Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Magma and cuBLAS incompatible in unit tests #2177

Open
ndellingwood opened this issue Apr 12, 2024 · 3 comments
Open

Magma and cuBLAS incompatible in unit tests #2177

ndellingwood opened this issue Apr 12, 2024 · 3 comments
Labels

Comments

@ndellingwood
Copy link
Contributor

Cuda builds with Magma and CuBlas are currently incompatible, resulting in the following compilation error:

17:08:00 /home/jenkins/weaver/workspace/KokkosKernels_Weaver_Cuda_Serial_cuda_1180_gcc_850_cusparse_cublas_magma/kokkos-kernels/tpls/gtest/gtest/gtest.h(18865): error: operand types are incompatible ("magma_queue_t *" and "cublasHandle_t *")
17:08:00           detected during:
17:08:00             instantiation of "testing::AssertionResult testing::internal::CmpHelperEQ(const char *, const char *, const T1 &, const T2 &) [with T1=magma_queue_t *, T2=cublasHandle_t *]" 
17:08:00 (18902): here
17:08:00             instantiation of "testing::AssertionResult testing::internal::EqHelper<lhs_is_null_literal>::Compare(const char *, const char *, const T1 &, const T2 &) [with lhs_is_null_literal=false, T1=magma_queue_t *, T2=cublasHandle_t *]" 
17:08:00 /home/jenkins/weaver/workspace/KokkosKernels_Weaver_Cuda_Serial_cuda_1180_gcc_850_cusparse_cublas_magma/kokkos-kernels/batched/dense/unit_test/Test_Batched_BatchedGemm.hpp(236): here
17:08:00             instantiation of "void Test::impl_test_batched_gemm<DeviceType,ViewType,ScalarType,ParamTagType>(int, int, int, int, int, int, int) [with DeviceType=Kokkos::Serial, ViewType=Kokkos::View<double ***, Kokkos::LayoutLeft, Kokkos::Serial>, ScalarType=double, ParamTagType=Test::SharedParamTag<KokkosBlas::Trans::NoTranspose, KokkosBlas::Trans::NoTranspose, KokkosBatched::BatchLayout::Right>]" 
17:08:00 /home/jenkins/weaver/workspace/KokkosKernels_Weaver_Cuda_Serial_cuda_1180_gcc_850_cusparse_cublas_magma/kokkos-kernels/batched/dense/unit_test/Test_Batched_BatchedGemm.hpp(320): here
17:08:00             instantiation of "void test_batched_gemm_with_layout<ViewType,DeviceType,ValueType,ScalarType,ParamTagType>(int) [with ViewType=Kokkos::View<double ***, Kokkos::LayoutLeft, Kokkos::Serial>, DeviceType=Kokkos::Serial, ValueType=double, ScalarType=double, ParamTagType=Test::SharedParamTag<KokkosBlas::Trans::NoTranspose, KokkosBlas::Trans::NoTranspose, KokkosBatched::BatchLayout::Right>]" 
17:08:00 /home/jenkins/weaver/workspace/KokkosKernels_Weaver_Cuda_Serial_cuda_1180_gcc_850_cusparse_cublas_magma/kokkos-kernels/batched/dense/unit_test/Test_Batched_BatchedGemm.hpp(388): here
17:08:00             instantiation of "int test_batched_gemm<DeviceType,ValueType,ScalarType,ParamTagType>() [with DeviceType=Kokkos::Serial, ValueType=double, ScalarType=double, ParamTagType=Test::SharedParamTag<KokkosBlas::Trans::NoTranspose, KokkosBlas::Trans::NoTranspose, KokkosBatched::BatchLayout::Right>]" 
17:08:00 /home/jenkins/weaver/workspace/KokkosKernels_Weaver_Cuda_Serial_cuda_1180_gcc_850_cusparse_cublas_magma/kokkos-kernels/batched/dense/unit_test/Test_Batched_BatchedGemm_Real.hpp(260): here
17:08:00 
17:08:01 1 error detected in the compilation of "/home/jenkins/weaver/workspace

Reproducer (weaver):

module load gcc/8.5.0 cuda/11.8.0 cmake magma/2.7.0 openblas/0.3.23

${KOKKOSKERNELS_PATH}/cm_generate_makefile.bash --with-cuda --with-serial --compiler=${KOKKOS_PATH}/bin/nvcc_wrapper --arch=Volta70,Power9 --with-cuda-options=enable_lambda --kokkos-path=${KOKKOS_PATH} --kokkoskernels-path=${KOKKOSKERNELS_PATH} ${DEBUG} --cxxflags=\"${CXXFLAGS}\" --with-scalars='double,complex_double' --with-ordinals=int --with-offsets=int,size_t --with-layouts=LayoutLeft --with-tpls=cusparse,cublas,magma
@ndellingwood ndellingwood changed the title Magma and CuBlas incompatible Magma and cuBLAS incompatible Apr 12, 2024
@ndellingwood ndellingwood changed the title Magma and cuBLAS incompatible Magma and cuBLAS incompatible in unit tests Apr 13, 2024
@ndellingwood
Copy link
Contributor Author

ndellingwood commented Apr 13, 2024

This is the culprit region in the BatchedGemmHandle for the unit tests:

BatchedGemmHandle(int kernelAlgoType = BaseHeuristicAlgos::SQUARE,
int teamSize = 0, int vecLength = 0)
: BatchedKernelHandle(kernelAlgoType, teamSize, vecLength) {
#if defined(KOKKOSKERNELS_ENABLE_TPL_CUBLAS)
if (!_tplParamsSet && kernelAlgoType == GemmTplAlgos::CUBLAS) {
static cublasHandle_t cublas_handle;
_tplParamsSingleton.cublas_handle = &cublas_handle;
_tplParamsSet = true;
}
#endif // CUBLAS
#if defined(KOKKOSKERNELS_ENABLE_TPL_MAGMA)
if (!_tplParamsSet && kernelAlgoType == GemmTplAlgos::MAGMA) {
static magma_queue_t magma_queue;
_tplParamsSingleton.magma_queue = &magma_queue;
_tplParamsSet = true;
}
#endif // MAGMA
};
BatchedGemmHandle(bool tplParamsSet,
int kernelAlgoType = BaseHeuristicAlgos::SQUARE,
int teamSize = 0, int vecLength = 0)
: BatchedKernelHandle(kernelAlgoType, teamSize, vecLength) {
_tplParamsSet = tplParamsSet;
};
#if defined(KOKKOSKERNELS_ENABLE_TPL_CUBLAS)
BatchedGemmHandle(cublasHandle_t &cublas_handle,
int kernelAlgoType = BaseHeuristicAlgos::SQUARE,
int teamSize = 0, int vecLength = 0)
: BatchedGemmHandle(true, kernelAlgoType, teamSize, vecLength) {
_tplParamsSingleton.cublas_handle = &cublas_handle;
};
#endif // CUBLAS
#if defined(KOKKOSKERNELS_ENABLE_TPL_MAGMA)
BatchedGemmHandle(magma_queue_t &magma_queue,
int kernelAlgoType = BaseHeuristicAlgos::SQUARE,
int teamSize = 0, int vecLength = 0)
: BatchedGemmHandle(true, kernelAlgoType, teamSize, vecLength) {
_tplParamsSingleton.magma_queue = &magma_queue;
};
#endif // MAGMA
decltype(auto) get_tpl_params() {
#if _kernelAlgoType == CUBLAS && defined(KOKKOSKERNELS_ENABLE_TPL_CUBLAS)
return _tplParamsSingleton.cublas_handle;
#elif _kernelAlgoType == MAGMA && defined(KOKKOSKERNELS_ENABLE_TPL_MAGMA)
return _tplParamsSingleton.magma_queue;
#else
return this->BatchedKernelHandle::get_tpl_params();
#endif
}

