Public interface and basic reference kernels for fixed-size block CSR #671

Slaedr · 2020-11-24T15:04:10Z

Adds a fixed block CSR matrix format. "Fixed" refers to the fact that the block size is uniform for the entire matrix, as opposed to what is sometimes called variable BCSR.

This PR adds the core functions and reference kernels.

TODO:

Squash some unnecessary commits

tcojean

Nice work! I have multiple comments. I concentrate on trimming this PR for now and did not review in depth several implemented kernels/tests.

For now, I really would like to give this PR a big haircut and remove all the dead code, I did not count but there is thousands of line of codes which are effectively useless. I tried as much as possible to highlight everything to be removed, but sometimes (as the non implemented module tests and kernels) that would be pretty much the whole file so I do not highlight the portions to remove.

Most of them I think stem directly from the script conversion, since CSR is a very bloated class and has pretty much every feature that Ginkgo offers for matrix formats. I think if the code comes from a script, that means it's very easy to get it back (a few copy paste and search replace), and therefore there is no need to keep such codes. It's even worst in the sense that if the code was uncommented, it wouldn't even work because that is code designed for CSR itself, so there is little information provided by the code. Here is an overview of what I think should be thrown out:

Kernels which are not used anywhere but still instantiated. It's also unclear whether they will ever be implemented, or when (a lot of conversions fall here).
Code generated by the script such as all the TODO kernels.
Related tests of commented kernels.
A few header files.

core/CMakeLists.txt

cuda/test/matrix/CMakeLists.txt

hip/test/matrix/CMakeLists.txt

omp/components/fbcsr_spgeam.hpp

omp/test/matrix/CMakeLists.txt

reference/test/matrix/fbcsr_kernels.cpp

tcojean

Two small comments to hopefully fix the compilation issues.

hip/matrix/fbcsr_kernels.hip.cpp

core/test/matrix/fbcsr_builder.cpp

core/test/matrix/fbcsr.cpp

reference/test/matrix/fbcsr_kernels.cpp

tcojean · 2020-11-30T12:05:14Z

label!

tcojean · 2020-11-30T12:05:21Z

check-format!

github-actions · 2020-11-30T12:06:41Z

Error: The following files need to be formatted:

core/components/fixed_block.hpp
core/test/matrix/fbcsr.cpp
core/test/matrix/fbcsr_sample.cpp
core/test/matrix/fbcsr_sample.hpp
include/ginkgo/core/base/blockutils.hpp
include/ginkgo/core/matrix/fbcsr.hpp
include/ginkgo/core/matrix/matrix_strategies.hpp
include/ginkgo/core/matrix/sparsity_csr.hpp

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

pratikvn · 2020-11-30T12:36:47Z

I think the reference and core labels are more accurate here than the all label. Even though the .cu and .hip files are also being added, the kernel implementations are only for reference

upsj · 2020-11-30T12:39:29Z

That's true. Unfortunately, there is no way to determine which files are scaffolding and which contain actual content based on the diff metadata alone, so the question is: Do we want any changes on the label! functionality or leave it as-is?

pratikvn · 2020-11-30T12:57:52Z

I think as long as we can modify it ourselves and the bot does not automatically override our selection, it should be fine. I dont think there is any reason to keep calling the label! again once it has been assigned by the bot at creation time. If needed we can add/modify the labels.

- blockutils.hpp is removed for now, its contents moved to pre-existing files - added braces after control-flow statements - More public functions in class Fbcsr are marked noexecpt

Added tests for BlockSizeError, getNumBlocks and especially FbcsrSample classes.

Co-authored-by: Terry Cojean <[email protected]>

Also in fbcsr.hpp, detail::get_num_blocks is now declared inline.

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

- removal of unnecessary lines of code - snake case convention for non-type template param Co-authored-by: Pratik Nayak <[email protected]>

Co-authored-by: Thomas Grützmacher <[email protected]>

pratikvn

LGTM! I think for some things that we discussed, can you create one issue with all FBCsr TODOs so that we don't lose it in the review comments ? I have marked some of them with See this

pratikvn · 2021-03-29T07:55:41Z

reference/matrix/fbcsr_kernels.cpp

+ for (int ib = 0; ib < blksz; ib++) {
+ for (int jb = 0; jb < blksz; jb++) {
+ cvalues(dest_idx, ib, jb) =
+ op(transpose_blocks ? rvalues(i, jb, ib)


pratikvn · 2021-03-29T07:59:54Z

include/ginkgo/core/matrix/fbcsr.hpp

+ public ConvertibleTo<Fbcsr<next_precision<ValueType>, IndexType>>,
+ public ConvertibleTo<Dense<ValueType>>,
+ public ConvertibleTo<Csr<ValueType, IndexType>>,
+ public ConvertibleTo<SparsityCsr<ValueType, IndexType>>,


pratikvn · 2021-03-29T08:46:32Z

core/matrix/fbcsr.cpp

+ const auto nnz = static_cast<index_type>(data.nonzeros.size());
+ const int bs = this->bs_;
+
+ using Block_t = detail::DenseBlock<value_type>;


sonarcloud · 2021-03-29T09:11:26Z

Kudos, SonarCloud Quality Gate passed!

0 Bugs
0 Vulnerabilities
0 Security Hotspots
80 Code Smells

84.0% Coverage
1.5% Duplication

Slaedr · 2021-03-29T11:46:58Z

@pratikvn

The transpose_blocks is there just to enable a future kernel that might require only transposing the sparsity pattern and not the blocks themselves; I don't see the harm in letting that be there. Since we don't know if a such a kernel will actually be needed, I don't think it's necessary to raise an issue about this.
We discussed the SparsityCsr; I believe it's fine to use it for the sparsity pattern of block CSR as well, at least for Fbcsr. It makes sense because, I think, any operation that makes sense on the sparsity pattern of a CSR matrix would also make sense on that of a fixed-block CSR matrix. It's just up to the Fbcsr class on how to use the results. So for example, the output of RCM reordering of a SparsityCsr would be used differently by Csr and Fbcsr, so the (block) permute functions in Fbcsr would be different. As @upsj said, we might need a BlockPermutable interface. Should we create an issue about block permutations now? I guess since the permute-related functions are currently unimplemented for Fbcsr, it's not really necessary right now. We will get to it when we implement those. ~~One option would be to remove all permutation-related stuff from Fbcsr for now. Would that be better?~~ Never mind, I've already removed it.
About the detail::DenseBlock, I can create an issue about it.

