* init
* fix source dir
* miopen specify test build dir
* fix test build dir
* revert change
* fix test build again
* move to ultra temporarily
* miopen-get-ck, working dir
* exclude flaky test
* move back to high
* Add MIVisionX and AMDMIGraphX downstream jobs to MIOpen
* comment sparsecheckoutdir
* quote component names
* fix artifact name
* miopen ck script exit on fail
* add downstream checkout repos
* mivisionx, add aomp
- Add knobs to toggle aggregate build options.
- Aggregate build pipeline will pull ROCm dependencies from earlier in the same pipeline.
- Changing build pool of some components for more compute power.
- Deleting deprecated component.
- Add Ninja to dependency compilation in MIOpen.
- Add retries to wget for MIOpen CK build case.
---------
Co-authored-by: Daniel Su <danielsu@amd.com>
- Also, fixes to rocprofiler-sdk when incorporating recent features.
- URL encoding algorithm converts trailing '=' in the base64 string to an integer representing the number of those trailing '=' characters.
- Dynamically write a Dockerfile based on the environment for the failing job.
- Account for additional dependencies that need to be installed and setup.
- Build and push a custom container based on that dynamic Dockerfile to capture that failing environment.
- Documenting additional setup to install Docker on VMSS during provisioning.
* Change AMDMIGraphX to use local-artifact-download for half 5.6
* Refactor dependencies-rocm & artifact-download, consolidate component variable lists
* Add mainline option to nightly
* Change all components to new dependencies-rocm usage
* rm aqlprofile checkoutRef
* simplify dependencies-rocm, add gpuTarget back to rocMLIR
* rm tag-builds from aqlprofile
* Make review changes
- Swap to default gcc on OS to resolve build errors from recent commits.
- Added libdnnl-dev dependency from iterative attempts with compiler change.
- Referred to the passing GitHub checks to observe the compilers that was used.
- Build CK jit lib and include in AMDMIGraphX build.
* Added all variables found in the library page on Azure
* removed extra space
* copied the example of referencing variables from variables-global.yml and add HALF560_PIPELINE_ID to the file
* introduced variables-global.yml to this file and pointed the path to variables.CCACHE_DIR
* introduced variables-global.yml and changed all variables in stagingPipelineIdentifiers and taggedPipelineIdentifiers to match the identifier names in variables-global.yml
* adjusted how the variables are introduced into the file
* tried adding ./ to variables-global.yml path
* copied the formatting from develop branch but changed identifiers to match them in variables-global.yml
* changed build pool to high to test if variable works
* recopied variables from library page to account for any changes
* changed build pool back to medium
* removed extra whitespace
* remove whitespace
* added all the variables from the page on azure
* fix merge
fix merge
---------
Co-authored-by: Daniel Su <danielsu@amd.com>
Extension of PR #3544 and additional logic for ROCm dependency downloads to account for gpu target for components that can specify GPU target when building or have direct dependencies of these components. Plus, refactoring if statements to reduce lines of code.
* Switch all pipeline gpu targets to gfx942
* Change more pipelines target to gfx942
* set variables for manual testing
* Switch all pipeline gpu targets to gfx942
* Change more pipelines target to gfx942
* set variables for manual testing
* add test pipeline id
* revert test changes
* correct gpu target name
* remove unused flags; change hipSPARSELt target to be gfx942
With MIOpen now building with latest source on External CI, this unblocked AMDMIGraphX from building with latest source.
Determined rocMLIR also needed to be built with latest source as a dependency.
- Fixed compilers to pick amdclang.
- Added ldconfig step for setting up linking of shared libraries.
- Set Azure VMs to medium only.
- Remove empty directories in published tarballs.
To best determine hardware specs per repo, added more build pool options with varying number of vCPUs, RAM size, etc. and will kick off builds with test targets enabled to determine long-term cost values.
Co-authored-by: alexxu-amd <alexxu12@amd.com>
