Add relative and absolute residual norm stopping criteria #577

fritzgoebel · 2020-06-24T16:59:13Z

This PR adds stopping criteria for relative (to norm(b)) and absolute residual norm.
It abstracts the three criteria ResidualNormReduction, RelativeResidualNorm and AbsoluteResidualNorm, which only differ in the way they initialize starting_tau_ from ResidualNorm.

If someone has a better name for the base class, I am open to suggestions, maybe something line ResidualNormCriterion?

Closes #147

upsj

LGTM! Do you actually test with different values for the different tests? It didn't look to me like that was the case when skimming over the CUDA tests.

yhmtsai

First part of my review. ( I do not check the test yet)
I think we need to keep the include/ginkgo/core/stop/residual_norm_reduction.hpp to keep the header interface. include/ginkgo/core/stop/residual_norm_reduction.hpp containing include/ginkgo/core/stop/residual_norm.hpp should be enough here.

yhmtsai · 2020-06-24T18:37:37Z

core/stop/residual_norm.cpp

- stoppingId, setFinalized, stop_status, &this->device_storage_,
- &all_converged, one_changed));
+ if (dense_tau->get_size()[1] > 1) {
+ }


Is anything missing here?

include/ginkgo/core/stop/residual_norm.hpp

pratikvn · 2020-06-24T20:00:43Z

I would still keep the abstract class name as ResidualNormReduction as we are still reducing the residual norm (but different types of the residual norm). The derived classes can be AbsoluteResNormReduction and RelativeResNormReduction or AbsResNormReduction and RelResNormReduction if the first two names are too long.

upsj · 2020-06-24T20:08:32Z

@pratik I would disagree there, since to me the Reduction in the name refers to the amount we reduce the residual norm by, while e.g. for Absolute/RelativeResidualNorm we give an exact tolerance, not a reduction factor. Basically to me, the different derived classes are different types of stopping criteria that are based on the residual norm in general, not necessarily its reduction.

pratikvn

Some minor things.

cuda/test/stop/residual_norm_kernels.cpp

include/ginkgo/core/stop/residual_norm.hpp

omp/test/stop/residual_norm_kernels.cpp

codecov · 2020-06-25T16:08:21Z

Codecov Report

Merging #577 into develop will increase coverage by 0.11%.
The diff coverage is 95.50%.

@@             Coverage Diff             @@
##           develop     #577      +/-   ##
===========================================
+ Coverage    84.10%   84.22%   +0.11%     
===========================================
  Files          296      296              
  Lines        20471    20649     +178     
===========================================
+ Hits         17218    17392     +174     
- Misses        3253     3257       +4

Impacted Files	Coverage Δ
core/device_hooks/common_kernels.inc.cpp	`0.00% <0.00%> (ø)`
core/test/solver/bicg.cpp	`100.00% <ø> (ø)`
core/test/solver/bicgstab.cpp	`100.00% <ø> (ø)`
core/test/solver/cg.cpp	`100.00% <ø> (ø)`
core/test/solver/cgs.cpp	`100.00% <ø> (ø)`
core/test/solver/fcg.cpp	`100.00% <ø> (ø)`
core/test/solver/gmres.cpp	`99.15% <ø> (ø)`
core/test/solver/ir.cpp	`98.37% <ø> (ø)`
include/ginkgo/core/preconditioner/ilu.hpp	`44.89% <ø> (ø)`
omp/test/solver/bicg_kernels.cpp	`100.00% <ø> (ø)`
... and 27 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 c6a469e...9684088. Read the comment docs.

fritzgoebel · 2020-06-25T16:14:50Z

LGTM! Do you actually test with different values for the different tests? It didn't look to me like that was the case when skimming over the CUDA tests.

