Skip to content

[DO NOT MERGE] Collate all ROCm component include directory for spack build #2001

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.

Sign up for GitHub

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

Jump to bottom
Closed
wants to merge 7 commits into from
Closed

[DO NOT MERGE] Collate all ROCm component include directory for spack build #2001

wants to merge 7 commits into from

Conversation

renjithravindrankannath
Copy link

@renjithravindrankannath renjithravindrankannath commented Mar 26, 2025 •
edited by jithunnair-amd
Loading

For Spack builds, each ROCm component will have a separate include directory, so we cannot assume /opt/rocm/include to be the path where all the ROCm component headers will be located.

This PR uses the ROCM_INCLUDE_DIRS CMake variable to collate all the ROCm component include directories set by find_package calls.

Fixes # SWDEV-524001

@rocm-repo-management-api
Copy link

rocm-repo-management-api bot commented Mar 26, 2025 •
edited
Loading

Jenkins build for 7901d8c998ed20f5f0358f351a8a80c28ad93a3b commit finished as FAILURE
Links: Blue Ocean view / Build artifacts

Detected error during base docker image building:

#32 12.38 The following packages have unmet dependencies:
#32 12.45  rocm-dev : Depends: rocm-cmake (= 0.14.0.60304-76~22.04) but 5.0.0-1 is to be installed
#32 12.45             Depends: rocm-device-libs (= 1.0.0.60304-76~22.04) but 5.0.0-1 is to be installed
#32 12.45  rocm-utils : Depends: rocminfo (= 1.0.0.60304-76~22.04) but 5.0.0-1 is to be installed
#32 12.45               Depends: rocm-cmake (= 0.14.0.60304-76~22.04) but 5.0.0-1 is to be installed
#32 12.46 E: Unable to correct problems, you have held broken packages.
#32 ERROR: process "/bin/sh -c bash ./install_rocm.sh" did not complete successfully: exit code: 100
------
 > [stage-0 24/50] RUN bash ./install_rocm.sh:
12.38 distribution that some required packages have not yet been created
12.38 or been moved out of Incoming.

@renjithravindrankannath renjithravindrankannath mentioned this pull request Mar 26, 2025
Merged
@jithunnair-amd
Copy link
Collaborator

@renjithravindrankannath This PR should target one of the release/ or internal_testing branches, not main. But I am not in favor of these changes anyway, because it seems to bring back spaghetti code. Can we explore other ways?

@rocm-repo-management-api
Copy link

rocm-repo-management-api bot commented Mar 28, 2025 •
edited
Loading

Jenkins build for 7901d8c998ed20f5f0358f351a8a80c28ad93a3b commit finished as FAILURE
Links: Blue Ocean view / Build artifacts

Detected error during base docker image building:

#31 15.03 The following packages have unmet dependencies:
#31 15.11  rocm-dev : Depends: rocm-cmake (= 0.14.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 15.11             Depends: rocm-device-libs (= 1.0.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 15.11  rocm-utils : Depends: rocminfo (= 1.0.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 15.11               Depends: rocm-cmake (= 0.14.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 15.11 E: Unable to correct problems, you have held broken packages.
#31 ERROR: process "/bin/sh -c bash ./install_rocm.sh" did not complete successfully: exit code: 100
------
 > [stage-0 23/61] RUN bash ./install_rocm.sh:
15.03 distribution that some required packages have not yet been created
15.03 or been moved out of Incoming.

@rocm-repo-management-api
Copy link

rocm-repo-management-api bot commented Apr 14, 2025 •
edited
Loading

Jenkins build for 7901d8c998ed20f5f0358f351a8a80c28ad93a3b commit finished as FAILURE
Links: Blue Ocean view / Build artifacts

@rocm-repo-management-api
Copy link

rocm-repo-management-api bot commented Apr 22, 2025 •
edited
Loading

Jenkins build for e16801444b11115bdaca3255d79884a619a3ba60 commit finished as FAILURE
Links: Blue Ocean view / Build artifacts

Detected error during base docker image building:

#31 11.29 The following packages have unmet dependencies:
#31 11.36  rocm-dev : Depends: rocm-cmake (= 0.14.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 11.36             Depends: rocm-device-libs (= 1.0.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 11.36  rocm-utils : Depends: rocminfo (= 1.0.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 11.36               Depends: rocm-cmake (= 0.14.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 11.36 E: Unable to correct problems, you have held broken packages.
#31 ERROR: process "/bin/sh -c bash ./install_rocm.sh" did not complete successfully: exit code: 100
------
 > [stage-0 23/61] RUN bash ./install_rocm.sh:
11.29 distribution that some required packages have not yet been created
11.29 or been moved out of Incoming.

