Changes for CCCL 2.7#97
Closed
assistant-librarian[bot] wants to merge 3 commits into
Closed
Conversation
assistant-librarian
Bot
requested review from
RobsonRLemos,
lawruble13 and
umfranzw
as code owners
ammallya
pushed a commit
that referenced
this pull request
Sep 24, 2025
* Engine config descriptor integration, and changes to support it (#85) * Engine plugin resource manager: engine descriptor integration (#93) * Fix dl error * Engine_descriptor: integrate with engine plugin resource manager * Fix tests * Engine config descriptor integration: bugfix (#97) * Engine plugin resource manager: fixed incorrect usage of the flatbuffer verifier * test_engine_descriptor: fixed a segfault * Mock_engine_plugin_resource_manager: added more methods * Integration clean up changes (#94) * Add friend class to allow access to unsafe casts for mock objects * Standardize flatbuffer and plugin data across engine_config and graph * Format and test disable * Convert FakePlugin to actual test plugin (#96) * Removing dead code, cleaning up CMakeLists.txt * Using stream for the kernel * Moved test plugin to a better location (#98) * Add engine heuristic integration, and fix tests (#100) * Add the execute integration that was missing. * Engine plugin resource manager: execution plan descriptor integration (#101) * Integrate execution plan descriptor * Fix formatting * Fix formatting * [ALMIOPEN-139] Engine plugin loading paths & mode (#87) * Loading multiple plugin paths * Load .so or dir and frontend integration * Error handling * Plugin loading mode * Add file to cmake * .dll or .so and fix tests * Types and API logging * More tests * Update names to include "engine", refine loading mode, remove catch * Small fix * Trivial review concerns * Use platform opaque filenames in tests, defer to shared library logic, remove tests * Load default plugins based on address of parent library and load directory or file correctly * Delineate between default plugin directories and ambiguous paths * Shared library tests * Plugin loading sanity check and todos * Cleanup * Update docstring * Cleanup plugin core and shared library * Use test plugin for tests and integrate plugin loading * Fix engine API tests due to fake engine Ids now being -1 * Fix duplication missed on merge * Review concerns: refactor plugin loading, add tests, remove fallback, absolute override * Remove kernel launch from test_good_plugin and device linking. Comment out checks to allow tests to pass until we finalize integration. * Remove anchor function * Fix for ASAN builds in Clang 20 * Better ASAN output * Fix for memory stomp on engine descriptor * Add exceution plan details to execute, and update testing plugin. --------- Co-authored-by: Brian Harrison <brian.harrison@amd.com> Co-authored-by: mousdahl-amd <mitch.ousdahl@amd.com> * MiOpen Batchnorm fwd inference integration test (#102) * add first bit of integration test, fix bug in graph where it incorrectly passed the descriptor to the backend. * can now do the full frontend graph -> backend -> miopen plugin flow and get all success stats. * clean up test, fully template it so we can easilly use the other types. * clean up and add other data types * get code in place for new plugin functionality. * fixes * fixes * fix test failures * fixes * fix * refactor the type to type enum function to proper place * done need this stuff * fix * better comments * Fix compiler error (#104) * Set plugin logging callback after load --------- Co-authored-by: Adam Dickin <adam.dickin@amd.com> Co-authored-by: Evgenii Averin <86725875+averinevg@users.noreply.github.com> Co-authored-by: Mitchell Ousdahl <mitch.ousdahl@amd.com> Co-authored-by: Samuel Reeder <41528605+SamuelReeder@users.noreply.github.com> Co-authored-by: Samuel Reeder <samuel.reeder@amd.com>
ammallya
pushed a commit
that referenced
this pull request
Sep 24, 2025
* Engine config descriptor integration, and changes to support it (#85) * Engine plugin resource manager: engine descriptor integration (#93) * Fix dl error * Engine_descriptor: integrate with engine plugin resource manager * Fix tests * Engine config descriptor integration: bugfix (#97) * Engine plugin resource manager: fixed incorrect usage of the flatbuffer verifier * test_engine_descriptor: fixed a segfault * Mock_engine_plugin_resource_manager: added more methods * Integration clean up changes (#94) * Add friend class to allow access to unsafe casts for mock objects * Standardize flatbuffer and plugin data across engine_config and graph * Format and test disable * Convert FakePlugin to actual test plugin (#96) * Removing dead code, cleaning up CMakeLists.txt * Using stream for the kernel * Moved test plugin to a better location (#98) * Add engine heuristic integration, and fix tests (#100) * Add the execute integration that was missing. * Engine plugin resource manager: execution plan descriptor integration (#101) * Integrate execution plan descriptor * Fix formatting * Fix formatting * [ALMIOPEN-139] Engine plugin loading paths & mode (#87) * Loading multiple plugin paths * Load .so or dir and frontend integration * Error handling * Plugin loading mode * Add file to cmake * .dll or .so and fix tests * Types and API logging * More tests * Update names to include "engine", refine loading mode, remove catch * Small fix * Trivial review concerns * Use platform opaque filenames in tests, defer to shared library logic, remove tests * Load default plugins based on address of parent library and load directory or file correctly * Delineate between default plugin directories and ambiguous paths * Shared library tests * Plugin loading sanity check and todos * Cleanup * Update docstring * Cleanup plugin core and shared library * Use test plugin for tests and integrate plugin loading * Fix engine API tests due to fake engine Ids now being -1 * Fix duplication missed on merge * Review concerns: refactor plugin loading, add tests, remove fallback, absolute override * Remove kernel launch from test_good_plugin and device linking. Comment out checks to allow tests to pass until we finalize integration. * Remove anchor function * Fix for ASAN builds in Clang 20 * Better ASAN output * Fix for memory stomp on engine descriptor * Add exceution plan details to execute, and update testing plugin. --------- Co-authored-by: Brian Harrison <brian.harrison@amd.com> Co-authored-by: mousdahl-amd <mitch.ousdahl@amd.com> * MiOpen Batchnorm fwd inference integration test (#102) * add first bit of integration test, fix bug in graph where it incorrectly passed the descriptor to the backend. * can now do the full frontend graph -> backend -> miopen plugin flow and get all success stats. * clean up test, fully template it so we can easilly use the other types. * clean up and add other data types * get code in place for new plugin functionality. * fixes * fixes * fix test failures * fixes * fix * refactor the type to type enum function to proper place * done need this stuff * fix * better comments * Fix compiler error (#104) * Set plugin logging callback after load --------- Co-authored-by: Adam Dickin <adam.dickin@amd.com> Co-authored-by: Evgenii Averin <86725875+averinevg@users.noreply.github.com> Co-authored-by: Mitchell Ousdahl <mitch.ousdahl@amd.com> Co-authored-by: Samuel Reeder <41528605+SamuelReeder@users.noreply.github.com> Co-authored-by: Samuel Reeder <samuel.reeder@amd.com> [ROCm/hipDNN commit: ee34ef0]
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Most of the changes here are pulling in the commits from CCCL 2.7. With the fixes and workarounds attached to these commits.
🔁 Imported from ROCm/rocThrust#557
🧑💻 Originally authored by @NB4444