You mean different values for the tolerance? No, I use the same value as was used before for the ResidualNormReduction criterion..

yhmtsai

For the test, I would like to use the following way.
Each test uses same b and initial residual which are not equal to 1 and different.
(for example, b is 1e3 but initial residual 1e-3)
they will need different final residual to pass them such that we can make sure they have different behavior.

core/stop/residual_norm.cpp

cuda/test/stop/residual_norm_kernels.cpp

include/ginkgo/core/stop/residual_norm_reduction.hpp

yhmtsai

LGTM. Is any reason to use different cases in multiple rhs tests? Some uses 1 for rhs and initial residual, but others use 100 for initial residual and 10 for rhs.

cuda/test/stop/residual_norm_kernels.cpp

yhmtsai

LGTM! I like the usage of 0.9 and 1.1.

yhmtsai · 2020-06-29T17:35:45Z

include/ginkgo/core/stop/residual_norm_reduction.hpp

-#ifndef GKO_CORE_STOP_RESIDUAL_NORM_REDUCTION_HPP_
-#define GKO_CORE_STOP_RESIDUAL_NORM_REDUCTION_HPP_


tcojean

LGTM. A small comment on test_install.

tcojean · 2020-06-29T17:45:06Z

test_install/test_install.cpp

@@ -351,7 +351,7 @@ int main(int, char **)
 auto time = gko::stop::Time::build()
 .with_time_limit(std::chrono::milliseconds(10))
 .on(refExec);
- // residual_norm_reduction.hpp
+ // residual_norm.hpp
 auto res_red = gko::stop::ResidualNormReduction<>::build()


You need to add similar checks with the new RelativeResidualNorm and Absolute.

tcojean · 2020-06-29T17:45:16Z

test_install/test_install_cuda.cu

@@ -343,7 +343,7 @@ int main(int, char **)
 auto time = gko::stop::Time::build()
 .with_time_limit(std::chrono::milliseconds(10))
 .on(cudaExec);
- // residual_norm_reduction.hpp
+ // residual_norm.hpp


tcojean · 2020-06-30T14:04:00Z

Looks like the residual_norm_reduction.hpp file is still included in the main ginkgo.hpp although it should be removed. I think this file is generated by a script so the script needs to be updated to not include this file. The base file (which is overwritten in case .the script is ran) also needs to be updated.
https://gitlab.com/ginkgo-project/ginkgo-public-ci/-/jobs/616917734

pratikvn

LGTM!

include/ginkgo/core/stop/residual_norm.hpp

sonarcloud · 2020-07-01T12:59:46Z

Kudos, SonarCloud Quality Gate passed!

0 Bugs
0 Vulnerabilities (and 0 Security Hotspots to review)
26 Code Smells

92.4% Coverage
3.0% Duplication

The version of Java (1.8.0_121) you have used to run this analysis is deprecated and we will stop accepting it from October 2020. Please update to at least Java 11.
Read more here