Ginkgo release 1.4.0 The Ginkgo team is proud to announce the new Ginkgo minor release 1.4.0. This release brings most of the Ginkgo functionality to the Intel DPC++ ecosystem which enables Intel-GPU and CPU execution. The only Ginkgo features which have not been ported yet are some preconditioners. Ginkgo's mixed-precision support is greatly enhanced thanks to: 1. The new Accessor concept, which allows writing kernels featuring on-the-fly memory compression, among other features. The accessor can be used as header-only, see the [accessor BLAS benchmarks repository](https://github.com/ginkgo-project/accessor-BLAS/tree/develop) as a usage example. 2. All LinOps now transparently support mixed-precision execution. By default, this is done through a temporary copy which may have a performance impact but already allows mixed-precision research. Native mixed-precision ELL kernels are implemented which do not see this cost. The accessor is also leveraged in a new CB-GMRES solver which allows for performance improvements by compressing the Krylov basis vectors. Many other features have been added to Ginkgo, such as reordering support, a new IDR solver, Incomplete Cholesky preconditioner, matrix assembly support (only CPU for now), machine topology information, and more! Supported systems and requirements: + For all platforms, cmake 3.13+ + C++14 compliant compiler + Linux and MacOS + gcc: 5.3+, 6.3+, 7.3+, all versions after 8.1+ + clang: 3.9+ + Intel compiler: 2018+ + Apple LLVM: 8.0+ + CUDA module: CUDA 9.0+ + HIP module: ROCm 3.5+ + DPC++ module: Intel OneAPI 2021.3. Set the CXX compiler to `dpcpp`. + Windows + MinGW and Cygwin: gcc 5.3+, 6.3+, 7.3+, all versions after 8.1+ + Microsoft Visual Studio: VS 2019 + CUDA module: CUDA 9.0+, Microsoft Visual Studio + OpenMP module: MinGW or Cygwin. Algorithm and important feature additions: + Add a new DPC++ Executor for SYCL execution and other base utilities [#648](#648), [#661](#661), [#757](#757), [#832](#832) + Port matrix formats, solvers and related kernels to DPC++. For some kernels, also make use of a shared kernel implementation for all executors (except Reference). [#710](#710), [#799](#799), [#779](#779), [#733](#733), [#844](#844), [#843](#843), [#789](#789), [#845](#845), [#849](#849), [#855](#855), [#856](#856) + Add accessors which allow multi-precision kernels, among other things. [#643](#643), [#708](#708) + Add support for mixed precision operations through apply in all LinOps. [#677](#677) + Add incomplete Cholesky factorizations and preconditioners as well as some improvements to ILU. [#672](#672), [#837](#837), [#846](#846) + Add an AMGX implementation and kernels on all devices but DPC++. [#528](#528), [#695](#695), [#860](#860) + Add a new mixed-precision capability solver, Compressed Basis GMRES (CB-GMRES). [#693](#693), [#763](#763) + Add the IDR(s) solver. [#620](#620) + Add a new fixed-size block CSR matrix format (for the Reference executor). [#671](#671), [#730](#730) + Add native mixed-precision support to the ELL format. [#717](#717), [#780](#780) + Add Reverse Cuthill-McKee reordering [#500](#500), [#649](#649) + Add matrix assembly support on CPUs. [#644](#644) + Extends ISAI from triangular to general and spd matrices. [#690](#690) Other additions: + Add the possibility to apply real matrices to complex vectors. [#655](#655), [#658](#658) + Add functions to compute the absolute of a matrix format. [#636](#636) + Add symmetric permutation and improve existing permutations. [#684](#684), [#657](#657), [#663](#663) + Add a MachineTopology class with HWLOC support [#554](#554), [#697](#697) + Add an implicit residual norm criterion. [#702](#702), [#818](#818), [#850](#850) + Row-major accessor is generalized to more than 2 dimensions and a new "block column-major" accessor has been added. [#707](#707) + Add an heat equation example. [#698](#698), [#706](#706) + Add ccache support in CMake and CI. [#725](#725), [#739](#739) + Allow tuning and benchmarking variables non intrusively. [#692](#692) + Add triangular solver benchmark [#664](#664) + Add benchmarks for BLAS operations [#772](#772), [#829](#829) + Add support for different precisions and consistent index types in benchmarks. [#675](#675), [#828](#828) + Add a Github bot system to facilitate development and PR management. [#667](#667), [#674](#674), [#689](#689), [#853](#853) + Add Intel (DPC++) CI support and enable CI on HPC systems. [#736](#736), [#751](#751), [#781](#781) + Add ssh debugging for Github Actions CI. [#749](#749) + Add pipeline segmentation for better CI speed. [#737](#737) Changes: + Add a Scalar Jacobi specialization and kernels. [#808](#808), [#834](#834), [#854](#854) + Add implicit residual log for solvers and benchmarks. [#714](#714) + Change handling of the conjugate in the dense dot product. [#755](#755) + Improved Dense stride handling. [#774](#774) + Multiple improvements to the OpenMP kernels performance, including COO, an exclusive prefix sum, and more. [#703](#703), [#765](#765), [#740](#740) + Allow specialization of submatrix and other dense creation functions in solvers. [#718](#718) + Improved Identity constructor and treatment of rectangular matrices. [#646](#646) + Allow CUDA/HIP executors to select allocation mode. [#758](#758) + Check if executors share the same memory. [#670](#670) + Improve test install and smoke testing support. [#721](#721) + Update the JOSS paper citation and add publications in the documentation. [#629](#629), [#724](#724) + Improve the version output. [#806](#806) + Add some utilities for dim and span. [#821](#821) + Improved solver and preconditioner benchmarks. [#660](#660) + Improve benchmark timing and output. [#669](#669), [#791](#791), [#801](#801), [#812](#812) Fixes: + Sorting fix for the Jacobi preconditioner. [#659](#659) + Also log the first residual norm in CGS [#735](#735) + Fix BiCG and HIP CSR to work with complex matrices. [#651](#651) + Fix Coo SpMV on strided vectors. [#807](#807) + Fix segfault of extract_diagonal, add short-and-fat test. [#769](#769) + Fix device_reset issue by moving counter/mutex to device. [#810](#810) + Fix `EnableLogging` superclass. [#841](#841) + Support ROCm 4.1.x and breaking HIP_PLATFORM changes. [#726](#726) + Decreased test size for a few device tests. [#742](#742) + Fix multiple issues with our CMake HIP and RPATH setup. [#712](#712), [#745](#745), [#709](#709) + Cleanup our CMake installation step. [#713](#713) + Various simplification and fixes to the Windows CMake setup. [#720](#720), [#785](#785) + Simplify third-party integration. [#786](#786) + Improve Ginkgo device arch flags management. [#696](#696) + Other fixes and improvements to the CMake setup. [#685](#685), [#792](#792), [#705](#705), [#836](#836) + Clarification of dense norm documentation [#784](#784) + Various development tools fixes and improvements [#738](#738), [#830](#830), [#840](#840) + Make multiple operators/constructors explicit. [#650](#650), [#761](#761) + Fix some issues, memory leaks and warnings found by MSVC. [#666](#666), [#731](#731) + Improved solver memory estimates and consistent iteration counts [#691](#691) + Various logger improvements and fixes [#728](#728), [#743](#743), [#754](#754) + Fix for ForwardIterator requirements in iterator_factory. [#665](#665) + Various benchmark fixes. [#647](#647), [#673](#673), [#722](#722) + Various CI fixes and improvements. [#642](#642), [#641](#641), [#795](#795), [#783](#783), [#793](#793), [#852](#852) Related PR: #857

