Skip to content

Conversation

@jinsolp
Copy link
Contributor

@jinsolp jinsolp commented May 30, 2025

@jinsolp jinsolp requested a review from a team as a code owner May 30, 2025 17:12
@jinsolp jinsolp requested review from divyegala and viclafargue May 30, 2025 17:12
@jinsolp jinsolp changed the base branch from branch-25.08 to branch-25.06 May 30, 2025 17:12
@jinsolp jinsolp added ! - Hotfix Hotfix is a bug that affects the majority of users for which there is no reasonable workaround non-breaking Non-breaking change improvement Improvement / enhancement to an existing function labels May 30, 2025
@csadorf csadorf added bug Something isn't working and removed improvement Improvement / enhancement to an existing function labels May 30, 2025
RAFT_CUDA_TRY(cudaPointerGetAttributes(&attr, inputsA.X));
float* ptr = reinterpret_cast<float*>(attr.devicePointer);
bool data_on_device = ptr != nullptr;
float* ptr = reinterpret_cast<float*>(attr.hostPointer);
Copy link
Member

Choose a reason for hiding this comment

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

Do pointers allocated with new and malloc also have a valid attr.hostPointer?

@csadorf
Copy link
Contributor

csadorf commented May 30, 2025

Looks like the patch does not work on CUDA 11.

Copy link
Member

@divyegala divyegala left a comment

Choose a reason for hiding this comment

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

Here's @robertmaynard's relevant PR, and why I believe this fix is correct: rapidsai/raft#2480

@jinsolp
Copy link
Contributor Author

jinsolp commented May 30, 2025

Checked that this fix works on the GH system too

@raydouglass raydouglass merged commit 64c7d31 into rapidsai:branch-25.06 May 30, 2025
91 of 93 checks passed
@jinsolp jinsolp deleted the umap-batch-hotfix branch May 31, 2025 05:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

! - Hotfix Hotfix is a bug that affects the majority of users for which there is no reasonable workaround bug Something isn't working CUDA/C++ non-breaking Non-breaking change

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants