Kokkos promotion #5432
Merged
Kokkos promotion #5432
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Internal implementation details for Kokkos::ParallelFor with Cuda changed with the enhancement for Cuda streams support. This commit updates Stokhos internal implementation to be compatible with Kokkos.
Changes made following removal of StaticAssert from Kokkos Changes to be committed: modified: packages/stokhos/test/Performance/FadMPAssembly/VectorImport.hpp modified: packages/stokhos/test/Performance/MPAssembly/VectorImport.hpp
Add enum is_assignable_data_type for compatibility with Kokkos changes for release 2.9
Add enum is_assignable_data_type for compatibility with Kokkos changes for release 2.9
See kokkos/kokkos#2140 The fence as a static member function of execution spaces was deprecated and made non-static. Impacted packages: ifpack2, intrepid2, moertel, panzer, phalanx, sacado, shylu (fastilu and tacho), stokhos, teuchos, tpetra, trilinoscouplings
Updates corresponding to kokkos/kokkos#2136 Changes to be committed: modified: ifpack2/src/Ifpack2_BlockTriDiContainer_impl.hpp
1. Change when_all to be called as member of scheduler 2. Change typedef of scheduler to DeprecatedTaskScheduler 3. Updates for Kokkos memory trait check names (non-deprecated)
currently, tasking api does not deallocate tasks properly
…deprecated schedulers
Call explicit modify_host method instead of templated version to prevent issues with Cuda UVM
Script used for kokkos + kokkos-kernels integration testing
exec_space::fence() deprecated in Kokkos 2.9; replace with call from instantiated exec_space - exec_space().fence()
SharedMemoryCapacity was removed from Kokkos, this commit updates stokhos with the accompanying change in Kokkos. Changes to be committed: modified: packages/stokhos/src/kokkos/Cuda/Stokhos_Cuda_CooProductTensor.hpp modified: packages/stokhos/src/kokkos/Cuda/Stokhos_Cuda_SimpleTiledCrsProductTensor.hpp modified: packages/stokhos/src/kokkos/Cuda/Stokhos_Cuda_TiledCrsProductTensor.hpp
Updates needed for compatibility with Kokkos deprecated code changes
… into kokkos-promotion
|
Status Flag 'Pre-Test Inspection' - Auto Inspected - Inspection Is Not Necessary for this Pull Request. |
|
Status Flag 'Pull Request AutoTester' - Testing Jenkins Projects: Pull Request Auto Testing STARTING (click to expand)Build InformationTest Name: Trilinos_pullrequest_gcc_4.8.4
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_intel_17.0.1
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_gcc_4.9.3_SERIAL
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_gcc_7.2.0
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_cuda_9.2
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_python_2
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_python_3
Jenkins Parameters
Using Repos:
Pull Request Author: ndellingwood |
|
Status Flag 'Pull Request AutoTester' - Jenkins Testing: 1 or more Jobs FAILED Note: Testing will normally be attempted again in approx. 2 Hrs 30 Mins. If a change to the PR source branch occurs, the testing will be attempted again on next available autotester run. Pull Request Auto Testing has FAILED (click to expand)Build InformationTest Name: Trilinos_pullrequest_gcc_4.8.4
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_intel_17.0.1
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_gcc_4.9.3_SERIAL
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_gcc_7.2.0
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_cuda_9.2
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_python_2
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_python_3
Jenkins Parameters
Console Output (last 100 lines) : Trilinos_pullrequest_gcc_4.8.4 # 3900 (click to expand)Console Output (last 100 lines) : Trilinos_pullrequest_intel_17.0.1 # 3716 (click to expand)Console Output (last 100 lines) : Trilinos_pullrequest_gcc_4.9.3_SERIAL # 2161 (click to expand)Console Output (last 100 lines) : Trilinos_pullrequest_gcc_7.2.0 # 1940 (click to expand)Console Output (last 100 lines) : Trilinos_pullrequest_cuda_9.2 # 1565 (click to expand)Console Output (last 100 lines) : Trilinos_pullrequest_python_2 # 79 (click to expand)Console Output (last 100 lines) : Trilinos_pullrequest_python_3 # 79 (click to expand) |
ndellingwood
added
the
AT: WIP
|
One more Tacho update for the The intel build died due to license server issues link. |
ndellingwood
removed
the
AT: WIP
|
Status Flag 'Pre-Test Inspection' - Auto Inspected - Inspection Is Not Necessary for this Pull Request. |
|
Status Flag 'Pull Request AutoTester' - Testing Jenkins Projects: Pull Request Auto Testing STARTING (click to expand)Build InformationTest Name: Trilinos_pullrequest_gcc_4.8.4