@lucbv
Copy link
Contributor

lucbv commented Apr 15, 2024

Thanks for catching this, we should go clean it up.
Setting up the build with MAGMA is going to be very helpful!

ndellingwood added a commit to ndellingwood/kokkos-kernels that referenced this issue Apr 23, 2024
- temporary workaround to skip magma test when cublas enabled to avoid issues like kokkos#2177
lucbv pushed a commit that referenced this issue Apr 23, 2024
* magma: fix linker errors for builds without cusolver

* BatchedGemm test: workaround testing cublas+magma

- temporary workaround to skip magma test when cublas enabled to avoid issues like #2177
@ndellingwood
Copy link
Contributor Author

ndellingwood commented Apr 25, 2024

Adding a note that #2181 has a temporary workaround in the unit test to allow compiling tests with both magma and cublas enabled. I think the BatchedGemmHandle will need some refactoring in order to remove the workaround

ndellingwood added a commit that referenced this issue May 1, 2024
* magma: fix linker errors for builds without cusolver

* BatchedGemm test: workaround testing cublas+magma

- temporary workaround to skip magma test when cublas enabled to avoid issues like #2177
lucbv pushed a commit to lucbv/kokkos-kernels that referenced this issue May 8, 2024
* magma: fix linker errors for builds without cusolver

* BatchedGemm test: workaround testing cublas+magma

- temporary workaround to skip magma test when cublas enabled to avoid issues like kokkos#2177
lucbv added a commit that referenced this issue Aug 12, 2024
* fence Kokkos before timed interations

* Deprecate KOKKOSLINALG_OPT_LEVEL

* Add CMake warning message if KokkosKernels_LINALG_OPT_LEVEL is used

* Async matrix release for MKL >= 2023.2

* Support CUBLAS_{LIBRARIES,LIBRARY_DIRS,INCLUDE_DIRS,ROOT} and KokkosKernels_CUBLAS_ROOT

* KokkosSparse_spmv_impl_merge.hpp: use capture by reference

Resolve warnings in builds with c++20 support enabled:
"kokkos-kernels/sparse/impl/KokkosSparse_spmv_impl_merge.hpp:166:81: warning: implicit capture of 'this' via '[=]' is deprecated in C++20 [-Wdeprecated]"

* KokkosSparse_par_ilut_numeric_impl.hpp: use capture by reference

Resolve warnings in builds with c++20 support enabled:
"kokkos-kernels/sparse/impl/KokkosSparse_par_ilut_numeric_impl.hpp(591):
warning #2908-D: the implicit by-copy capture of "this" is deprecated"

* Backup

* Backup

* Backup

* Backup

* Formatting

* Correcting compilation error

* Typo

* Changes for syr and syr2, to be tested at weaver

* Formatting

* Changes for axpby

* Backup

* Formatting

* Just to force new checking tests in github

* Addressing feedback from Luc.

* Don't call optimize_gemv for one-shot spmv

* Add HIPManagedSpace support

- CMake option for ETI
- Run unit tests with a Kokkos::Device, not just Kokkos::HIP
	- Like we do for Cuda
	- Still use HIPSpace unless Managed is the only enabled memspace
- Couple of minor fixes
	- Allow querying free HIPManagedSpace memory for SpGEMM
	- Disable VBD coloring (not a huge deal, had to do same on CUDA)
	- Use correct memory space in SpTRSV solve

* Backup

* Backup

* Backup

* Minor typo

* Add block support to all SPILUK algorithms (#2064)

* Interface for block iluk

* Progress. Test hooked up

* Progress on test refactoring

* More test reorg

* Fix test

* Refactor spiluk numeric a bit with a struct wrapper

* Add good logging

* progress

* Fix block test

* Progress but potential dead end

* Giving up on this approach for now

* progress

* Make verbose

* Progress

* Progress

* RP working?

* Progress on TP alg

* Bug fix

* Progress on template stuff

* Progress on block TP

* Progress

* Get rid of all the static_casts

* More cleanup. Steams now support blocks

* Tests not passing

* Serail tests all working, both algs, blocked

* Remove output coming from spiluk test

* Final fixes for CPU

* Cuda req full template specification for SerialGemm::invoke

* Don't use scratch for now

* Formatting

* Fix warnings

* Formatting

* Add tolerance to view checks. Use macro and remove redundant test util

* Fix for HIP

* formatting

* Another test reorg to fix weirdness on solo

* formatting

* Remove unused var

* Github feedback

* Remove test cout

* formatting

* Zero-size arrays can cause problems

* Fix unused var warning

* Add CUDA/HIP TPL support for KokkosSparse::spadd (#1962)

* spadd: change arguments to ctor of SPADDHandle

add a default value to input_sorted;
add a second argument input_merged to indicate unqiue entries;
So that we can easily know whether we can use TPLs on the input matrices

* spadd: add cuda/rocm TPL support for spadd_symbolic/numeric

* Make spiluk_handle::reset backwards compatible (#2087)

* Make spiluk_handle::reset backwards compatible

By making block_size default to -1, which means don't change
block size.

* Switch default val for block_size for reset_handle

* formatting

* Fix comment

* spadd: add APIs without an execution space argument (#2090)

* Lapack - SVD: adding initial files that do not implement anything (#2092)

Adding SVD feature to Lapack component, the interface is similar
to classic Lapack and the implementation relies on the TPL layer
to provide initial capabilities. The TPL supported are LAPACK,
MKL, cuSOLVER and rocSOLVER.

Testing three analytical cases 2x2, 2x3 and 3x2 and then some
randomly generated matrices.

* Hands off namespace `Kokkos::Impl` - cleanup couple violations that snuck in (#2094)

* Do not use things from namespace Kokkos::Impl (Kokkos::{Impl:: -> }ALL_t)

* Do not use things from namespace Kokkos::Impl (Kokkos::Impl::DeepCopy)

Can achieve the same with Kokkos::deep_copy

* Fix warning `declaration of ‘std::size_t n’ shadows a parameter`

* Change name of yaml-cpp to yamlcpp

* Fix macro setting in CMakeLists

* GMRES: Add support for BSR matrices

Also, add a test for this.

