Row gather linop #901

yhmtsai · 2021-10-09T10:10:58Z

This PR adds the row gather linop (Rowgatherer) interface to use dense row gather operation.
Also add mixed operation and the advanced row gather operation

codecov · 2021-10-09T14:22:38Z

Codecov Report

Merging #901 (164ddae) into develop (68b41ef) will increase coverage by 1.11%.
The diff coverage is 98.00%.

❗ Current head 164ddae differs from pull request most recent head 43d3e8e. Consider uploading reports for the commit 43d3e8e to get more accurate results

@@             Coverage Diff             @@
##           develop     #901      +/-   ##
===========================================
+ Coverage    92.27%   93.38%   +1.11%     
===========================================
  Files          476      470       -6     
  Lines        39410    38214    -1196     
===========================================
- Hits         36367    35688     -679     
+ Misses        3043     2526     -517

Impacted Files	Coverage Δ
core/device_hooks/common_kernels.inc.cpp	`0.00% <0.00%> (ø)`
core/solver/multigrid.cpp	`93.95% <ø> (+0.47%)`	⬆️
core/test/matrix/permutation.cpp	`98.68% <ø> (ø)`
include/ginkgo/core/matrix/dense.hpp	`96.32% <ø> (+0.73%)`	⬆️
include/ginkgo/core/matrix/permutation.hpp	`93.02% <ø> (ø)`
core/base/dispatch_helper.hpp	`66.66% <66.66%> (ø)`
core/test/matrix/row_gatherer.cpp	`97.82% <97.82%> (ø)`
common/unified/matrix/dense_kernels.cpp	`93.15% <100.00%> (-2.26%)`	⬇️
core/matrix/dense.cpp	`99.18% <100.00%> (+5.12%)`	⬆️
core/multigrid/amgx_pgm.cpp	`100.00% <100.00%> (ø)`
... and 135 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1cf901d...43d3e8e. Read the comment docs.

yhmtsai · 2021-10-13T08:46:24Z

Any feedback is welcome, especially for the name.

thoasm

My feedback so far.

common/unified/matrix/dense_kernels.cpp

thoasm · 2021-10-15T14:34:42Z

core/CMakeLists.txt

@@ -32,17 +32,18 @@ target_sources(ginkgo
    matrix/permutation.cpp
    matrix/sellp.cpp
    matrix/sparsity_csr.cpp
+    #matrix/row_gatherer.cpp


why is this commented? Because this is not finalized (you marked it as need-feedback)?

Thanks, I forget to delete it.
all row gather call dense function, so we do not really need the core.
still need that to instantiate all class

include/ginkgo/core/base/lin_op.hpp

include/ginkgo/core/matrix/dense.hpp

thoasm

Mostly formatting and documentation comments.

include/ginkgo/core/base/array.hpp

include/ginkgo/ginkgo.hpp

core/matrix/dense.cpp

include/ginkgo/core/matrix/row_gatherer.hpp

reference/test/matrix/dense_kernels.cpp

test/matrix/dense_kernels.cpp

greole

Not a full review, but some minor comments ...

greole · 2021-11-05T09:15:04Z

core/test/utils/assertions.hpp

@@ -624,6 +625,55 @@ ::testing::AssertionResult array_equal(const std::string& first_expression,
                                    second);
 }

+/**
+ * array_equal overload: first is ConstArrayView.


Could one use doxygens \overload command here, see https://www.doxygen.nl/manual/commands.html#cmdoverload?

it requires us to set SORT_MEMBER_DOCS = NO, to avoid the reordering issue

core/test/utils/assertions.hpp

reference/test/matrix/dense_kernels.cpp

greole · 2021-11-08T21:18:00Z

core/matrix/dense.cpp

@@ -1010,7 +1029,6 @@ std::unique_ptr<Dense<ValueType>> Dense<ValueType>::row_gather(
    return result;
 }

-


I believe this empty line should stay

It is overloaded, so one empty is enough

greole · 2021-11-08T21:18:46Z

core/matrix/dense.cpp

@@ -1022,20 +1040,125 @@ std::unique_ptr<Dense<ValueType>> Dense<ValueType>::row_gather(
    return result;
 }