The Ginkgo team is proud to announce the new minor release of Ginkgo version 1.2.0. This release brings full HIP support to Ginkgo, new preconditioners (ParILUT, ISAI), conversion between double and float for all LinOps, and many more features and fixes. Supported systems and requirements: + For all platforms, cmake 3.9+ + Linux and MacOS + gcc: 5.3+, 6.3+, 7.3+, all versions after 8.1+ + clang: 3.9+ + Intel compiler: 2017+ + Apple LLVM: 8.0+ + CUDA module: CUDA 9.0+ + HIP module: ROCm 2.8+ + Windows + MinGW and CygWin: gcc 5.3+, 6.3+, 7.3+, all versions after 8.1+ + Microsoft Visual Studio: VS 2017 15.7+ + CUDA module: CUDA 9.0+, Microsoft Visual Studio + OpenMP module: MinGW or CygWin. The current known issues can be found in the [known issues page](https://github.com/ginkgo-project/ginkgo/wiki/Known-Issues). # Additions Here are the main additions to the Ginkgo library. Other thematic additions are listed below. + Add full HIP support to Ginkgo [#344](#344), [#357](#357), [#384](#384), [#373](#373), [#391](#391), [#396](#396), [#395](#395), [#393](#393), [#404](#404), [#439](#439), [#443](#443), [#567](#567) + Add a new ISAI preconditioner [#489](#489), [#502](#502), [#512](#512), [#508](#508), [#520](#520) + Add support for ParILUT and ParICT factorization with ILU preconditioners [#400](#400) + Add a new BiCG solver [#438](#438) + Add a new permutation matrix format [#352](#352), [#469](#469) + Add CSR SpGEMM support [#386](#386), [#398](#398), [#418](#418), [#457](#457) + Add CSR SpGEAM support [#556](#556) + Make all solvers and preconditioners transposable [#535](#535) + Add CsrBuilder and CooBuilder for intrusive access to matrix arrays [#437](#437) + Add a standard-compliant allocator based on the Executors [#504](#504) + Support conversions for all LinOp between double and float [#521](#521) + Add a new boolean to the CUDA and HIP executors to control DeviceReset (default off) [#557](#557) + Add a relaxation factor to IR to represent Richardson Relaxation [#574](#574) + Add two new stopping criteria, for relative (to `norm(b)`) and absolute residual norm [#577](#577) ### Example additions + Templatize all examples to simplify changing the precision [#513](#513) + Add a new adaptive precision block-Jacobi example [#507](#507) + Add a new IR example [#522](#522) + Add a new Mixed Precision Iterative Refinement example [#525](#525) + Add a new example on iterative trisolves in ILU preconditioning [#526](#526), [#536](#536), [#550](#550) ### Compilation and library changes + Auto-detect compilation settings based on environment [#435](#435), [#537](#537) + Add SONAME to shared libraries [#524](#524) + Add clang-cuda support [#543](#543) ### Other additions + Add sorting, searching and merging kernels for GPUs [#403](#403), [#428](#428), [#417](#417), [#455](#455) + Add `gko::as` support for smart pointers [#493](#493) + Add setters and getters for criterion factories [#527](#527) + Add a new method to check whether a solver uses `x` as an initial guess [#531](#531) + Add contribution guidelines [#549](#549) # Fixes ### Algorithms + Improve the classical CSR strategy's performance [#401](#401) + Improve the CSR automatical strategy [#407](#407), [#559](#559) + Memory, speed improvements to the ELL kernel [#411](#411) + Multiple improvements and fixes to ParILU [#419](#419), [#427](#427), [#429](#429), [#456](#456), [#544](#544) + Fix multiple issues with GMRES [#481](#481), [#523](#523), [#575](#575) + Optimize OpenMP matrix conversions [#505](#505) + Ensure the linearity of the ILU preconditioner [#506](#506) + Fix IR's use of the advanced apply [#522](#522) + Fix empty matrices conversions and add tests [#560](#560) ### Other core functionalities + Fix complex number support in our math header [#410](#410) + Fix CUDA compatibility of the main ginkgo header [#450](#450) + Fix isfinite issues [#465](#465) + Fix the Array::view memory leak and the array/view copy/move [#485](#485) + Fix typos preventing use of some interface functions [#496](#496) + Fix the `gko::dim` to abide to the C++ standard [#498](#498) + Simplify the executor copy interface [#516](#516) + Optimize intermediate storage for Composition [#540](#540) + Provide an initial guess for relevant Compositions [#561](#561) + Better management of nullptr as criterion [#562](#562) + Fix the norm calculations for complex support [#564](#564) ### CUDA and HIP specific + Use the return value of the atomic operations in our wrappers [#405](#405) + Improve the portability of warp lane masks [#422](#422) + Extract thread ID computation into a separate function [#464](#464) + Reorder kernel parameters for consistency [#474](#474) + Fix the use of `pragma unroll` in HIP [#492](#492) ### Other + Fix the Ginkgo CMake installation files [#414](#414), [#553](#553) + Fix the Windows compilation [#415](#415) + Always use demangled types in error messages [#434](#434), [#486](#486) + Add CUDA header dependency to appropriate tests [#452](#452) + Fix several sonarqube or compilation warnings [#453](#453), [#463](#463), [#532](#532), [#569](#569) + Add shuffle tests [#460](#460) + Fix MSVC C2398 error [#490](#490) + Fix missing interface tests in test install [#558](#558) # Tools and ecosystem ### Benchmarks + Add better norm support in the benchmarks [#377](#377) + Add CUDA 10.1 generic SpMV support in benchmarks [#468](#468), [#473](#473) + Add sparse library ILU in benchmarks [#487](#487) + Add overhead benchmarking capacities [#501](#501) + Allow benchmarking from a matrix list file [#503](#503) + Fix benchmarking issue with JSON and non-finite numbers [#514](#514) + Fix benchmark logger crashers with OpenMP [#565](#565) ### CI related + Improvements to the CI setup with HIP compilation [#421](#421), [#466](#466) + Add MacOSX CI support [#470](#470), [#488](#488) + Add Windows CI support [#471](#471), [#488](#488), [#510](#510), [#566](#566) + Use sanitizers instead of valgrind [#476](#476) + Add automatic container generation and update facilities [#499](#499) + Fix the CI parallelism settings [#517](#517), [#538](#538), [#539](#539) + Make the codecov patch check informational [#519](#519) + Add support for LLVM sanitizers with improved thread sanitizer support [#578](#578) ### Test suite + Add an assertion for sparsity pattern equality [#416](#416) + Add core and reference multiprecision tests support [#448](#448) + Speed up GPU tests by avoiding device reset [#467](#467) + Change test matrix location string [#494](#494) ### Other + Add Ginkgo badges from our tools [#413](#413) + Update the `create_new_algorithm.sh` script [#420](#420) + Bump copyright and improve license management [#436](#436), [#433](#433) + Set clang-format minimum requirement [#441](#441), [#484](#484) + Update git-cmake-format [#446](#446), [#484](#484) + Disable the development tools by default [#442](#442) + Add a script for automatic header formatting [#447](#447) + Add GDB pretty printer for `gko::Array` [#509](#509) + Improve compilation speed [#533](#533) + Add editorconfig support [#546](#546) + Add a compile-time check for header self-sufficiency [#552](#552) # Related PR: #583

fritzgoebel added is:enhancement An improvement of an existing feature. mod:core This is related to the core module. labels Jun 24, 2020

fritzgoebel requested review from upsj, pratikvn, thoasm, yhmtsai, hartwiganzt and tcojean June 24, 2020 16:59

fritzgoebel self-assigned this Jun 24, 2020

upsj approved these changes Jun 24, 2020

View reviewed changes

yhmtsai requested changes Jun 24, 2020

View reviewed changes

pratikvn requested changes Jun 25, 2020

View reviewed changes

pratikvn added 1:ST:ready-for-review This PR is ready for review type:stopping-criteria This is related to the stopping criteria labels Jun 25, 2020

yhmtsai requested changes Jun 25, 2020

View reviewed changes

yhmtsai requested changes Jun 29, 2020

View reviewed changes

cuda/test/stop/residual_norm_kernels.cpp Show resolved Hide resolved

cuda/test/stop/residual_norm_kernels.cpp Outdated Show resolved Hide resolved

fritzgoebel added 7 commits June 29, 2020 17:37

Add relative and absolute residual norm stopping criteria

b0974c2

review updates

4a7b539

Add include/ginkgo/core/stop/residual_norm_reduction.hpp

00cc07a

Fix cuda and hip tests

c181e36

Fix cuda and hip tests

e242593

review updates

16d74d9

Fix tests

99e3e1e

fritzgoebel force-pushed the add_more_stopping_criteria branch from 08d98db to 99e3e1e Compare June 29, 2020 15:38

tcojean added this to the Ginkgo 1.2.0 milestone Jun 29, 2020

yhmtsai approved these changes Jun 29, 2020

View reviewed changes

yhmtsai reviewed Jun 29, 2020

View reviewed changes

tcojean approved these changes Jun 29, 2020

View reviewed changes

Add test_install checks for the new criteria

48bbffb

fritzgoebel force-pushed the add_more_stopping_criteria branch from 925b264 to 48bbffb Compare June 30, 2020 07:11

Remove .../residual_norm_reduction.hpp from ginkgo/ginkgo.hpp

bd2345f

pratikvn approved these changes Jul 1, 2020

View reviewed changes

include/ginkgo/core/stop/residual_norm.hpp Show resolved Hide resolved

include/ginkgo/core/stop/residual_norm.hpp Show resolved Hide resolved

include/ginkgo/core/stop/residual_norm.hpp Outdated Show resolved Hide resolved

Add equations to explanation of the stopping criteria

9684088

fritzgoebel merged commit e089243 into develop Jul 1, 2020

fritzgoebel deleted the add_more_stopping_criteria branch July 1, 2020 13:08

tcojean mentioned this pull request Jul 1, 2020

Release/1.2.0 #576

Merged

tcojean 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 Jul 7, 2020

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add relative and absolute residual norm stopping criteria #577

Add relative and absolute residual norm stopping criteria #577

fritzgoebel commented Jun 24, 2020 •

edited

Loading

upsj left a comment

yhmtsai left a comment

yhmtsai Jun 24, 2020

pratikvn commented Jun 24, 2020

upsj commented Jun 24, 2020 •

edited

Loading

pratikvn left a comment

codecov bot commented Jun 25, 2020 •

edited

Loading

fritzgoebel commented Jun 25, 2020

yhmtsai left a comment

yhmtsai left a comment

yhmtsai left a comment

yhmtsai Jun 29, 2020

tcojean left a comment

tcojean Jun 29, 2020

tcojean Jun 29, 2020

tcojean commented Jun 30, 2020 •

edited

Loading

pratikvn left a comment

sonarcloud bot commented Jul 1, 2020

		#ifndef GKO_CORE_STOP_RESIDUAL_NORM_REDUCTION_HPP_
		#define GKO_CORE_STOP_RESIDUAL_NORM_REDUCTION_HPP_

Add relative and absolute residual norm stopping criteria #577

Add relative and absolute residual norm stopping criteria #577

Conversation

fritzgoebel commented Jun 24, 2020 • edited Loading

upsj left a comment

Choose a reason for hiding this comment

yhmtsai left a comment

Choose a reason for hiding this comment

yhmtsai Jun 24, 2020

Choose a reason for hiding this comment

pratikvn commented Jun 24, 2020

upsj commented Jun 24, 2020 • edited Loading

pratikvn left a comment

Choose a reason for hiding this comment

codecov bot commented Jun 25, 2020 • edited Loading

Codecov Report

fritzgoebel commented Jun 25, 2020

yhmtsai left a comment

Choose a reason for hiding this comment

yhmtsai left a comment

Choose a reason for hiding this comment

yhmtsai left a comment

Choose a reason for hiding this comment

yhmtsai Jun 29, 2020

Choose a reason for hiding this comment

tcojean left a comment

Choose a reason for hiding this comment

tcojean Jun 29, 2020

Choose a reason for hiding this comment

tcojean Jun 29, 2020

Choose a reason for hiding this comment

tcojean commented Jun 30, 2020 • edited Loading

pratikvn left a comment

Choose a reason for hiding this comment

sonarcloud bot commented Jul 1, 2020

fritzgoebel commented Jun 24, 2020 •

edited

Loading

upsj commented Jun 24, 2020 •

edited

Loading

codecov bot commented Jun 25, 2020 •

edited

Loading

tcojean commented Jun 30, 2020 •

edited

Loading