* Remove all mentions of HBWSpace

* Reintroduce EXECSPACE_(SERIAL,OPENMP,THREADS}_VALID_MEM_SPACES

Drop HBWSPACE as an option

* Lapack: adding svd benchmark

Fixing unit-test for CUSOLVER and adding benchmark to check the
algorithm performance on various platforms.

* Fix Cuda TPL finding (#2098)

- Allow finding cusparse, cusolver based on manually provided paths
  - This is necessary when using an nvhpc toolchain instead of a
    standard cuda toolchain
- Set header paths correctly (this is redundant in a cuda installation,
  in which $CUDA_ROOT/include is already a system include dir, but
  needed in other cases)

* Add support for BSR matrices to some trsv routines (#2104)

* Add support for BSR matrices to some trsv routines
* Change trsv to gesv

* Lapack - SVD: adding quick return when cuSOLVER is skipped (#2107)

Currently we still run the tests on U, S and Vt which does not
make sense since we actively skip this test because cuSOLVER does
not support more columns than rows...

* Fix build error in trsv on gcc8

* Add a workaround for compilation errors with cuda-12.2.0 + gcc-12.3 (#2108)

On Perlmutter@NERSC, I met this error

/usr/lib64/gcc/x86_64-suse-linux/12/include/avx512fp16intrin.h(38): error: vector_size attribute requires an arithmetic or enum type
   typedef __half __v8hf __attribute__ ((__vector_size__ (16)));

The workaround was mentioned at https://forums.developer.nvidia.com/t/including-cub-header-breakes-compilation-with-gcc-12-and-sse2-or-better/255018

* Lapack - SVD: fix for unit-test when MKL is enabled (#2110)

This is really a problem with our implementation of the BLAS
interface when MKL is enabled since MKL redefines the function
signatures of blas functions using MKL_INT instead if int...

* Revert "Merge pull request #2037 from ndellingwood/remove-rocsolver-optional-dependency" (#2106)

This reverts commit 5a36d577e725546062af3b297eec87e23a40ab58, reversing
changes made to 2c66d291f9b5512e17f9375304902b6ba42133b2.

* Fixing missing inclusion in source file

* BLAS - MKL: fixing HostBlas calls to handle MKL_INT type (#2112)

MKL redefines the BLAS interface based on how MKL_INT is defined
we need to wrap that definition with our own Kokkos Kernels INT
type to make both compatible with regular BLAS.

applying clang-format

* Fix weird Trilinos compiler error

It seemed to have a problem with these deep_copies, so just do
the copy by hand like it was being done before my recent trsv
PR.

* Update changelog

* Update changelog

* Block spiluk follow up (#2085)

* Fix for gemm
* Remove unused divide method
* Enhancements to spiluk test
* Progress. Block spiluk now checks out against analytical results
* LUPrec test with spiluk woring
* Disable spiluk LU test on non-host
* Enhancements to spiluk test
* Clean up a few issues uncovered by gh review

* github workflows: update to v4 (use Node 20)

* Refactor Test_Sparse_sptrsv (#2102)

* Refactor Test_Sparse_sptrsv

* More cleanups

* Remove old commented-out code

* CMake: error out in certain case (#2115)

Graph unit tests are unique in that they use default_scalar for the
KokkosKernelsHandle. So if test-eti-only is ON, but neither float nor
double is instatiated, then error out for the graph unit tests.

Users can still build without float or double if they want, but only if
they turn off tests or the graph component.

* Wiki examples for BLAS2 functions are added (#2122)

Some small additional change the the function headers themselves
to add some missing header file inclusions.

Applying clang-format

Removing constexpr since it won't happen before some work in Core.

* Increase tolerance on gesv test (Fix #2123) (#2124)

And uncomment the verbose output for when tolerance is exceeded,
since that helps debug this sort of issue.
This is only printed at most once so it won't spam the output if
the entire vector is wrong.

* Spmv handle (#2126)

* spmv handle, TPL reuse

* using handle in unification layer and hooking up new algorithm
enums with old Controls options

* Update spmv_merge perf test
Compare KK merge vs. default and KK native

* Small changes to help text of spmv_merge perf test

* Complete backwards compatibility with Controls interface
- copy over spmv algorithm selection correctly
- copy expert tuning parameters

* Controls spmv: accept other name for bsr algo

* bsr spmv test: disable tensor core
It was not actually being run before due to a different name
actually enabling it (experimental_bsr_tc rather than experimental_tc)

* Disable OneMKL spmv for complex types
oneapi 2023.2 throws error saying complex isn't supported

* OneMKL: call optimize_gemv during setup

* Option to apply RCM reordering to extracted CRS diagonal blocks (#2125)

* Add rcm option when extracting diagonal blocks

* Update kk_extract_diagonal_blocks_crsmatrix_sequential

* Add test for extracting diagonal blocks with rcm

* Update RCM checking

* cm_test_all_sandia: various updates

- updates for blake

* cm_test_all_sandia: drop decommissioned/unavailable machines

- remove voltrino, mayer

* Fix2130 (#2132)

* Fix #2130

- Do not call BsrMatrix spmv impl if block size is 1
- Instead, convert it to unmanaged CrsMatrix and call spmv again
  - cuSPARSE returned an error code in this case
  - Better performance

* Formatting

* Remove redundant remove_pointer_t

Handle is already a non-pointer type

* Benchmark: modifying spmv benchmark to run range of spmv tests (#2135)

This could be further automated to run on matrix from suite sparse

* Kokkos Kernels: update version guards to drop old version of Kokkos (#2133)

Since we are now in the 4.2 series we only support up to 4.1.00.
Older version of Kokkos Core will require older version of Kokkos
Kernels for compatibility. Once 4.3.00 is out we will move to
drop support for the 4.1 series and only keep 4.2 and 4.3 series.

* ODE: BDF methods (#1930)

* ODE: adding BDF algorithms

Implementing BDF formula for stiff ODEs.
Orders 1 to 5 are available and tested.
The integrators can be called on GPU to
solve multiple systems in parallel.

* ODE: fixing storage handling for start-up RK stack

* ODE: clang-format

* ODE: first adaptive version of BDF

The current implementation only allows for adaptivity in time,
at this point the BDF Step actually converges as expected with
first order integration!

* ODE: fixing issues with adaptive BDF

The unit-test BDF_adaptive now shows the integration
of the logistic equation using adaptive time steps and
increasing integration order from 1 to 5.

* ODE: running BDF on StiffChemistry problem

The problem runs fine and is solved but there are oscillations
while the behavior of the solution is smooth. More investigation
is needed...

* BDF: fixing types and template parameters in batched calls

Bascially we need template parameters to be more versatile
and cannot assume that all rank1 views will have the exact
same underlying type, for instance layouts can be different.

* More fixes for GPUs only in tests this time.

* ODE: BDF adaptive, fix small bug

After adding rhs and update vectors to temp the subviews taken for
other variables need to be offset appropriately...

* Revert "More fixes for GPUs only in tests this time."

This reverts commit 2f70432761485bc6a4c65a1833e7299dd2c340e2.

* Revert "Revert "More fixes for GPUs only in tests this time.""

This reverts commit 836012bb529551727b3f5913057acad94dfe60df.

* ODE: BDF small change to temporarily avoid compile time issue

True fix involving a KOKKOS_VERSION check is upcoming after more
tests on GPU side...

* ODE: BDF fix for some printf statements that will go away soon...

* ODE: adding benchmark for BDF

The benchmark helps us monitor the performance of the BDF
implementaiton across multiple platforms as well as impact of
changes over time.

* ODE: improve benchmark interface...

* ODE: BDF changes to use RMS norm and change some default values

Small changes to compare more closely with reference implementation.
Some of these might be reverted eventually but that's fine for now.

* ODE: BDF convergence more stable and results look pretty good now!

Changing the Newton solver convergence criteria as well as changing
a few default input parameters leads to a more stable algorithms
which can now integrate the stiff Henderson autocatalytic example
well in 66 time steps instead of 200k for fixed order integration...

* ODE: BDF fix bug in initial time step calculation

The initial step routine was overwriting the initial right hand side
which led to obvious issues further down the road... now things should
work fine. Need to figure out if I can re-initialize the variables in
the perf test while excluding that time from each iteration.

* ODE: BDF removing bad print statement...

std::cout in device code

* ODE - BDF: improving perf test

Basically adding new untimed setup within the main loop of the
benchmark to reset the intial conditions, buffers and vectors
ahead of each iteration.

* Modifying unit-test to catch proper return type

* Applying clang-format

* cm_test_all_sandia: update caraway compilers

add rocm/5.6.1 and rocm/6.0.0, and openblas/0.3.23 as tpl

* Sparse MKL: changing the location of the MKL_SAFE_CALL macro (#2134)

* Sparse MKL: changing the location of the MKL_SAFE_CALL macro

Moving the macro outside of namespaces to ensure that it will be
interpreted correctly when called from any other location in the
library.

It does not make much sense to guard Impl code in the Experimental
namespace and in this case it cleans up a problem with namespace
disambiguation for the compiler...

* Sparse BsrSpMV: removing Experimental namespace from Impl namespace

* Applying clang-format

* Sparse SpMV: fixing more namespace issues!

* Fixing missing descriptor for bsr spmv

* Kokkos Kernels: change the default offset ETI from size_t to int (#2140)

This change makes it easier for customer to leverage TPL support
which almost always requires offset=int, ordinal=int to be enabled
meaning that no TPL support is available with our default ETI...

* KokkosSparse_spmv_bsrmatrix_spec: fix Bsr_TC_Precision namespacing

Resolve compilation errors in nightly cuda/12.2 A100 build

* Drop comment for cleaner clang-format fix

* Fix usage of RAII to set cusparse/rocsparse stream (#2141)

Temporary objects like "A()" get destructed immediately.
For the object to have scope lifetime, it needs a name like "A a();".
This was causing cusparse/rocsparse spmv to always execute on the default stream,
causing incorrect timing in the spmv perf test.

* Use execution space operator== (#2136)

It actually is part of the public interface

* cm_test_all_sandia: more caraway module updates and cleanup (#2145)

* Spmv perftest improvements (#2146)

* Spmv perf test improvements

- Add option to flush caches by filling a dummy buffer between
iterations
- Add option to call the non-reuse interface instead of handle/reuse
interface
- Fix modes T, H in nonsquare case (make x,y the correct length)

* Fix mode help text

* KokkosKernels Utils: cleaning the zero_vector interface

One of the overload requires an unused template, removing that
extraneous template and simplify how that function is called in
a second overload.

* Kokkos Kernels: fixing call to zero_vector in Gauss-Seidel

* CMakeLists.txt: Update develop to 4.3.99

* Address reviewer comments

* Update CMakeLists.txt

Co-authored-by: brian-kelley <brian.honda11@gmail.com>

* cm_test_all_sandia: solo updates

module updates post TOSS upgrade

* Fix signed/unsigned comparison warnings (#2150)

This is only hit when spmv is called with integer scalars,
which doesn't happen in our CI but does often in Tpetra.

* SPMV tpl fixes, cusparse workaround (#2152)

* SPMV tpl fixes, workaround

* Avoid possible integer conversion warnings

* Document cusparseSpMM algos that were tested

* Applying clang-format

* Use default_size_type as default offset in matrix types (#2149)

Now a declaration like CrsMatrix<Scalar, Ordinal, Device>
will by default use an ETI'd type combination (as int is the default
ETI'd offset)

* spmv bsr perftest: move fences to right place (#2153)

* Kokkos Kernels: removing old code branches ahead of 4.3.00 release (#2139)

Some old code branches kept for compatibility with Kokkos Core
versions less than 4.2.00 are dropped after release 4.3.00.
The largest changes are the removal of support the Kokkos print
macro in favor of Kokkos::printf() and the removal of half support
from Kokkos Kernels since it is now in Kokkos Core.

* Code for running performance measurements on ger() (#2082)

* Correct flop count

* Addressing feedbacks from Luc

* Using 'zero()' instead of '0.'

* KokkosBlas1_axpby.hpp: change debug macro guard for printInformation (#2157)

* KokkosBlas1_axpby.hpp: change debug macro guard for printInformation

- resolves test failures in Trilinos (MueLu) that rely on gold file diff
comparisons by removing extra output in debug builds

* fix compilation error

* Add user tolerance to Serial SVD (#2120)

* Add user tolerance to Serial SVD
---------

Co-authored-by: whorne <whorne@sandia.gov>
Co-authored-by: Carl Pearson <cwpears@sandia.gov>

* Add a simple Harwell-Boeing file reader (#2155)

* Add a simple Harwell-Boeing file reader

And a test that validates against the MM reader.

* Support for symmetrize

* This loop can be simplified, there's no diag duplication

* Improve IO test

* spmv tpls: use correct bool for eti template param (#2160)

(It should just be the default,
KokkosSpars::Impl::eti_spec_avail<..>::value)

* Fix sparse_ioutils test on kokkos-dev (#2162)

Adding a ss.sync call fixes it but I have no idea why it
was needed here. All the other stringstream reads work fine.

* Fix #2156 (#2164)

spmv: add special path for rank-2 x/y, but where both have 1 column
and a TPL is available for rank-1 but not rank-2.

Also call "subhandle->set_exec_space" correctly in the TPLs to ensure
proper synchronization between setup, spmv and cleanup (in the case that
different exec instances are used in different calls)

* Updates from feedback runnig Trilinos testing

- Update debug level to > 1 guarding `printInformation(...)` in
  KokkosBlas1_axpby.hpp to reduce noisy test output
- Loosen tolerance of lapack.svd test to avoid random failures that
  occur near prior tolerance level

* Fix #2167: classic MKL doesn't use space instance (#2168)

* CHANGELOG.md: 4.3.00 update

* KokkosLapack_svd_tpl_spec_decl: defer to MKL spec when LAPACK also enabled

Resolves redefintion of struct SVD compilation errors with both MKL and LAPACK are enabled
Reported by @maartenarnst in https://github.com/trilinos/Trilinos/issues/12891

Co-authored-by: brian-kelley <brian.honda11@gmail.com>

* .github/workflows: Added bdw.yml

* .github/workflows: Added spr.yml

* .github/workflows: Added mi210.yml

* .github/workflows: Added h100.yml

* .github/workflows: Added volta70.yml

* .github/workflows: Added power9.yml and power9_tpls.yml

* Jgfouca/block spiluk fixes (#2172)

* Progress

* Attempt 1, fix multiplication order

* Converges in 1 step

* Various cleanups

* Be sure not to reduce performance of unblocked impl

Also add some comments.

* Remove test mangling

* Fixes for GPU

* Fix warning

* formatting

* Increase eps for floats

* This is no longer needed

* .github/workflows: Add PR_VEGA908_ROCM561_HIP_SERIAL_LEFT_OPENBLAS_OPENLAPACK_REL

* Add guard for cusparse spmv_mv_tpl_spec_avail

Address issue #2175
Configuring with magma tpl enabled and cusparse disabled mistakenly triggers the cusparse tpl avail check to be true
Guard the KOKKOSSPARSE_SPMV_MV_TPL_SPEC_AVAIL_CUSPARSE macros when CUSPARSE is enabled to prevent this

* .github/workflows: Remove OPENLAPACK from names

* .github/workflows: Remove power9 until we have hardware to test it on

* .github/workflows: Enable rocblas in rocm tpl check

* .github/workflows: Remove volta70 until we have hardware to test it on

* Add early return if numRows == 0 in trsv to avoid integer divide-by-zero error

* Resolves multiple definition of Magma and Cuda singletons (#2178)

Address issue #2175

* magma: fix linker errors for builds without cusolver (#2181)

* magma: fix linker errors for builds without cusolver

* BatchedGemm test: workaround testing cublas+magma

- temporary workaround to skip magma test when cublas enabled to avoid issues like #2177

* .github/workflows/mi210: Fix include paths

* workflows/osx.yml: test against most recent kokkos tag

- test against most recent kokkos release rather than develop branch, as
  done with AT CI, to avoid compatibility breakages

* Resolve vortex compilation issue by resolving (potentially) duplicate symbol (#2183)

Stick to pattern of removing leading 'c' or 'z' in method name and relying on the template type

Co-authored-by: malphil <malphil@sandia.gov>

* Changes to enable OneAPI usage with Trilinos build (#2185)

- Get rid of SYCL_OVERRIDE setting
- Add MKL_PROVIDES_BLAS_LAPACK cmake variable and define so that
  code knows whether to use int, or MKL_INT
  - Trilinos builds might link with OneAPI for GPU but standard
    BLAS/LAPACK on CPU

* Fix macOS docs build (#2190)

* Fix docs build

* try docs fix

* make sphinx available at config time

* GH-Actions: adding security actions and scorecard (#2192)

Pretty much taking the new files from PR #2191 and re-creating
it in a clean PR on top of develop with small changes relevant
to our repository.

Cleaning up some workflows to tailor it for our needs

* Scorecard: adding manual dispatch and target default branch (#2195)

* BsrMatrix: Fix HostMirror typedef (#2196)

It needed to have size_type.

* update changelog for 4.3.1

* Update CHANGELOG.md

* docs.yml: change kokkos version to latest release

- avoid version range check issues for release tests

* Sparse - SpGEMM: labeling spgemm_symbolic in TPL layer a bit more clearly (#2193)

This just improves the readability of the output from the tools as it now
has a symbolic matching the numeric phase. Previously we only had spgemm
in the label which is a bit confusing as it could be the whole spgemm time
i.e. both symbolic and numeric, additionally we had symbolic in the MKL
path but not in cusparse, rocsparse...

* SpMV: Test NaN, fix NaN handling when beta=0 (#2188)

* Test_Sparse_spmv_bsr.hpp: add NaNs to tests

* handle NaN in spmv_beta_transpose when beta=0

* handle nan in SpmvMergeHierarchical when beta=0

* Test NaNs in Y, don't reuse modifed Y, catch NaNs in results test

* remove unused <iostream> include

* explicit casting of zero

* Test_sparse_spmv.hpp: remove unused nans parameter

* KokkosSparse_spmv.hpp: CUDA11 can't detect this function always returns

* Test_Sparse_spmv.hpp: remove unused variable

* Run unit tests in correct execution space

* Test_Sparse_spmv.hpp: remove unused type aliases

* Kokkos::nan() -> KokkosKernels::Impl::quiet_NaN()

* Disable cuBLAS dot wrapper (#2206)

(not deleted, just guarded with #if 0 and comments explaining)

It performs significantly worse than our native impl on 11.2, 11.8 and 12.0 on V100.
This is in the dot perf test with a warm-up call.

https://github.com/trilinos/Trilinos/issues/12982 was a symptom of this.

* Fix spmv regressions (#2204)

* Restore cusparse spmv ALG2 path for imbalanced

With correct version cutoffs

* spmv: use separate rank-1 and rank-2 tpl subhandles

* Remove redundant single-column path in native spmv_mv

* Fix unused param warning

* c++17: add [[fallthrough]] attribute (#1493)

* c++17: add [[fallthrough]] attribute

* cm_test_all_sandia: -Wimplicit-fallthrough

* Enable 3 at2 builds (#2210)

* .github/mi210: Enable on PRs

* .github/mi210: Disable non-tpl build

* .github/bdw: Enable PR_BDW_GNU1020_OPENMP_SERIAL_LEFT_OPENBLAS_REL

* .github/h100: Enable PR_HOPPER90_CUDA1180_CUDA_LEFT_RIGHT_REL

* Bump ossf/scorecard-action from 2.0.6 to 2.3.3 (#2214)

Bumps [ossf/scorecard-action](https://github.com/ossf/scorecard-action) from 2.0.6 to 2.3.3.
- [Release notes](https://github.com/ossf/scorecard-action/releases)
- [Changelog](https://github.com/ossf/scorecard-action/blob/main/RELEASE.md)
- [Commits](https://github.com/ossf/scorecard-action/compare/99c53751e09b9529366343771cc321ec74e9bd3d...dc50aa9510b46c811795eb24b2f1ba02a914e534)

---
updated-dependencies:
- dependency-name: ossf/scorecard-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump actions/upload-artifact from 3.1.3 to 4.3.3 (#2215)

Bumps [actions/upload-artifact](https://github.com/actions/upload-artifact) from 3.1.3 to 4.3.3.
- [Release notes](https://github.com/actions/upload-artifact/releases)
- [Commits](https://github.com/actions/upload-artifact/compare/a8a3f3ad30e3422c9c7b888a15615d19a852ae32...65462800fd760344b1a7b4382951275a0abb4808)