Release 1.4.0 to master The Ginkgo team is proud to announce the new Ginkgo minor release 1.4.0. This release brings most of the Ginkgo functionality to the Intel DPC++ ecosystem which enables Intel-GPU and CPU execution. The only Ginkgo features which have not been ported yet are some preconditioners. Ginkgo's mixed-precision support is greatly enhanced thanks to: 1. The new Accessor concept, which allows writing kernels featuring on-the-fly memory compression, among other features. The accessor can be used as header-only, see the [accessor BLAS benchmarks repository](https://github.com/ginkgo-project/accessor-BLAS/tree/develop) as a usage example. 2. All LinOps now transparently support mixed-precision execution. By default, this is done through a temporary copy which may have a performance impact but already allows mixed-precision research. Native mixed-precision ELL kernels are implemented which do not see this cost. The accessor is also leveraged in a new CB-GMRES solver which allows for performance improvements by compressing the Krylov basis vectors. Many other features have been added to Ginkgo, such as reordering support, a new IDR solver, Incomplete Cholesky preconditioner, matrix assembly support (only CPU for now), machine topology information, and more! Supported systems and requirements: + For all platforms, cmake 3.13+ + C++14 compliant compiler + Linux and MacOS + gcc: 5.3+, 6.3+, 7.3+, all versions after 8.1+ + clang: 3.9+ + Intel compiler: 2018+ + Apple LLVM: 8.0+ + CUDA module: CUDA 9.0+ + HIP module: ROCm 3.5+ + DPC++ module: Intel OneAPI 2021.3. Set the CXX compiler to `dpcpp`. + Windows + MinGW and Cygwin: gcc 5.3+, 6.3+, 7.3+, all versions after 8.1+ + Microsoft Visual Studio: VS 2019 + CUDA module: CUDA 9.0+, Microsoft Visual Studio + OpenMP module: MinGW or Cygwin. Algorithm and important feature additions: + Add a new DPC++ Executor for SYCL execution and other base utilities [#648](#648), [#661](#661), [#757](#757), [#832](#832) + Port matrix formats, solvers and related kernels to DPC++. For some kernels, also make use of a shared kernel implementation for all executors (except Reference). [#710](#710), [#799](#799), [#779](#779), [#733](#733), [#844](#844), [#843](#843), [#789](#789), [#845](#845), [#849](#849), [#855](#855), [#856](#856) + Add accessors which allow multi-precision kernels, among other things. [#643](#643), [#708](#708) + Add support for mixed precision operations through apply in all LinOps. [#677](#677) + Add incomplete Cholesky factorizations and preconditioners as well as some improvements to ILU. [#672](#672), [#837](#837), [#846](#846) + Add an AMGX implementation and kernels on all devices but DPC++. [#528](#528), [#695](#695), [#860](#860) + Add a new mixed-precision capability solver, Compressed Basis GMRES (CB-GMRES). [#693](#693), [#763](#763) + Add the IDR(s) solver. [#620](#620) + Add a new fixed-size block CSR matrix format (for the Reference executor). [#671](#671), [#730](#730) + Add native mixed-precision support to the ELL format. [#717](#717), [#780](#780) + Add Reverse Cuthill-McKee reordering [#500](#500), [#649](#649) + Add matrix assembly support on CPUs. [#644](#644) + Extends ISAI from triangular to general and spd matrices. [#690](#690) Other additions: + Add the possibility to apply real matrices to complex vectors. [#655](#655), [#658](#658) + Add functions to compute the absolute of a matrix format. [#636](#636) + Add symmetric permutation and improve existing permutations. [#684](#684), [#657](#657), [#663](#663) + Add a MachineTopology class with HWLOC support [#554](#554), [#697](#697) + Add an implicit residual norm criterion. [#702](#702), [#818](#818), [#850](#850) + Row-major accessor is generalized to more than 2 dimensions and a new "block column-major" accessor has been added. [#707](#707) + Add an heat equation example. [#698](#698), [#706](#706) + Add ccache support in CMake and CI. [#725](#725), [#739](#739) + Allow tuning and benchmarking variables non intrusively. [#692](#692) + Add triangular solver benchmark [#664](#664) + Add benchmarks for BLAS operations [#772](#772), [#829](#829) + Add support for different precisions and consistent index types in benchmarks. [#675](#675), [#828](#828) + Add a Github bot system to facilitate development and PR management. [#667](#667), [#674](#674), [#689](#689), [#853](#853) + Add Intel (DPC++) CI support and enable CI on HPC systems. [#736](#736), [#751](#751), [#781](#781) + Add ssh debugging for Github Actions CI. [#749](#749) + Add pipeline segmentation for better CI speed. [#737](#737) Changes: + Add a Scalar Jacobi specialization and kernels. [#808](#808), [#834](#834), [#854](#854) + Add implicit residual log for solvers and benchmarks. [#714](#714) + Change handling of the conjugate in the dense dot product. [#755](#755) + Improved Dense stride handling. [#774](#774) + Multiple improvements to the OpenMP kernels performance, including COO, an exclusive prefix sum, and more. [#703](#703), [#765](#765), [#740](#740) + Allow specialization of submatrix and other dense creation functions in solvers. [#718](#718) + Improved Identity constructor and treatment of rectangular matrices. [#646](#646) + Allow CUDA/HIP executors to select allocation mode. [#758](#758) + Check if executors share the same memory. [#670](#670) + Improve test install and smoke testing support. [#721](#721) + Update the JOSS paper citation and add publications in the documentation. [#629](#629), [#724](#724) + Improve the version output. [#806](#806) + Add some utilities for dim and span. [#821](#821) + Improved solver and preconditioner benchmarks. [#660](#660) + Improve benchmark timing and output. [#669](#669), [#791](#791), [#801](#801), [#812](#812) Fixes: + Sorting fix for the Jacobi preconditioner. [#659](#659) + Also log the first residual norm in CGS [#735](#735) + Fix BiCG and HIP CSR to work with complex matrices. [#651](#651) + Fix Coo SpMV on strided vectors. [#807](#807) + Fix segfault of extract_diagonal, add short-and-fat test. [#769](#769) + Fix device_reset issue by moving counter/mutex to device. [#810](#810) + Fix `EnableLogging` superclass. [#841](#841) + Support ROCm 4.1.x and breaking HIP_PLATFORM changes. [#726](#726) + Decreased test size for a few device tests. [#742](#742) + Fix multiple issues with our CMake HIP and RPATH setup. [#712](#712), [#745](#745), [#709](#709) + Cleanup our CMake installation step. [#713](#713) + Various simplification and fixes to the Windows CMake setup. [#720](#720), [#785](#785) + Simplify third-party integration. [#786](#786) + Improve Ginkgo device arch flags management. [#696](#696) + Other fixes and improvements to the CMake setup. [#685](#685), [#792](#792), [#705](#705), [#836](#836) + Clarification of dense norm documentation [#784](#784) + Various development tools fixes and improvements [#738](#738), [#830](#830), [#840](#840) + Make multiple operators/constructors explicit. [#650](#650), [#761](#761) + Fix some issues, memory leaks and warnings found by MSVC. [#666](#666), [#731](#731) + Improved solver memory estimates and consistent iteration counts [#691](#691) + Various logger improvements and fixes [#728](#728), [#743](#743), [#754](#754) + Fix for ForwardIterator requirements in iterator_factory. [#665](#665) + Various benchmark fixes. [#647](#647), [#673](#673), [#722](#722) + Various CI fixes and improvements. [#642](#642), [#641](#641), [#795](#795), [#783](#783), [#793](#793), [#852](#852) Related PR: #866

