI version in cross-compilation settings [#1446](#1446) + Fix false-positive deprecation warnings in Ginkgo, especially for the old Rcm (it doesn't emit deprecation warnings anymore as a result but is still considered deprecated) [#1444](#1444) ### Related PR: #1451

Release 1.7.0 to develop The Ginkgo team is proud to announce the new Ginkgo minor release 1.7.0. This release brings new features such as: - Complete GPU-resident sparse direct solvers feature set and interfaces, - Improved Cholesky factorization performance, - A new MC64 reordering, - Batched iterative solver support with the BiCGSTAB solver with batched Dense and ELL matrix types, - MPI support for the SYCL backend, - Improved ParILU(T)/ParIC(T) preconditioner convergence, and more! If you face an issue, please first check our [known issues page](https://github.com/ginkgo-project/ginkgo/wiki/Known-Issues) and the [open issues list](https://github.com/ginkgo-project/ginkgo/issues) and if you do not find a solution, feel free to [open a new issue](https://github.com/ginkgo-project/ginkgo/issues/new/choose) or ask a question using the [github discussions](https://github.com/ginkgo-project/ginkgo/discussions). Supported systems and requirements: + For all platforms, CMake 3.16+ + C++14 compliant compiler + Linux and macOS + GCC: 5.5+ + clang: 3.9+ + Intel compiler: 2019+ + Apple Clang: 14.0 is tested. Earlier versions might also work. + NVHPC: 22.7+ + Cray Compiler: 14.0.1+ + CUDA module: CMake 3.18+, and CUDA 10.1+ or NVHPC 22.7+ + HIP module: ROCm 4.5+ + DPC++ module: Intel oneAPI 2022.1+ with oneMKL and oneDPL. Set the CXX compiler to `dpcpp` or `icpx`. + MPI: standard version 3.1+, ideally GPU Aware, for best performance + Windows + MinGW: GCC 5.5+ + Microsoft Visual Studio: VS 2019+ + CUDA module: CUDA 10.1+, Microsoft Visual Studio + OpenMP module: MinGW. ### Version support changes + CUDA 9.2 is no longer supported and 10.0 is untested [#1382](#1382) + Ginkgo now requires CMake version 3.16 (and 3.18 for CUDA) [#1368](#1368) ### Interface changes + `const` Factory parameters can no longer be modified through `with_*` functions, as this breaks const-correctness [#1336](#1336) [#1439](#1439) ### New Deprecations + The `device_reset` parameter of CUDA and HIP executors no longer has an effect, and its `allocation_mode` parameters have been deprecated in favor of the `Allocator` interface. [#1315](#1315) + The CMake parameter `GINKGO_BUILD_DPCPP` has been deprecated in favor of `GINKGO_BUILD_SYCL`. [#1350](#1350) + The `gko::reorder::Rcm` interface has been deprecated in favor of `gko::experimental::reorder::Rcm` based on `Permutation`. [#1418](#1418) + The Permutation class' `permute_mask` functionality. [#1415](#1415) + Multiple functions with typos (`set_complex_subpsace()`, range functions such as `conj_operaton` etc). [#1348](#1348) ### Summary of previous deprecations + `gko::lend()` is not necessary anymore. + The classes `RelativeResidualNorm` and `AbsoluteResidualNorm` are deprecated in favor of `ResidualNorm`. + The class `AmgxPgm` is deprecated in favor of `Pgm`. + Default constructors for the CSR `load_balance` and `automatical` strategies + The PolymorphicObject's move-semantic `copy_from` variant + The templated `SolverBase` class. + The class `MachineTopology` is deprecated in favor of `machine_topology`. + Logger constructors and create functions with the `executor` parameter. + The virtual, protected, Dense functions `compute_norm1_impl`, `add_scaled_impl`, etc. + Logger events for solvers and criterion without the additional `implicit_tau_sq` parameter. + The global `gko::solver::default_krylov_dim`, use instead `gko::solver::gmres_default_krylov_dim`. ### Added features + Adds a batch::BatchLinOp class that forms a base class for batched linear operators such as batched matrix formats, solver and preconditioners [#1379](#1379) + Adds a batch::MultiVector class that enables operations such as dot, norm, scale on batched vectors [#1371](#1371) + Adds a batch::Dense matrix format that stores batched dense matrices and provides gemv operations for these dense matrices. [#1413](#1413) + Adds a batch::Ell matrix format that stores batched Ell matrices and provides spmv operations for these batched Ell matrices. [#1416](#1416) [#1437](#1437) + Add a batch::Bicgstab solver (class, core, and reference kernels) that enables iterative solution of batched linear systems [#1438](#1438). + Add device kernels (CUDA, HIP, and DPCPP) for batch::Bicgstab solver. [#1443](#1443). + New MC64 reordering algorithm which optimizes the diagonal product or sum of a matrix by permuting the rows, and computes additional scaling factors for equilibriation [#1120](#1120) + New interface for (non-symmetric) permutation and scaled permutation of Dense and Csr matrices [#1415](#1415) + LU and Cholesky Factorizations can now be separated into their factors [#1432](#1432) + New symbolic LU factorization algorithm that is optimized for matrices with an almost-symmetric sparsity pattern [#1445](#1445) + Sorting kernels for SparsityCsr on all backends [#1343](#1343) + Allow passing pre-generated local solver as factory parameter for the distributed Schwarz preconditioner [#1426](#1426) + Add DPCPP kernels for Partition [#1034](#1034), and CSR's `check_diagonal_entries` and `add_scaled_identity` functionality [#1436](#1436) + Adds a helper function to create a partition based on either local sizes, or local ranges [#1227](#1227) + Add function to compute arithmetic mean of dense and distributed vectors [#1275](#1275) + Adds `icpx` compiler supports [#1350](#1350) + All backends can be built simultaneously [#1333](#1333) + Emits a CMake warning in downstream projects that use different compilers than the installed Ginkgo [#1372](#1372) + Reordering algorithms in sparse_blas benchmark [#1354](#1354) + Benchmarks gained an `-allocator` parameter to specify device allocators [#1385](#1385) + Benchmarks gained an `-input_matrix` parameter that initializes the input JSON based on the filename [#1387](#1387) + Benchmark inputs can now be reordered as a preprocessing step [#1408](#1408) ### Improvements + Significantly improve Cholesky factorization performance [#1366](#1366) + Improve parallel build performance [#1378](#1378) + Allow constrained parallel test execution using CTest resources [#1373](#1373) + Use arithmetic type more inside mixed precision ELL [#1414](#1414) + Most factory parameters of factory type no longer need to be constructed explicitly via `.on(exec)` [#1336](#1336) [#1439](#1439) + Improve ParILU(T)/ParIC(T) convergence by using more appropriate atomic operations [#1434](#1434) ### Fixes + Fix an over-allocation for OpenMP reductions [#1369](#1369) + Fix DPCPP's common-kernel reduction for empty input sizes [#1362](#1362) + Fix several typos in the API and documentation [#1348](#1348) + Fix inconsistent `Threads` between generations [#1388](#1388) + Fix benchmark median condition [#1398](#1398) + Fix HIP 5.6.0 compilation [#1411](#1411) + Fix missing destruction of rand_generator from cuda/hip [#1417](#1417) + Fix PAPI logger destruction order [#1419](#1419) + Fix TAU logger compilation [#1422](#1422) + Fix relative criterion to not iterate if the residual is already zero [#1079](#1079) + Fix memory_order invocations with C++20 changes [#1402](#1402) + Fix `check_diagonal_entries_exist` report correctly when only missing diagonal value in the last rows. [#1440](#1440) + Fix checking OpenMPI version in cross-compilation settings [#1446](#1446) + Fix false-positive deprecation warnings in Ginkgo, especially for the old Rcm (it doesn't emit deprecation warnings anymore as a result but is still considered deprecated) [#1444](#1444) ### Related PR: #1454

upsj added the 1:ST:need-feedback The PR is somewhat ready but feedback on a blocking topic is required before a proper review. label Oct 6, 2023

upsj requested a review from a team October 6, 2023 19:35

upsj self-assigned this Oct 6, 2023

upsj added 1:ST:ready-for-review This PR is ready for review and removed 1:ST:need-feedback The PR is somewhat ready but feedback on a blocking topic is required before a proper review. labels Oct 12, 2023

MarcelKoch self-requested a review October 13, 2023 06:56

MarcelKoch requested changes Oct 13, 2023

View reviewed changes

pratikvn reviewed Oct 13, 2023

View reviewed changes

upsj mentioned this pull request Oct 13, 2023

Refactor code duplications in dense_kernels tests #1423

Merged

yhmtsai reviewed Oct 16, 2023

View reviewed changes

common/unified/matrix/csr_kernels.cpp Outdated Show resolved Hide resolved

common/cuda_hip/matrix/csr_kernels.hpp.inc Show resolved Hide resolved

yhmtsai reviewed Oct 17, 2023

View reviewed changes

yhmtsai reviewed Oct 23, 2023

View reviewed changes

yhmtsai reviewed Oct 24, 2023

View reviewed changes

yhmtsai added this to the Release 1.7.0 milestone Oct 24, 2023

upsj force-pushed the scaled_permutation branch from 0779b54 to 8c4a46d Compare October 24, 2023 16:51

upsj requested review from MarcelKoch, yhmtsai and pratikvn October 24, 2023 16:51