---
updated-dependencies:
- dependency-name: actions/upload-artifact
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump actions/dependency-review-action from 2.5.1 to 4.3.2 (#2217)

Bumps [actions/dependency-review-action](https://github.com/actions/dependency-review-action) from 2.5.1 to 4.3.2.
- [Release notes](https://github.com/actions/dependency-review-action/releases)
- [Commits](https://github.com/actions/dependency-review-action/compare/0efb1d1d84fc9633afcdaad14c485cbbc90ef46c...0c155c5e8556a497adf53f2c18edabf945ed8e70)

---
updated-dependencies:
- dependency-name: actions/dependency-review-action
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump step-security/harden-runner from 2.7.1 to 2.8.0 (#2218)

Bumps [step-security/harden-runner](https://github.com/step-security/harden-runner) from 2.7.1 to 2.8.0.
- [Release notes](https://github.com/step-security/harden-runner/releases)
- [Commits](https://github.com/step-security/harden-runner/compare/a4aa98b93cab29d9b1101a6143fb8bce00e2eac4...f086349bfa2bd1361f7909c78558e816508cdc10)

---
updated-dependencies:
- dependency-name: step-security/harden-runner
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump github/codeql-action from 2.25.3 to 3.25.6 (#2216)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 2.25.3 to 3.25.6.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/ceaec5c11a131e0d282ff3b6f095917d234caace...9fdb3e49720b44c48891d036bb502feb25684276)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Kokkos Kernels: initial security policy (#2220)

We state were patches will be made and how to report security vulnerabilities to the project.

* Kokkos Kernels: adding SHA for github actions (#2221)

* Kokkos Kernels: adding SHA for github actions

* Adding sha for docker images

* More dependencies fix (#2222)

* workflows: fix sha for label checker docker image

* workflows: fixing Docker files dependencies.

* Fix Docker files (#2223)

* SPMV TPLs: improve profile region labels (#2219)

- Mark rank-2 versions as "spmv_mv" not "spmv" (the native impl has
  this, and it's useful to know which one is being run)
- Add missing commas separating "BSRMATRIX" and the scalar type name

* cusparse spgemm: provide non-null row-ptr (#2213)

* Bump github/codeql-action from 3.25.6 to 3.25.7 (#2225)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.25.6 to 3.25.7.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/9fdb3e49720b44c48891d036bb502feb25684276...f079b8493333aace61c81488f8bd40919487bd9f)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump actions/checkout from 3.6.0 to 4.1.6 (#2226)

Bumps [actions/checkout](https://github.com/actions/checkout) from 3.6.0 to 4.1.6.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](https://github.com/actions/checkout/compare/f43a0e5ff2bd294095638e18286ca9a3d1956744...a5ac7e51b41094c92402da3b24376905380afc29)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-type: direct:production
  update-type: version-update:semver-major
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* README: adding link to scorecard (#2230)

* spmv_mv wrappers for rocsparse (#2233)

* spmv_mv wrappers for rocsparse (rocsparse_spmm())

* Use consistent types for alpha/beta in spmv wrappers

* Bump step-security/harden-runner from 2.8.0 to 2.8.1 (#2236)

Bumps [step-security/harden-runner](https://github.com/step-security/harden-runner) from 2.8.0 to 2.8.1.
- [Release notes](https://github.com/step-security/harden-runner/releases)
- [Commits](https://github.com/step-security/harden-runner/compare/f086349bfa2bd1361f7909c78558e816508cdc10...17d0e2bd7d51742c71671bd19fa12bdc9d40a3d6)

---
updated-dependencies:
- dependency-name: step-security/harden-runner
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump github/codeql-action from 3.25.7 to 3.25.8 (#2237)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.25.7 to 3.25.8.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/f079b8493333aace61c81488f8bd40919487bd9f...2e230e8fe0ad3a14a340ad0815ddb96d599d2aff)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump actions/dependency-review-action from 4.3.2 to 4.3.3 (#2235)

Bumps [actions/dependency-review-action](https://github.com/actions/dependency-review-action) from 4.3.2 to 4.3.3.
- [Release notes](https://github.com/actions/dependency-review-action/releases)
- [Commits](https://github.com/actions/dependency-review-action/compare/0c155c5e8556a497adf53f2c18edabf945ed8e70...72eb03d02c7872a771aacd928f3123ac62ad6d3a)

---
updated-dependencies:
- dependency-name: actions/dependency-review-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Add testing for transpose corner cases (#2234)

* Add testing for transpose corner cases

crs, bsr, graph: test cases that are 0x0, 100x0 and 0x100. In these cases make
sure the matrix generator doesn't try to insert any entries (nnz = 0).

* Update sparse/unit_test/Test_Sparse_Transpose.hpp

Co-authored-by: Luc Berger <lberge@sandia.gov>

* Update sparse/unit_test/Test_Sparse_Transpose.hpp

Co-authored-by: Luc Berger <lberge@sandia.gov>

---------

Co-authored-by: Luc Berger <lberge@sandia.gov>

* Graph: removing executable from repo (#2239)

* Fix logic around merge path with TPLs (#2240)

SPMV_MERGE_PATH is not always a native algorithm. Add
SPMV_NATIVE_MERGE_PATH to cover that case specifically. Test this new
option.

* spgemm unit test: change matrix value distribution (#2241)

Change the distribution A, B values are sampled from so that
values in C can't end up close to 0 (as the result of summing terms
that are larger). The relative error metric in is_same_matrix is sensitive
to this.

Fixes #2232

* kokkoskernels_tpls.cmake: remove duplicates arguments when creating argument for exported INTERFACE_INCLUDE_DIRECTORIES

Attempt to workaround issue #2238

* Sparse - BsrMatrix: adding new wiki example for documentation (#2228)

There is already an example for this but it uses a CrsMatrix as
starting point to build a BsrMatrix which is not really helpful in
general as the hope is that you can use the BsrMatrix without needing
the CrsMatrix as it would double the storage needed...