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_intel_17.0.1
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_gcc_4.9.3_SERIAL
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_gcc_7.2.0
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_cuda_9.2
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_python_2
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_python_3
Jenkins Parameters
Using Repos:
Pull Request Author: ndellingwood |
|
Status Flag 'Pull Request AutoTester' - Jenkins Testing: all Jobs PASSED Pull Request Auto Testing has PASSED (click to expand)Build InformationTest Name: Trilinos_pullrequest_gcc_4.8.4
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_intel_17.0.1
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_gcc_4.9.3_SERIAL
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_gcc_7.2.0
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_cuda_9.2
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_python_2
Jenkins Parameters
Build InformationTest Name: Trilinos_pullrequest_python_3
Jenkins Parameters
|
|
Status Flag 'Pre-Merge Inspection' - - This Pull Request Requires Inspection... The code must be inspected by a member of the Team before Testing/Merging |
|
All Jobs Finished; status = PASSED, However Inspection must be performed before merge can occur... |
@ndellingwood - looks like this ran ok last night. These sorts of issues happen a bit and all we can really do is call the team that manages the license server. I will make sure they know they had an issue yesterday. |
|
@ndellingwood @trilinos/framework - there was an explanation for the intel license issue in one of my e-mails this morning
So it should be good to go at this point. |
|
Thanks everyone! @ndellingwood and I will merge this in coordination with Kokkos and KokkosKernels today. |
ibaned
approved these changes
Jun 26, 2019
|
Status Flag 'Pre-Merge Inspection' - SUCCESS: The last commit to this Pull Request has been INSPECTED AND APPROVED by [ ibaned ]! |
|
Status Flag 'Pull Request AutoTester' - AutoMerge IS ENABLED, but the Label AT: AUTOMERGE is not set. Either set Label AT: AUTOMERGE or manually merge the PR... |
|
@trilinos/kokkos, @trilinos/kokkos-kernels, @ndellingwood, From looking on CDash today and in the list of failing tests, it looks like most of these failures were triggered by this update of Kokkos and/or KokkkosKernels. I will wait until tomorrow to see how many of these repeat and then create new Issues for them. |
|
@bartlettroscoe I setup a clean build on Waterman with the develop branch to begin testing with shared libs on (the Trilinos-atdm-waterman_cuda-9.2_shared_opt build had the most failures). So far I am unable to reproduce the Panzer failures when running the tests in an interactive session. There is no option in waterman's SHA a69b6ab Configure Panzer test results: |
|
@ndellingwood said:
Let's see what happens in automated testing tomorrow to see if these failures are repeatable.
You are not restricted to the builds with build options listed in that file. That is just the builds that we are testing. You can try any combination of build options that you can construct described here. |
|
@bartlettroscoe if these are things triggered by Kokkos promotion that will block Trilinos PR testing, then we need the exact scripts and arguments to reproduce each of these builds, and we need to move this discussion into a Trilinos issue. |
|
The CDash-reported Tacho failure passed with this configuration when attempting to reproduce: Configure Ctest: |
|
Tested the CDash-reported Tpetra failures with the fpic configuration below on Waterman, Configure Ctest |
|
PR #5449 submitted to address the kokkos-kernels timeout with the serial graph tests. |
|
Back to the timed out Tpetra test Ctest - waterman7 node Previous timeout test occurred when on the waterman6 node. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This promotes Kokkos and KokkosKernels to Version 2.9.00.
@trilinos/kokkos @trilinos/kokkos-kernels
This update includes the following:
Kokkos
CudaStream support; TeamVectorRange (new policy); TaskScheduler updates (multi-queue); HPX backend; deep_copy within parallel regions; more reduction options to ScatterView
KokkosKernels
KokkosBatched - remove from Experimental namespace, add specializations for float2, float4, double4; Experimental sparse triangular solve initial capability; MAGMA GESV TPL support; CudaUVMSpace specializations for cuBLAS
Kokkos Changelog
2.9.00 (2019-06-24)
Full Changelog
Implemented enhancements:
create\_mirror\_view\_without\_initializing#1765Fixed bugs:
KokkosKernels Changelog
2.9.00 (2019-06-24)
Full Changelog
Implemented enhancements:
Fixed bugs: