Fix a bug in tensornet backend scratch pad allocation in multi-GPU mode#2516
Merged
1tnguyen merged 4 commits intoNVIDIA:mainfrom Jan 17, 2025
Merged
Conversation
…r we've set the device Signed-off-by: Thien Nguyen <thiennguyen@nvidia.com>
Signed-off-by: Thien Nguyen <thiennguyen@nvidia.com>
Signed-off-by: Thien Nguyen <thiennguyen@nvidia.com>
|
CUDA Quantum Docs Bot: A preview of the documentation can be found here. |
annagrin
pushed a commit
to annagrin/cuda-quantum
that referenced
this pull request
Jun 17, 2025
…de (NVIDIA#2516) * Fix a bug in default init of scratchpad: it must allocate memory after we've set the device Signed-off-by: Thien Nguyen <thiennguyen@nvidia.com> * Add test Signed-off-by: Thien Nguyen <thiennguyen@nvidia.com> * Add a check to prevent multiple allocate calls Signed-off-by: Thien Nguyen <thiennguyen@nvidia.com> --------- Signed-off-by: Thien Nguyen <thiennguyen@nvidia.com> Signed-off-by: Anna Gringauze <agringauze@nvidia.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Description
ScratchDeviceMemallocates memory based on memory availability on construction. This mechanism is not compatible with multi-GPU code path (MPI execution), whereby the CUDA device is selected in the simulator constructor; hence we need to defer the allocation until the device is selected.Fixed by having a separate
allocatemethod to be called once during the simulator backend constructor after device selection.This bug was introduced in #1865, where the scratch pad is allocated once (scratch pad is a member variable of the simulator class) rather than on-demand to improve performance.
Add a unit test for this case, to be executed when there are multiple GPUs.