Addressing Kim's comments

* Sparse - CrsToBsr: fix type mismatch (#2242)

* Update rocsparse algo defaults (#2245)

* Update default spmv algorithms for rocsparse

- Use stream for common cases (default, fast setup) as it has nearly zero
  setup cost and performs well for somewhat balanced matrices
- Use adaptive (which is rocsparse's default) only if SPMV_MERGE_PATH
  is the algorithm, as it has a very high setup cost

* Re-enable rocsparse spmv for SPMV_FAST_SETUP

* In deprecated spmv, fix Controls algorithm mapping (#2246)

native -> SPMV_NATIVE
native-merge -> SPMV_NATIVE_MERGE_PATH
merge -> SPMV_MERGE_PATH
tpl -> SPMV_FAST_SETUP

* Add batched serial tbsv (#2202)

* Add batched serial tbsv

* remove incx argument and use strided views instead

* Add a new line at the end of files

* fix random number generation for complex numbers

* remove unused variables from internal tbsv serial functions

* remove allclose for testing

---------

Co-authored-by: Yuuichi Asahi <y.asahi@nr.titech.ac.jp>

* Bump actions/checkout from 4.1.6 to 4.1.7 (#2248)

Bumps [actions/checkout](https://github.com/actions/checkout) from 4.1.6 to 4.1.7.
- [Release notes](https://github.com/actions/checkout/releases)
- [Changelog](https://github.com/actions/checkout/blob/main/CHANGELOG.md)
- [Commits](https://github.com/actions/checkout/compare/a5ac7e51b41094c92402da3b24376905380afc29...692973e3d937129bcbf40652eb9f2f61becf3332)

---
updated-dependencies:
- dependency-name: actions/checkout
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump github/codeql-action from 3.25.8 to 3.25.10 (#2249)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.25.8 to 3.25.10.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/2e230e8fe0ad3a14a340ad0815ddb96d599d2aff...23acc5c183826b7a8a97bce3cecc52db901f8251)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* FindTPLROC*: updates to fix export of import targets

Changes for the Rocm tpls to match the handling as done with the Cuda tpls

Should resolve issue #2238

* Fix warning about memcpy (#2252)

When building Stokhos BlockCrs, this util function gave a warning
about memcpy modifying a non-trivially-copyable type. Silence it
by casting to void*

* RCM fixes, improvements (#2254)

* Fix RCM starting vertex issue, improve testing
* apply reversing as labels are computed
instead of at the end. Saves a loop over all the labels
* use min-degree starting vertex within each connected component

* spgemm: add profiling regions to native implementations (#2253)

* spgemm: add profiling regions to native implementations

* Add profiling region to KokkosSPGEMM::KokkosSPGEMM_symbolic

* sparse: replace macros with constexpr bools (#2260)

* Rename `Impl::alignPtr` to `Impl::alignPtrTo`, allow it to infer argument type (#2261)

* KokkosKernels::Impl::alignPtr infers argument type

* Rename KokkosKernels::Impl::alignPtr -> alignPtrTo

* Bump github/codeql-action from 3.25.10 to 3.25.11 (#2263)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.25.10 to 3.25.11.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/23acc5c183826b7a8a97bce3cecc52db901f8251...b611370bb5703a7efb587f9d136a52ea24c5c38c)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* sparse: spadd_symbolic fences before device values used on host (#2259)

* sparse: spadd_symbolic fences before device values used on host

* sparse: use prefix sum to remove explicit spadd fence

* sycl: use alternative `alignPtrTo` when SYCL is enabled (SpGEMM) (#2262)

* sycl: use alternative alignPtr when SYCL is enabled

The current alignPtr, as well as two other alternatives below, do not work on SYCL on Intel PVC.

unsigned int f1(unsigned int i, unsigned int align) {
    return ((i + align - 1) / align * align);
}

unsigned int f2(unsigned int i, unsigned int align) {
    return (i + align - 1) & (-align);
}

* alignPtrTo unit tests

* Help gcc/8.3 with ctad issue

Resolves #2264

Co-authored-by: Carl Pearson <cwpears@sandia.gov>

* Bump actions/upload-artifact from 4.3.3 to 4.3.4 (#2266)

Bumps [actions/upload-artifact](https://github.com/actions/upload-artifact) from 4.3.3 to 4.3.4.
- [Release notes](https://github.com/actions/upload-artifact/releases)
- [Commits](https://github.com/actions/upload-artifact/compare/65462800fd760344b1a7b4382951275a0abb4808...0b2256b8c012f0828dc542b3febcab082c67f72b)

---
updated-dependencies:
- dependency-name: actions/upload-artifact
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* handle_t* -> unique_ptr<handle_t> in Bsr SpMV unit tests (#2269)

* Workarounds for removed cusparse functions (#2270)

cusparse 12.5 removed some functions that were deprecated, like the ILU
factorizations and the legacy csrsv (sparse triangular solve) functions.
As a workaround, if the cusparse version is >= 12.5 then disable the paths
in perftests that call those.

* BLAS - gemv: using fallback when mode is 't' or 'c' and onemkl is used (#2272)

* Implement batched serial pttrf (#2256)

* Batched serial pttrf implementation

* fix: use GEMM to add matrices

* fix: initialization order

* fformat

* fix: temporary variable in a test code

* fix: docstring of pttrf

* check_positive_definitiveness only if KOKKOSKERNELS_DEBUG_LEVEL > 0

* Improve the test for pttrf

* fix: int type

* fix: cleanup tests for SerialPttrf

* cleanup: remove unused deep_copies

* fix: docstrings and comments for pttrf

* ConjTranspose with conj and Transpose

* quick return in pttrf for size 1 or 0 matrix

* Add tests for invalid input

* fix: info computation

---------

Co-authored-by: Yuuichi Asahi <y.asahi@nr.titech.ac.jp>

* A little sptrsv cleanup before the main block effort (#2247)

* Some cleanup and refactoring

* First round of cleanup complete

* Fix a couple warnings

* formatting

* Sparse - SpMV: removing calls to unsuported oneapi - MKL functions (#2274)

* Sycl gemv beta (#2276)

* BLAS - GEMV: zero out Y when beta == 0 in SYCL TPL code path

* BLAS - GEMV: reverting wrong change from previous PR, my bad.

* Applying clang-format

* Unify alignPtrTo implementation (#2275)

* init (#2273)

* Bigger sptrsv cleanup (#2280)

* Some cleanup and refactoring
* Remove Upper/Lower TriLvlSchedTP2SolverFunctors
* Remove Upper/Lower single block functors
* Remove unused TriLvlSchedTP1SingleBlockFunctorDiagValues and merge upper/lower tri_solve_cg
* Merge two big upper/lower branch of tri_solve_chain
* Merge upper/lower tri_solve_streams
* Switch over block spiluk precond test to use new block sptrsv

* Bump actions/dependency-review-action from 4.3.3 to 4.3.4 (#2279)

Bumps [actions/dependency-review-action](https://github.com/actions/dependency-review-action) from 4.3.3 to 4.3.4.
- [Release notes](https://github.com/actions/dependency-review-action/releases)
- [Commits](https://github.com/actions/dependency-review-action/compare/72eb03d02c7872a771aacd928f3123ac62ad6d3a...5a2ce3f5b92ee19cbb1541a4984c76d921601d7c)

---
updated-dependencies:
- dependency-name: actions/dependency-review-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump github/codeql-action from 3.25.11 to 3.25.12 (#2278)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.25.11 to 3.25.12.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/b611370bb5703a7efb587f9d136a52ea24c5c38c...4fa2a7953630fd2f3fb380f21be14ede0169dd4f)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Propose increasing column limit to 120. (#2255)