And here is one empty line missing

Suggested change

It is overloaded, so one empty is enough

core/matrix/dense.cpp

core/test/matrix/row_gatherer.cpp

pratikvn

Some comments

core/multigrid/amgx_pgm.cpp

include/ginkgo/core/matrix/dense.hpp

thoasm

LGTM!

pratikvn

Only minor nits. Mostly LGTM!

pratikvn · 2021-11-22T15:28:52Z

core/test/matrix/row_gatherer.cpp

+using TestType = ::testing::Types<
+    std::tuple<float, float, gko::int32>, std::tuple<float, double, gko::int32>,
+    std::tuple<double, double, gko::int32>,
+    std::tuple<double, float, gko::int32>,
+    std::tuple<std::complex<float>, std::complex<float>, gko::int32>,
+    std::tuple<std::complex<float>, std::complex<double>, gko::int32>,
+    std::tuple<std::complex<double>, std::complex<double>, gko::int32>,
+    std::tuple<std::complex<double>, std::complex<float>, gko::int32>>;


Can probably be moved into core/test/utils.hpp ?

pratikvn · 2021-11-22T15:30:07Z

core/test/matrix/row_gatherer.cpp

+
+TYPED_TEST(RowGatherer, KnowsItsSizeAndValues)
+{
+    this->assert_equal_to_original_mtx(this->mtx.get());


Maybe to reduce the amount of code, GKO_ASSERT_MTX_NEAR may be sufficient ?

no, because the row gatherer does not have the write_to(matrix_data), can not use GKO_ASSERT_MTX_NEAR

Wouldn't it be useful to add that feature, then?

core/test/matrix/row_gatherer.cpp

include/ginkgo/core/base/lin_op.hpp

include/ginkgo/core/matrix/row_gatherer.hpp

upsj

LGTM for the most part, just a few points on the interface I would like to discuss

core/matrix/dense.cpp

core/solver/multigrid.cpp

core/test/utils/assertions.hpp

upsj · 2021-12-01T08:54:11Z

include/ginkgo/core/base/lin_op.hpp

+/**
+ * The RowGatherable is an interface that allows to get the row gather result of
+ * a LinOp.
+ *
+ * @tparam IndexType  the type of the indices used to extract/gather the rows
+ */
+template <typename IndexType>
+class RowGatherable {


Do we need a new Mixin here? row gather on CSR and Dense have two quite separate functions, and don't need to be made generic? Also the fact that std::unique_ptr is not covariant as a return type makes for a weird interface for the _linop overload.

Or do you need this for some mixed-precision application?

Yes, I think so? because the RowGatherer does not have value type information, I need this interface to call row_gather.
I thought CSR and Dense have the same interface for row_gather, doesn't it?
I do not understand the last sentence. could you describe it more detail?

you can't override a std::unique_ptr<LinOp> function() with a std::unique_ptr<Dense> function(). Covariance means that if your class is more specialized, your return type can also be more specialized. This is true for pointers, but not for smart pointers.

Good to know this! does the compiler show the override only partially message for this kind on pointers?
I think the return type is not considered in the function overload, so I still need to use two version for them?

include/ginkgo/core/matrix/row_gatherer.hpp

sonarqubecloud · 2021-12-01T17:33:39Z

Kudos, SonarCloud Quality Gate passed!

0 Bugs
0 Vulnerabilities
0 Security Hotspots
23 Code Smells

90.0% Coverage
2.0% Duplication

upsj

I forgot that I was blocking this, sorry. A few small nits, most of my comments have been adressed.
Still, I don't think we need a separate interface for behavior that is only implemented in Dense.
The advanced row_gather might also use a separate name like row_gather_add or similar, though I'm not 100% sure about that.
How about you call the dense kernels inside RowGatherer, which means you can use mixed_precision_dispatch to figure out their dimensions? That eliminates the need for an interface.
Also, a RowGatherer::write function might simplify things in the tests.

upsj · 2022-01-03T13:31:05Z

core/test/utils.hpp

 struct PairTypenameNameGenerator {
-    template <typename T>
+    template <typename T,
+              std::enable_if_t<std::tuple_size<T>::value == 2, bool> = true>


Nit: That would be a TupleTypenameNameGenerator, not Pair. No need to provide two overloads

upsj · 2022-01-03T13:33:40Z

include/ginkgo/core/matrix/row_gatherer.hpp

+    /**
+     * Returns a pointer to the array of row_gatherer_indices.
+     *
+     * @return the pointer to the row_gatherer_indices array.
+     */
+    index_type* get_row_gatherer_indices() noexcept
+    {
+        return row_gatherer_indices_.get_data();
+    }


Suggested change

/**

* Returns a pointer to the array of row_gatherer_indices.

*

* @return the pointer to the row_gatherer_indices array.

*/

index_type* get_row_gatherer_indices() noexcept

{

return row_gatherer_indices_.get_data();

}

/**

* Returns a pointer to the row index array.

*

* @return the pointer to the row index array.

*/

index_type* get_row_indices() noexcept

{

return row_gatherer_indices_.get_data();

}

upsj · 2022-01-03T13:33:50Z

include/ginkgo/core/matrix/row_gatherer.hpp

+    /**
+     * @copydoc get_row_gatherer_indices()
+     *
+     * @note This is the constant version of the function, which can be
+     *       significantly more memory efficient than the non-constant version,
+     *       so always prefer this version.
+     */
+    const index_type* get_const_row_gatherer_indices() const noexcept
+    {
+        return row_gatherer_indices_.get_const_data();
+    }


upsj · 2022-01-03T13:36:02Z

include/ginkgo/core/matrix/row_gatherer.hpp

+    template <typename IndicesArray>
+    RowGatherer(std::shared_ptr<const Executor> exec, const dim<2>& size,
+                IndicesArray&& row_gatherer_indices)
+        : EnableLinOp<RowGatherer>(exec, size),
+          row_gatherer_indices_{
+              exec, std::forward<IndicesArray>(row_gatherer_indices)}
+    {
+        GKO_ASSERT_EQ(size[0], row_gatherer_indices_.get_num_elems());
+    }


I know we do it like this in other places, but this makes the interface unnecessarily generic and forces it to be in the header.

Suggested change

template <typename IndicesArray>

RowGatherer(std::shared_ptr<const Executor> exec, const dim<2>& size,

IndicesArray&& row_gatherer_indices)

: EnableLinOp<RowGatherer>(exec, size),

row_gatherer_indices_{

exec, std::forward<IndicesArray>(row_gatherer_indices)}

{

GKO_ASSERT_EQ(size[0], row_gatherer_indices_.get_num_elems());

}

RowGatherer(std::shared_ptr<const Executor> exec, const dim<2>& size,

Array<IndexType> row_gatherer_indices)

: EnableLinOp<RowGatherer>(exec, size),

row_gatherer_indices_{

exec, std::move(row_gatherer_indices)}

{

GKO_ASSERT_EQ(size[0], row_gatherer_indices_.get_num_elems());

}

Ah nevermind, using this interface allows us to avoid a single copy if the array is on another executor than the RowGatherer

upsj · 2022-01-03T13:36:31Z

include/ginkgo/core/matrix/row_gatherer.hpp

+template <typename IndexType = int32>
+class RowGatherer : public EnableLinOp<RowGatherer<IndexType>>,
+                    public EnableCreateMethod<RowGatherer<IndexType>> {
+    friend class EnableCreateMethod<RowGatherer>;


Can also use a create_const method

upsj · 2022-01-03T13:36:59Z

include/ginkgo/core/matrix/row_gatherer.hpp

+     *
+     * @param exec  Executor associated to the LinOp
+     */
+    RowGatherer(std::shared_ptr<const Executor> exec)


All these member functions can be moved to the .cpp file

yhmtsai · 2022-01-03T16:51:33Z

@upsj

How about you call the dense kernels inside RowGatherer, which means you can use mixed_precision_dispatch to figure out their dimensions? That eliminates the need for an interface.

Do you mean call the dense kernels in the row gatherer apply function? I do not understand what dimensions to figure out. Could you give more detail? The interface can be used by Csr, too. Csr->row_gather(index) = Csr

Also, a RowGatherer::write function might simplify things in the tests.

Yes, but I am a little scared about setting the type which is not used by the class-self.

upsj · 2022-01-03T17:03:37Z

@yhmtsai
Sorry, I typed the wrong word, I meant precisions, not dimensions, i.e. the value type.

