-
Notifications
You must be signed in to change notification settings - Fork 578
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
Trilinos Master Merge PR Generator: Auto PR created to promote from master_merge_20190517_000612 branch to master #5195
Conversation
@trilinos/tpetra The result may be more amendable to kernel fusion. Tests pass with CUDA_LAUNCH_BLOCKING=1. Some test failures with that environment variable unset. I am working on diagnosing those.
@trilinos/tpetra It had worked before for multiple arguments, when I was using C++14 generic lambdas. My fix to make withLocalAccess work for C++11 broke the multiple arguments case. This commit fixes that. I found this issue when I wrote an example (actually a test, but it's really meant to be read) of how to extend withLocalAccess to support object types other than Tpetra::MultiVector and Tpetra::Vector. This commit also includes the example.
@trilinos/tpetra Tpetra developers argued against for_each taking indices as well as a reference to the value to modify. Thus, I have removed tests of Tpetra::for_each that exercise these overloads. I have also improved Tpetra::transform's tests.
@trilinos/tpetra Now that withLocalAccess works again with multiple arguments, exploit this to simplify transform's implementation for MultiVectors and Vectors.
@trilinos/tpetra We're getting rid of the Tpetra::for_each overloads that take (IST&,LO) and (IST&,LO,LO). This commit fixes the implementation of Tpetra::transform for Tpetra::MultiVector and Tpetra::Vector, so that it no longer calls these overloads of Tpetra::for_each.
@trilinos/tpetra Per request by @crtrott and the other Tpetra developers, purge the overloads of Tpetra::for_each that take user functions of the form (IST&,LO) or (IST&,LO,LO). (The first takes a nonconst reference and the row index, while the second takes a nonconst reference, the row index, and the column index.) This will prevent users from assuming that they can have loop-carried dependencies in their function (they cannot).
@trilinos/tpetra Make GetNonowningLocalObject templated on the same LocalAccess specialization as its corresponding GetMasterLocalObject specialization, rather than templating it on the result type of getMasterLocalObject. This will avoid "collisions" between users' specializations, if different GlobalObjectType types have the same master local object types.
@trilinos/tpetra 1. Intel 17.0.1 insists that a template parameter of type const AccessMode get static_cast to AccessMode before I can == or != it with another AccessMode instance. Otherwise, it gives build errors of the form "error: a built-in binary operator applied to a scoped enumeration requires two operands of the same type." Using static_cast in this way makes withLocalAccess tests build and pass for Intel 17.0.1. (They were already fine with Clang, GCC, and NVCC+GCC.) 2. Intel 17.0.1 did not like the private friend declarations of readOnly, writeOnly, and readWrite inside LocalAccess. The compiler thought they were redeclarations. We don't need these to be friends of LocalAccess anyway, and when I removed them, all the for_each and transform tests built and passed with Intel 17.0.1. (They were already fine with Clang, GCC, and NVCC+GCC.)
@trilinos/tpetra
Automatically Merged using Trilinos Pull Request AutoTester PR Title: SEACAS: ioss/cgns - fix update_db_zone_property PR Author: gsjaardema
Automatically Merged using Trilinos Pull Request AutoTester PR Title: Tpetra: "New RTI" (Reduction Transform Interface), Part 1 (for_each, transform) PR Author: mhoemmen
Automatically Merged using Trilinos Pull Request AutoTester PR Title: Teuchos: replacing not with ! for MSVC compiler PR Author: lucbv
…_final_point Automatically Merged using Trilinos Pull Request AutoTester PR Title: Piro: reporting final point to the wrapped ME at the end of NOX solve PR Author: bartgol
Status Flag 'Pre-Test Inspection' - Auto Inspected - Inspection Is Not Necessary for this Pull Request. |
Status Flag 'Master Merge 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
Using Repos:
Pull Request Author: trilinos-autotester |
Status Flag 'Master Merge 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
|
Status Flag 'Pre-Merge Inspection' - Auto Inspected - Inspection Is Not Necessary for this Pull Request. |
Status Flag 'Master Merge AutoTester' - Pull Request will be Automerged |
Merge on Pull Request# 5195: IS A SUCCESS - Pull Request successfully merged |
Auto PR created to promote from master_merge_20190517_000612 branch to master