* Change key files

* Full reformat

* Update format.yml

* Update ubuntu version for format checker

* A couple platforms do not correctly handle static complexes

* .github/workflows: Auto trigger AT2 ci upon review (#2243)

* Bump github/codeql-action from 3.25.12 to 3.25.13 (#2284)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.25.12 to 3.25.13.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/4fa2a7953630fd2f3fb380f21be14ede0169dd4f...2d790406f505036ef40ecba973cc774a50395aac)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump step-security/harden-runner from 2.8.1 to 2.9.0 (#2283)

Bumps [step-security/harden-runner](https://github.com/step-security/harden-runner) from 2.8.1 to 2.9.0.
- [Release notes](https://github.com/step-security/harden-runner/releases)
- [Commits](https://github.com/step-security/harden-runner/compare/17d0e2bd7d51742c71671bd19fa12bdc9d40a3d6...0d381219ddf674d61a7572ddd19d7941e271515c)

---
updated-dependencies:
- dependency-name: step-security/harden-runner
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Add big reformat commits to ignore revs for blame (#2286)

* Bump github/codeql-action from 3.25.13 to 3.25.15 (#2288)

Bumps [github/codeql-action](https://github.com/github/codeql-action) from 3.25.13 to 3.25.15.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/2d790406f505036ef40ecba973cc774a50395aac...afb54ba388a7dca6ecae48f608c4ff05ff4cc77a)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Bump ossf/scorecard-action from 2.3.3 to 2.4.0 (#2287)

Bumps [ossf/scorecard-action](https://github.com/ossf/scorecard-action) from 2.3.3 to 2.4.0.
- [Release notes](https://github.com/ossf/scorecard-action/releases)
- [Changelog](https://github.com/ossf/scorecard-action/blob/main/RELEASE.md)
- [Commits](https://github.com/ossf/scorecard-action/compare/dc50aa9510b46c811795eb24b2f1ba02a914e534...62b2cac7ed8198b15735ed49ab1e5cf35480ba46)

---
updated-dependencies:
- dependency-name: ossf/scorecard-action
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* Sparse - coo2csr: deactivating test on PVC (#2290)

* Sparse - coo2csr: deactivating test on PVC

With this test deactivated everything in Kokkos Kernels passes.
Let us confirm that with a nightly build and potentially create
of SYCL CI-build.

* Protecting SYCL check with preprocessor guard

* cmake: add CMake language support for CUDA/HIP (#2173)

* cmake: add CMake language support for CUDA/HIP

* Apply suggestions of @brian-kelley

* Make sure clang is using a newer GCC on weaver (#2289)

It's clear from the name of the AT job GCC930_Light_Tpls_GCC930_Tpls_CLANG13CUDA10 that
the intent was to use gcc-9.3 but gcc-7.4.0 was the one actually being
used because the clang-13 module loads that gcc.

* Set version number to 4.4.0

* Revert "Changes to enable OneAPI usage with Trilinos build (#2185)"

This reverts commit b87dc951f64463fe5d6d4d886c0496ee804b2715.

* Add support for BSRs to sptrsv (#2281)

* Some cleanup and refactoring

* First round of cleanup complete

* Fix a couple warnings

* formatting

* Cleanup progress

* Fixes

* Remove Upper/Lower TriLvlSchedTP2SolverFunctors

* Remove Upper/Lower single block functors

* Remove unused TriLvlSchedTP1SingleBlockFunctorDiagValues and merge upper/lower tri_solve_cg

* Merge two big upper/lower branch of tri_solve_chain

* Merge upper/lower tri_solve_streams

* progres

* progress

* Progress, test added

* progress and fixes

* prog

* prog but broken

* Prog with debug prints

* Remove extreme debug printing

* works

* all working

* Remove test mangling

* Switch over block spiluk precond test to use new block sptrsv

* More test cleanup

* Fixes for GPU warnings

* Conflicts resolved, still work needed

* builds

* Formatting

* Update work to latest format style

* Remove unused functions. Remove prints. Add barriers

* Minor fixes

* lset is not used

* Fix for clang

* formatting

* New impl approach

* Unset macro

* There's no reason to limit team-policy alg to MAX_VEC_SIZE

* formatting

* Add missing kokkos-inline-funcs

* Fix warnings

* format

* Some compilers throw shadow warnings in static functions (#2297)

* Update changelog for 4.4.00

* Remove SYCL GEMV update

The SYCL GEMV update has been reverted in a later PR as the fix required was for the GEMV algorithm in the Sparse namespace, also known as SpMV. The previous BLAS::gemv fix was reverted

* Add entry for #2281, remove #2185

* docs.yaml: update kokkos version to 4.3.01

* update min kokkos version to 4.3.01

* apply clang-format

* update yaml files to use kokkos v 4.3.01

* github workflow: adding release workflow to have provenance for release artifacts (#2304)

This adds a workflow based on the scorecard recommendation for software provenance.
The workflow requires some manual actions to verify that the signiture in the release is correct.

* update master_history.txt for 4.4.00

---------

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: Luc Berger <lberge@sandia.gov>
Co-authored-by: Carl Pearson <cwpearson@users.noreply.github.com>
Co-authored-by: Carl Pearson <cwpears@sandia.gov>
Co-authored-by: Ernesto Prudencio <eeprude@sandia.gov>
Co-authored-by: Brian Kelley <bmkelle@sandia.gov>
Co-authored-by: James Foucar <jgfouca@sandia.gov>
Co-authored-by: Junchao Zhang <jczhang@anl.gov>
Co-authored-by: Damien L-G <dalg24+github@gmail.com>
Co-authored-by: Caleb Schilly <cwschilly@gmail.com>
Co-authored-by: Damien L-G <dalg24@gmail.com>
Co-authored-by: Sean Miller <sean.miller@amd.com>
Co-authored-by: Vinh Dang <vqdang@sandia.gov>
Co-authored-by: Luc Berger-Vergiat (-EXP) <lberge@trappist.son.sandia.gov>
Co-authored-by: brian-kelley <brian.honda11@gmail.com>
Co-authored-by: eeprude <119708086+eeprude@users.noreply.github.com>
Co-authored-by: Wyatt Horne <68676884+wjhorne@users.noreply.github.com>
Co-authored-by: whorne <whorne@sandia.gov>
Co-authored-by: Evan Harvey <eharvey@sandia.gov>
Co-authored-by: malphil <malphil@sandia.gov>
Co-authored-by: Evan Harvey <57234914+e10harvey@users.noreply.github.com>
Co-authored-by: Malachi <malachi2@illinois.edu>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
Co-authored-by: yasahi-hpc <57478230+yasahi-hpc@users.noreply.github.com>
Co-authored-by: Yuuichi Asahi <y.asahi@nr.titech.ac.jp>
Co-authored-by: Daniel Arndt <arndtd@ornl.gov>
Co-authored-by: Baptiste Legouix <stilynx51@gmail.com>
Co-authored-by: Richard Berger <richard.berger@outlook.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
Projects
None yet
Development

No branches or pull requests

2 participants