Since this PR only involves Dense, I don't see the advantage of this approach. Adding Csr would introduce another feature hidden behind an apply(...) function like #942, which I would like to avoid as much as possible.

upsj · 2022-01-03T17:04:23Z

About write: Fft also declares a value_type, even though it doesn't have values strictly speaking.

yhmtsai · 2022-01-04T14:12:44Z

the precision is also dispatched in the row_gather of dense.
Yes, if we only restrict the apply in dense, do not need the interface.
I stand at the opposite side. I think it should be allowed for any kind of LinOp when we have the implementation.
The complex interface (dense input/output additionally) is to fit the original dense interface.
about the row gather, if we only allow dense case, what should we have the interface for Csr case when we need it?
dense can use row_gather->apply(dense, dense) or dense->row_gather(index)
but csr can only use csr->row_gather(index)

yhmtsai · 2022-02-09T08:50:35Z

@upsj row gather only consider dense now, so apply will go over all possible dense type to call row_gather

Co-authored-by: Thomas Grützmacher <thomas.gruetzmacher@kit.edu>

add overload of array_equal for ConstArrayView

Co-authored-by: Gregor Olenik <gregor.olenik@web.de> Co-authored-by: Thomas Grützmacher <thomas.gruetzmacher@kit.edu>

Co-authored-by: Thomas Grützmacher <thomas.gruetzmacher@kit.edu>

Co-authored-by: Pratik Nayak <pratikvn@protonmail.com> Co-authored-by: Tobias Ribizel <ribizel@kit.edu>

Co-authored-by: Tobias Ribizel <ribizel@kit.edu>

… create_const Co-authored-by: Tobias Ribizel <ribizel@kit.edu>

ginkgo-bot · 2022-02-10T23:20:22Z

Note: This PR changes the Ginkgo ABI:

Functions changes summary: 8 Removed, 60 Changed (154 filtered out), 306 Added functions
Variables changes summary: 0 Removed, 0 Changed, 0 Added variable

For details check the full ABI diff under Artifacts here

sonarqubecloud · 2022-02-11T09:34:58Z

Kudos, SonarCloud Quality Gate passed!

0 Bugs
0 Vulnerabilities
0 Security Hotspots
0 Code Smells

No Coverage information
No Duplication information

Advertise release 1.5.0 and last changes + Add changelog, + Update third party libraries + A small fix to a CMake file See PR: #1195 The Ginkgo team is proud to announce the new Ginkgo minor release 1.5.0. This release brings many important new features such as: - MPI-based multi-node support for all matrix formats and most solvers; - full DPC++/SYCL support, - functionality and interface for GPU-resident sparse direct solvers, - an interface for wrapping solvers with scaling and reordering applied, - a new algebraic Multigrid solver/preconditioner, - improved mixed-precision support, - support for device matrix assembly, and much 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 3.13+ + C++14 compliant compiler + Linux and macOS + GCC: 5.5+ + clang: 3.9+ + Intel compiler: 2018+ + Apple LLVM: 8.0+ + NVHPC: 22.7+ + Cray Compiler: 14.0.1+ + CUDA module: CUDA 9.2+ or NVHPC 22.7+ + HIP module: ROCm 4.0+ + DPC++ module: Intel OneAPI 2021.3 with oneMKL and oneDPL. Set the CXX compiler to `dpcpp`. + Windows + MinGW and Cygwin: GCC 5.5+ + Microsoft Visual Studio: VS 2019 + CUDA module: CUDA 9.2+, Microsoft Visual Studio + OpenMP module: MinGW or Cygwin. Algorithm and important feature additions: + Add MPI-based multi-node for all matrix formats and solvers (except GMRES and IDR). ([#676](#676), [#908](#908), [#909](#909), [#932](#932), [#951](#951), [#961](#961), [#971](#971), [#976](#976), [#985](#985), [#1007](#1007), [#1030](#1030), [#1054](#1054), [#1100](#1100), [#1148](#1148)) + Porting the remaining algorithms (preconditioners like ISAI, Jacobi, Multigrid, ParILU(T) and ParIC(T)) to DPC++/SYCL, update to SYCL 2020, and improve support and performance ([#896](#896), [#924](#924), [#928](#928), [#929](#929), [#933](#933), [#943](#943), [#960](#960), [#1057](#1057), [#1110](#1110), [#1142](#1142)) + Add a Sparse Direct interface supporting GPU-resident numerical LU factorization, symbolic Cholesky factorization, improved triangular solvers, and more ([#957](#957), [#1058](#1058), [#1072](#1072), [#1082](#1082)) + Add a ScaleReordered interface that can wrap solvers and automatically apply reorderings and scalings ([#1059](#1059)) + Add a Multigrid solver and improve the aggregation based PGM coarsening scheme ([#542](#542), [#913](#913), [#980](#980), [#982](#982), [#986](#986)) + Add infrastructure for unified, lambda-based, backend agnostic, kernels and utilize it for some simple kernels ([#833](#833), [#910](#910), [#926](#926)) + Merge different CUDA, HIP, DPC++ and OpenMP tests under a common interface ([#904](#904), [#973](#973), [#1044](#1044), [#1117](#1117)) + Add a device_matrix_data type for device-side matrix assembly ([#886](#886), [#963](#963), [#965](#965)) + Add support for mixed real/complex BLAS operations ([#864](#864)) + Add a FFT LinOp for all but DPC++/SYCL ([#701](#701)) + Add FBCSR support for NVIDIA and AMD GPUs and CPUs with OpenMP ([#775](#775)) + Add CSR scaling ([#848](#848)) + Add array::const_view and equivalent to create constant matrices from non-const data ([#890](#890)) + Add a RowGatherer LinOp supporting mixed precision to gather dense matrix rows ([#901](#901)) + Add mixed precision SparsityCsr SpMV support ([#970](#970)) + Allow creating CSR submatrix including from (possibly discontinuous) index sets ([#885](#885), [#964](#964)) + Add a scaled identity addition (M <- aI + bM) feature interface and impls for Csr and Dense ([#942](#942)) Deprecations and important changes: + Deprecate AmgxPgm in favor of the new Pgm name. ([#1149](#1149)). + Deprecate specialized residual norm classes in favor of a common `ResidualNorm` class ([#1101](#1101)) + Deprecate CamelCase non-polymorphic types in favor of snake_case versions (like array, machine_topology, uninitialized_array, index_set) ([#1031](#1031), [#1052](#1052)) + Bug fix: restrict gko::share to rvalue references (*possible interface break*) ([#1020](#1020)) + Bug fix: when using cuSPARSE's triangular solvers, specifying the factory parameter `num_rhs` is now required when solving for more than one right-hand side, otherwise an exception is thrown ([#1184](#1184)). + Drop official support for old CUDA < 9.2 ([#887](#887)) Improved performance additions: + Reuse tmp storage in reductions in solvers and add a mutable workspace to all solvers ([#1013](#1013), [#1028](#1028)) + Add HIP unsafe atomic option for AMD ([#1091](#1091)) + Prefer vendor implementations for Dense dot, conj_dot and norm2 when available ([#967](#967)). + Tuned OpenMP SellP, COO, and ELL SpMV kernels for a small number of RHS ([#809](#809)) Fixes: + Fix various compilation warnings ([#1076](#1076), [#1183](#1183), [#1189](#1189)) + Fix issues with hwloc-related tests ([#1074](#1074)) + Fix include headers for GCC 12 ([#1071](#1071)) + Fix for simple-solver-logging example ([#1066](#1066)) + Fix for potential memory leak in Logger ([#1056](#1056)) + Fix logging of mixin classes ([#1037](#1037)) + Improve value semantics for LinOp types, like moved-from state in cross-executor copy/clones ([#753](#753)) + Fix some matrix SpMV and conversion corner cases ([#905](#905), [#978](#978)) + Fix uninitialized data ([#958](#958)) + Fix CUDA version requirement for cusparseSpSM ([#953](#953)) + Fix several issues within bash-script ([#1016](#1016)) + Fixes for `NVHPC` compiler support ([#1194](#1194)) Other additions: + Simplify and properly name GMRES kernels ([#861](#861)) + Improve pkg-config support for non-CMake libraries ([#923](#923), [#1109](#1109)) + Improve gdb pretty printer ([#987](#987), [#1114](#1114)) + Add a logger highlighting inefficient allocation and copy patterns ([#1035](#1035)) + Improved and optimized test random matrix generation ([#954](#954), [#1032](#1032)) + Better CSR strategy defaults ([#969](#969)) + Add `move_from` to `PolymorphicObject` ([#997](#997)) + Remove unnecessary device_guard usage ([#956](#956)) + Improvements to the generic accessor for mixed-precision ([#727](#727)) + Add a naive lower triangular solver implementation for CUDA ([#764](#764)) + Add support for int64 indices from CUDA 11 onward with SpMV and SpGEMM ([#897](#897)) + Add a L1 norm implementation ([#900](#900)) + Add reduce_add for arrays ([#831](#831)) + Add utility to simplify Dense View creation from an existing Dense vector ([#1136](#1136)). + Add a custom transpose implementation for Fbcsr and Csr transpose for unsupported vendor types ([#1123](#1123)) + Make IDR random initilization deterministic ([#1116](#1116)) + Move the algorithm choice for triangular solvers from Csr::strategy_type to a factory parameter ([#1088](#1088)) + Update CUDA archCoresPerSM ([#1175](#1116)) + Add kernels for Csr sparsity pattern lookup ([#994](#994)) + Differentiate between structural and numerical zeros in Ell/Sellp ([#1027](#1027)) + Add a binary IO format for matrix data ([#984](#984)) + Add a tuple zip_iterator implementation ([#966](#966)) + Simplify kernel stubs and declarations ([#888](#888)) + Simplify GKO_REGISTER_OPERATION with lambdas ([#859](#859)) + Simplify copy to device in tests and examples ([#863](#863)) + More verbose output to array assertions ([#858](#858)) + Allow parallel compilation for Jacobi kernels ([#871](#871)) + Change clang-format pointer alignment to left ([#872](#872)) + Various improvements and fixes to the benchmarking framework ([#750](#750), [#759](#759), [#870](#870), [#911](#911), [#1033](#1033), [#1137](#1137)) + Various documentation improvements ([#892](#892), [#921](#921), [#950](#950), [#977](#977), [#1021](#1021), [#1068](#1068), [#1069](#1069), [#1080](#1080), [#1081](#1081), [#1108](#1108), [#1153](#1153), [#1154](#1154)) + Various CI improvements ([#868](#868), [#874](#874), [#884](#884), [#889](#889), [#899](#899), [#903](#903), [#922](#922), [#925](#925), [#930](#930), [#936](#936), [#937](#937), [#958](#958), [#882](#882), [#1011](#1011), [#1015](#1015), [#989](#989), [#1039](#1039), [#1042](#1042), [#1067](#1067), [#1073](#1073), [#1075](#1075), [#1083](#1083), [#1084](#1084), [#1085](#1085), [#1139](#1139), [#1178](#1178), [#1187](#1187))