Slaedr requested review from upsj, pratikvn, thoasm, yhmtsai, tcojean and fritzgoebel November 24, 2020 15:04

tcojean assigned Slaedr Nov 25, 2020

tcojean requested changes Nov 25, 2020

View reviewed changes

tcojean reviewed Nov 27, 2020

View reviewed changes

hip/matrix/fbcsr_kernels.hip.cpp Outdated Show resolved Hide resolved

core/test/matrix/fbcsr_builder.cpp Outdated Show resolved Hide resolved

Slaedr force-pushed the fbcsr_ref branch from e3fe272 to 572a6de Compare November 27, 2020 08:59

Slaedr commented Nov 27, 2020

View reviewed changes

core/test/matrix/fbcsr.cpp Outdated Show resolved Hide resolved

Slaedr commented Nov 27, 2020

View reviewed changes

reference/test/matrix/fbcsr_kernels.cpp Outdated Show resolved Hide resolved

Slaedr commented Nov 27, 2020

View reviewed changes

reference/test/matrix/fbcsr_kernels.cpp Outdated Show resolved Hide resolved

github-actions bot added mod:all This touches all Ginkgo modules. reg:build This is related to the build system. reg:testing This is related to testing. labels Nov 30, 2020

upsj removed mod:core This is related to the core module. mod:reference This is related to the reference module. labels Nov 30, 2020

pratikvn removed the mod:all This touches all Ginkgo modules. label Nov 30, 2020

Slaedr and others added 20 commits March 29, 2021 00:50

removed extra getter overloads

d67d61a

incorporated some suggestions from SonarCloud

a06bbae

added test for fbcsr calculate max nnz per row

a13ed33

fixed doc-string as suggested in review

8a0216e

added reference kernel for sorting fbcsr matrix by column

818b44c

Addressed code structures issues raised in review

4336552

- blockutils.hpp is removed for now, its contents moved to pre-existing files - added braces after control-flow statements - More public functions in class Fbcsr are marked noexecpt

Improved test coverage

323083c

Added tests for BlockSizeError, getNumBlocks and especially FbcsrSample classes.