@jithunnair-amd jithunnair-amd changed the title rocthrust, rocprim, rocrand, hipcub, ck and aotriton include paths Collate all ROCm component include directory for spack build Apr 25, 2025
@jithunnair-amd jithunnair-amd requested review from jeffdaily and removed request for jeffdaily April 25, 2025 15:59
@rocm-repo-management-api
Copy link

rocm-repo-management-api bot commented Apr 25, 2025 •
edited
Loading

Jenkins build for d0a90e857829cb0b27f5118b9411bb1748de00af commit finished as FAILURE
Links: Blue Ocean view / Build artifacts

@jithunnair-amd jithunnair-amd changed the title Collate all ROCm component include directory for spack build [DO NOT MERGE] Collate all ROCm component include directory for spack build Apr 25, 2025
@jithunnair-amd jithunnair-amd marked this pull request as draft April 25, 2025 21:16
@rocm-repo-management-api
Copy link

rocm-repo-management-api bot commented Apr 25, 2025 •
edited
Loading

Jenkins build for d0a90e857829cb0b27f5118b9411bb1748de00af commit finished as FAILURE
Links: Blue Ocean view / Build artifacts

@rocm-repo-management-api
Copy link

rocm-repo-management-api bot commented Apr 27, 2025 •
edited
Loading

Jenkins build for 301a1788e8b7251480423801264b474dca98fedb commit finished as FAILURE
Links: Blue Ocean view / Build artifacts

Detected error during base docker image building:

#31 16.30 The following packages have unmet dependencies:
#31 16.38  rocm-dev : Depends: rocm-cmake (= 0.14.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 16.38             Depends: rocm-device-libs (= 1.0.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 16.38  rocm-utils : Depends: rocminfo (= 1.0.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 16.38               Depends: rocm-cmake (= 0.14.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 16.38 E: Unable to correct problems, you have held broken packages.
#31 ERROR: process "/bin/sh -c bash ./install_rocm.sh" did not complete successfully: exit code: 100
------
 > [stage-0 23/61] RUN bash ./install_rocm.sh:
16.30 distribution that some required packages have not yet been created
16.30 or been moved out of Incoming.

@jithunnair-amd jithunnair-amd requested a review from m-gallus April 28, 2025 12:31
@jithunnair-amd
Copy link
Collaborator

Added @m-gallus as reviewer so he can help us ensure we don't break Windows builds of PyTorch on ROCm.

m-gallus
m-gallus reviewed Apr 28, 2025
View reviewed changes
Copy link

@m-gallus m-gallus left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I confirm that the internal Windows build is still passing with those changes.

@rocm-repo-management-api
Copy link

rocm-repo-management-api bot commented Apr 28, 2025 •
edited
Loading

Jenkins build for 03800325e96e64fb01c32b95ead19bf7b71fee74 commit finished as FAILURE
Links: Blue Ocean view / Build artifacts

m-gallus
m-gallus approved these changes Apr 30, 2025
View reviewed changes
Copy link

@m-gallus m-gallus left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Looks good to me, Windows build passes.

@rocm-repo-management-api
Copy link

rocm-repo-management-api bot commented Apr 30, 2025 •
edited
Loading

Jenkins build for 2ea898acd3f1133873653cc5ff8a095832769786 commit finished as FAILURE
Links: Blue Ocean view / Build artifacts

Detected error during base docker image building:

#31 11.14 The following packages have unmet dependencies:
#31 11.21  rocm-dev : Depends: rocm-cmake (= 0.14.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 11.21             Depends: rocm-device-libs (= 1.0.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 11.21  rocm-utils : Depends: rocminfo (= 1.0.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 11.21               Depends: rocm-cmake (= 0.14.0.60304-76~22.04) but 5.0.0-1 is to be installed
#31 11.21 E: Unable to correct problems, you have held broken packages.
#31 ERROR: process "/bin/sh -c bash ./install_rocm.sh" did not complete successfully: exit code: 100
------
 > [stage-0 23/61] RUN bash ./install_rocm.sh:
11.14 distribution that some required packages have not yet been created
11.14 or been moved out of Incoming.

@jithunnair-amd jithunnair-amd mentioned this pull request May 1, 2025
Closed
@jithunnair-amd
Copy link
Collaborator

Closing this in favor of pytorch#152569

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@jithunnair-amd jithunnair-amd jithunnair-amd left review comments

@m-gallus m-gallus m-gallus approved these changes

Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
3 participants
@renjithravindrankannath @jithunnair-amd @m-gallus