Enable ROCm CI support#1786

Open

akashveramd wants to merge 88 commits intomainfrom

torchtitan_rocm_support_main

Commits on Sep 26, 2025

Added support to run torchtitan tests on ROCm.
akashveramd
committed
Added rocm ci support for integration_test_h100.
akashveramd
committed
Fixed a bug in build script. Removed ubuntu-cuda folder, instead using ubuntu folder for cuda Dockerfile.
akashveramd
committed
Added tests.integration_tests.features during rebase.
akashveramd
committed
Modified docker-builds.yml to build rocm docker image for torchtitan.
akashveramd
committed
Fixed runner for cuda and rocm images in docker-builds.yml.
akashveramd
committed
Added TEST_WITH_ROCM environment variable for running tests on rocm. Fixed error in integration_tests.py. Fixed lint errors.
akashveramd
committed
Made additional changes to tests.integration_tests.features during rebase.
akashveramd
committed
Changed runner to i-0962598bd0e8298b3 for building ROCm docker image.
akashveramd
committed
Changed runner to linux.12xlarge for building ROCm docker image.
akashveramd
committed
Changed runner to linux.2xlarge for building ROCm docker image.
akashveramd
committed
Resolved conflict in .github.workflows.integration_test_8gpu_models during rebase.
akashveramd
committed
Changed rocm docker image name in docker-builds.yml.
akashveramd
committed
Reverted the changes to integration_test_8gpu_h100.yaml.
akashveramd
committed
Empty dummy commit.
akashveramd
committed
Increased the timeout to 45 minutes to override timeout used in linux_job_v2.yml for integration_test_8gpu.yaml.
akashveramd
committed
Empty dummy commit.
akashveramd
committed
Added aws setup in the integration_test_8gpu workflow.
akashveramd
committed
Performed rebase and made changes to include code refactoring done upstream.
akashveramd
committed
Changed rocm runner name.
akashveramd
committed
Added a change to run build-test after aws-setup.
akashveramd
committed
Changed the test name in integration_test_8gpu.yaml workflow file.
akashveramd
committed

Commits on Sep 29, 2025

Commits on Sep 30, 2025

Moved permissions section for id-token outside the aws-setup job.
akashveramd
committed

Commits on Oct 6, 2025

Using move_aws_steps_inside_setup_rocm branch to do aws authentication in linux_job_v2.yml.
akashveramd
committed

Commits on Oct 7, 2025

Commits on Oct 8, 2025

Removed integration_test_8gpu.yaml and added ROCm workflow to run features tests inside integration_test_8gpu_features.yaml. Using linux_job_v2.yml from the main branch. Rolled back to using 8 GPU …
akashveramd
committed

Commits on Oct 9, 2025

Commits on Oct 11, 2025

Switched back to main branch for linux_job_v2.yml.
akashveramd
committed

Commits on Oct 16, 2025

Commits on Oct 18, 2025

Commits on Oct 21, 2025

DEBUG:checking disk space on ROCm runner before container is created.
akashveramd
committed