yhmtsai self-assigned this Oct 9, 2021

yhmtsai 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 11, 2021

yhmtsai requested a review from a team October 11, 2021 12:57

thoasm self-requested a review October 15, 2021 15:34

thoasm reviewed Oct 20, 2021

View reviewed changes

yhmtsai force-pushed the row_gather_linop branch 2 times, most recently from aa287b3 to 7cc8fad Compare October 25, 2021 18:35

yhmtsai 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 25, 2021

thoasm reviewed Nov 5, 2021

View reviewed changes

greole reviewed Nov 8, 2021

View reviewed changes

pratikvn reviewed Nov 11, 2021

View reviewed changes

core/multigrid/amgx_pgm.cpp Show resolved Hide resolved

include/ginkgo/core/matrix/dense.hpp Show resolved Hide resolved

include/ginkgo/core/matrix/dense.hpp Show resolved Hide resolved

yhmtsai force-pushed the row_gather_linop branch from 7cc8fad to 193d7ac Compare November 18, 2021 18:13

thoasm approved these changes Nov 19, 2021

View reviewed changes

pratikvn approved these changes Nov 22, 2021

View reviewed changes

upsj requested changes Dec 1, 2021

View reviewed changes

yhmtsai force-pushed the row_gather_linop branch from 97c838f to 164ddae Compare December 1, 2021 13:48

upsj reviewed Jan 3, 2022

View reviewed changes

yhmtsai force-pushed the row_gather_linop branch 2 times, most recently from 0446199 to dff2afb Compare February 8, 2022 13:00

upsj approved these changes Feb 9, 2022

View reviewed changes

yhmtsai force-pushed the row_gather_linop branch 2 times, most recently from 78649d3 to e2b9a95 Compare February 10, 2022 12:24

yhmtsai 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 Feb 10, 2022

yhmtsai force-pushed the row_gather_linop branch from b938b6d to 21dad5a Compare February 10, 2022 13:53

tcojean added this to the Ginkgo 1.5.0 milestone Feb 10, 2022

yhmtsai and others added 11 commits February 10, 2022 23:41

Add RowGather, mixed apply, corresponding test

0ace524

add advanced row gather

45dcc84

update documentation, use consistent variable name

9e1056b

Co-authored-by: Thomas Grützmacher <thomas.gruetzmacher@kit.edu>

use RowGatherer, add ConstArrayView copy_to_array

93afbb0

add overload of array_equal for ConstArrayView

update documentation and format

766bf01

Co-authored-by: Gregor Olenik <gregor.olenik@web.de> Co-authored-by: Thomas Grützmacher <thomas.gruetzmacher@kit.edu>

rename row_gather_index -> row_gatherer_indices

47f55fa

Co-authored-by: Thomas Grützmacher <thomas.gruetzmacher@kit.edu>

update format/doc and remove duplicated function

df0d002

Co-authored-by: Pratik Nayak <pratikvn@protonmail.com> Co-authored-by: Tobias Ribizel <ribizel@kit.edu>

move the dispatch and TwoValueIndexType

cc6b504

Co-authored-by: Pratik Nayak <pratikvn@protonmail.com> Co-authored-by: Tobias Ribizel <ribizel@kit.edu>

update pair to tuple name generator

5c1673f

Co-authored-by: Tobias Ribizel <ribizel@kit.edu>

rowgatherer only support dense, update doc, remove interface, and add…

6976cc7

… create_const Co-authored-by: Tobias Ribizel <ribizel@kit.edu>

change row_indices to row_idxs

43d3e8e

yhmtsai force-pushed the row_gather_linop branch from 21dad5a to 43d3e8e Compare February 10, 2022 22:43

yhmtsai merged commit e1f17e4 into develop Feb 11, 2022

yhmtsai deleted the row_gather_linop branch February 11, 2022 05:53

		@@ -1010,7 +1029,6 @@ std::unique_ptr<Dense<ValueType>> Dense<ValueType>::row_gather(
		return result;
		}

		@@ -1022,20 +1040,125 @@ std::unique_ptr<Dense<ValueType>> Dense<ValueType>::row_gather(
		return result;
		}

Row gather linop #901

Row gather linop #901

Conversation

yhmtsai commented Oct 9, 2021

codecov bot commented Oct 9, 2021 • edited Loading

Codecov Report

yhmtsai commented Oct 13, 2021

thoasm left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

yhmtsai Oct 21, 2021 • edited Loading

Choose a reason for hiding this comment

thoasm left a comment

Choose a reason for hiding this comment

greole 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

Choose a reason for hiding this comment

pratikvn left a comment

Choose a reason for hiding this comment

thoasm 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

Choose a reason for hiding this comment

Choose a reason for hiding this comment

upsj 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

upsj Dec 1, 2021 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

sonarqubecloud bot commented Dec 1, 2021

upsj left a comment • 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

yhmtsai commented Jan 3, 2022 • edited Loading

upsj commented Jan 3, 2022

upsj commented Jan 3, 2022

yhmtsai commented Jan 4, 2022

yhmtsai commented Feb 9, 2022

ginkgo-bot commented Feb 10, 2022

sonarqubecloud bot commented Feb 11, 2022

codecov bot commented Oct 9, 2021 •

edited

Loading

yhmtsai Oct 21, 2021 •

edited

Loading

upsj Dec 1, 2021 •

edited

Loading

upsj left a comment •

edited

Loading

yhmtsai commented Jan 3, 2022 •

edited

Loading