MarcelKoch approved these changes Oct 25, 2023

View reviewed changes

core/matrix/csr.cpp Outdated Show resolved Hide resolved

reference/test/matrix/permutation.cpp Show resolved Hide resolved

reference/test/matrix/scaled_permutation.cpp Show resolved Hide resolved

pratikvn approved these changes Oct 25, 2023

View reviewed changes

yhmtsai reviewed Oct 25, 2023

View reviewed changes

yhmtsai approved these changes Oct 25, 2023

View reviewed changes

upsj and others added 12 commits October 27, 2023 09:33

improve-doc

288c437

doc-and-new-interface

62c88ff

add new interface and kernels

9845c2a

improve permutation interface consistency

e2048ee

- remove permute_mask - deprecate permute_mask and dim<2> parameters

swap order of scaling and permutation

1223cb8

add SYCL kernels

33c75f9

add permutation combination functions

caafb12

Format files

3fc05a1

Co-authored-by: Tobias Ribizel <[email protected]>

fix warning

6815279

fix incorrect combination function

9ea0076

also adds a few more tests to be sure.

upsj force-pushed the scaled_permutation branch from 5668746 to 67152e9 Compare October 27, 2023 13:44

upsj added 1:ST:ready-to-merge This PR is ready to merge. and removed 1:ST:ready-for-review This PR is ready for review labels Oct 27, 2023

upsj added 4 commits October 27, 2023 09:49

fix DPC++ compilation

ebb97af

fix test precision

202aba9

rename get_scale to get_scaling_factors

97a5f0c

fix misc test-related warnings

6c8ec9b

upsj merged commit 6d9e5fd into develop Oct 28, 2023
12 of 13 checks passed

upsj deleted the scaled_permutation branch October 28, 2023 03:02

tcojean mentioned this pull request Nov 6, 2023

Release 1.7.0 to master #1451

Merged

upsj mentioned this pull request Nov 13, 2023

Improve reordering interface #940

Closed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Scaled permutation and improved permutation interface #1415

Scaled permutation and improved permutation interface #1415

upsj commented Oct 6, 2023 •

edited

Loading

upsj commented Oct 12, 2023

sonarqubecloud bot commented Oct 13, 2023

MarcelKoch left a comment

pratikvn left a comment

yhmtsai Oct 17, 2023

upsj Oct 24, 2023 •

edited

Loading

yhmtsai Oct 17, 2023

yhmtsai Oct 17, 2023

upsj Oct 17, 2023

MarcelKoch Oct 17, 2023

yhmtsai Oct 17, 2023

upsj Oct 24, 2023

yhmtsai left a comment

yhmtsai Oct 24, 2023

upsj Oct 24, 2023

yhmtsai Oct 24, 2023

yhmtsai Oct 24, 2023

upsj Oct 24, 2023

MarcelKoch left a comment

pratikvn left a comment

pratikvn Oct 25, 2023

upsj Oct 26, 2023

yhmtsai left a comment

yhmtsai Oct 25, 2023 •

edited

Loading

yhmtsai Oct 25, 2023

upsj Oct 26, 2023

yhmtsai Oct 25, 2023

upsj Oct 26, 2023

yhmtsai Oct 27, 2023

upsj Oct 27, 2023

yhmtsai Oct 25, 2023

upsj Oct 26, 2023

yhmtsai left a comment •

edited

Loading

		* From the linear algebra perspective, with $P_{ij} = \delta_{i
		* \pi(i)}$, this represents the operation $P^{-1} A P^{-T}$.

		}


		TYPED_TEST(Permutation, AppliesRowAndColPermutationToDenseWithOneArray)

		run<const Dense<double>, const Dense<float>,
		const Dense<complex<double>>, const Dense<complex<float>>>(op, fn);

		std::unique_ptr<Permutation> combine(
		ptr_param<const Permutation> other) const;

Scaled permutation and improved permutation interface #1415

Scaled permutation and improved permutation interface #1415

Conversation

upsj commented Oct 6, 2023 • edited Loading

upsj commented Oct 12, 2023

sonarqubecloud bot commented Oct 13, 2023

MarcelKoch left a comment

Choose a reason for hiding this comment

pratikvn left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

upsj Oct 24, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

yhmtsai left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

MarcelKoch left a comment

Choose a reason for hiding this comment

pratikvn left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

yhmtsai left a comment

Choose a reason for hiding this comment

yhmtsai Oct 25, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

yhmtsai left a comment • edited Loading

Choose a reason for hiding this comment

upsj commented Oct 6, 2023 •

edited

Loading

upsj Oct 24, 2023 •

edited

Loading

yhmtsai Oct 25, 2023 •

edited

Loading

yhmtsai left a comment •

edited

Loading