Addressed points raised in review

688311c

Co-authored-by: Terry Cojean <[email protected]>

small dense blocks are now column-major

aebb433

New exception helper for divisibility by block size

a501787

Also in fbcsr.hpp, detail::get_num_blocks is now declared inline.

fixed get_num_blocks test

8ced1e9

Fbcsr write now writes in row-major order

6bff7cf

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

changed macros to variables in fbcsr sample

e370d36

review updates

8dc3d85

- removal of unnecessary lines of code - snake case convention for non-type template param Co-authored-by: Pratik Nayak <[email protected]>

added block size check to in fbcsr read, updated contributors list

a5d7929

fixed a bug in ref fbcsr transpose

5b57d56

replaced new DenseBlocksView with range and col_major accessor

1e49c91

rebased onto renamed block_col_major accessor

7fb1068

addressed changes to accessor sub-library

7669264

Fixed C-style casts

b2ea21f

Co-authored-by: Thomas Grützmacher <[email protected]>

Slaedr force-pushed the fbcsr_ref branch from f9a4ae8 to b2ea21f Compare March 28, 2021 22:56

pratikvn approved these changes Mar 29, 2021

View reviewed changes

Slaedr mentioned this pull request Mar 29, 2021

Repeated allocations while reading Fbcsr matrix #729

Closed

Slaedr merged commit 3ab51db into develop Mar 29, 2021

Slaedr deleted the fbcsr_ref branch March 29, 2021 12:08

Slaedr added this to Done in Fixed point-block preconditioners May 14, 2021

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Public interface and basic reference kernels for fixed-size block CSR #671

Public interface and basic reference kernels for fixed-size block CSR #671

Slaedr commented Nov 24, 2020 •

edited

Loading

tcojean left a comment

tcojean left a comment

tcojean commented Nov 30, 2020

tcojean commented Nov 30, 2020

github-actions bot commented Nov 30, 2020

pratikvn commented Nov 30, 2020

upsj commented Nov 30, 2020

pratikvn commented Nov 30, 2020

pratikvn left a comment

pratikvn Mar 29, 2021

pratikvn Mar 29, 2021

pratikvn Mar 29, 2021

sonarcloud bot commented Mar 29, 2021

Slaedr commented Mar 29, 2021 •

edited

Loading

Public interface and basic reference kernels for fixed-size block CSR #671

Public interface and basic reference kernels for fixed-size block CSR #671

Conversation

Slaedr commented Nov 24, 2020 • edited Loading

tcojean left a comment

Choose a reason for hiding this comment

tcojean left a comment

Choose a reason for hiding this comment

tcojean commented Nov 30, 2020

tcojean commented Nov 30, 2020

github-actions bot commented Nov 30, 2020

pratikvn commented Nov 30, 2020

upsj commented Nov 30, 2020

pratikvn commented Nov 30, 2020

pratikvn left a comment

Choose a reason for hiding this comment

pratikvn Mar 29, 2021

Choose a reason for hiding this comment

pratikvn Mar 29, 2021

Choose a reason for hiding this comment

pratikvn Mar 29, 2021

Choose a reason for hiding this comment

sonarcloud bot commented Mar 29, 2021

Slaedr commented Mar 29, 2021 • edited Loading

Slaedr commented Nov 24, 2020 •

edited

Loading

Slaedr commented Mar 29, 2021 •

edited

Loading