-
Notifications
You must be signed in to change notification settings - Fork 575
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
Tpetra: improve performance of unpackAndCombineWithOwningPIDsCount #11694
Labels
Comments
6 tasks
Note: |
jhux2
added a commit
that referenced
this issue
Jul 14, 2023
Merges Tpetra_CrsMatrix methods unpackAndCombineWithOwningPIDsCount and unpackAndCombineIntoCrsArrays. Reduces number of deep copies. Part of larger effort to have TAFC run on device. Temporary change in Tpetra_CrsMatrix_def.hpp: destMat->numImportPacketsPerLID_.modify_host() because numImportPacketsPerLID_ is a Kokkos::DualView and hasn't been properly marked as modified on host Addresses #11693 and #11694.
jhux2
added a commit
that referenced
this issue
Sep 2, 2023
Merges Tpetra_CrsMatrix methods unpackAndCombineWithOwningPIDsCount and unpackAndCombineIntoCrsArrays. Reduces number of deep copies. Part of larger effort to have TAFC run on device. Temporary change in Tpetra_CrsMatrix_def.hpp: destMat->numImportPacketsPerLID_.modify_host() because numImportPacketsPerLID_ is a Kokkos::DualView and hasn't been properly marked as modified on host Addresses #11693 and #11694.
csiefer2
pushed a commit
that referenced
this issue
Sep 12, 2023
Merges Tpetra_CrsMatrix methods unpackAndCombineWithOwningPIDsCount and unpackAndCombineIntoCrsArrays. Reduces number of deep copies. Part of larger effort to have TAFC run on device. Temporary change in Tpetra_CrsMatrix_def.hpp: destMat->numImportPacketsPerLID_.modify_host() because numImportPacketsPerLID_ is a Kokkos::DualView and hasn't been properly marked as modified on host Addresses #11693 and #11694.
cwpearson
pushed a commit
to cwpearson/Trilinos
that referenced
this issue
Sep 19, 2023
Merges Tpetra_CrsMatrix methods unpackAndCombineWithOwningPIDsCount and unpackAndCombineIntoCrsArrays. Reduces number of deep copies. Part of larger effort to have TAFC run on device. Temporary change in Tpetra_CrsMatrix_def.hpp: destMat->numImportPacketsPerLID_.modify_host() because numImportPacketsPerLID_ is a Kokkos::DualView and hasn't been properly marked as modified on host Addresses trilinos#11693 and trilinos#11694.
cwpearson
pushed a commit
to cwpearson/Trilinos
that referenced
this issue
Sep 19, 2023
Replace unpackAndCombineIntoCrsArrays version that accepts only Teuchos Arrays/ArrayRCPs with one that accepts some Kokkos views. (The conversion isn't complete, as upstream/downstream changes are still to be done.) Part of trilinos#11693 and trilinos#11694.
cwpearson
pushed a commit
to cwpearson/Trilinos
that referenced
this issue
Sep 19, 2023
Merges Tpetra_CrsMatrix methods unpackAndCombineWithOwningPIDsCount and unpackAndCombineIntoCrsArrays. Reduces number of deep copies. Part of larger effort to have TAFC run on device. Temporary change in Tpetra_CrsMatrix_def.hpp: destMat->numImportPacketsPerLID_.modify_host() because numImportPacketsPerLID_ is a Kokkos::DualView and hasn't been properly marked as modified on host Addresses trilinos#11693 and trilinos#11694.
cwpearson
pushed a commit
to cwpearson/Trilinos
that referenced
this issue
Sep 19, 2023
Replace unpackAndCombineIntoCrsArrays version that accepts only Teuchos Arrays/ArrayRCPs with one that accepts some Kokkos views. (The conversion isn't complete, as upstream/downstream changes are still to be done.) Part of trilinos#11693 and trilinos#11694.
completed |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Enhancement
@trilinos/tpetra
unpackAndCombineWithOwningPIDsCount
is a hotspot when profiling transfer-and-fill-complete on device.Timer name: "TAFC unpack-count-resize"
See extensive discussions in #2267 and #2556 about previous refactoring.
Part of #11689.
The text was updated successfully, but these errors were encountered: