Fix the internal compiler error with gcc 12/13 and CUDA 12.4/12.5 #2282
Conversation
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
|
|
Author
|
Is there any interest to get this merged? I need to patch newer releases with Grace Hopper support. |
Member
|
I am curious: Does it work if, instead of removing |
Author
|
I can try it next week. |
Member
|
Thank you. If that does not work, let me know because I have another suggestion in mind. |
Author
|
@garrison I changed the first two ones and now I get |
garrison
reviewed
Nov 25, 2025
Comment on lines
-481
to
483
|
|
||
| const uint_t mask_u = ~MASKS[x_max + 1]; | ||
| const uint_t mask_l = MASKS[x_max]; | ||
| uint_t mask_u = ~MASKS[x_max + 1]; | ||
| uint_t mask_l = MASKS[x_max]; | ||
| auto lambda = [&](const int_t i, double &val_re, double &val_im) -> void { |
Member
There was a problem hiding this comment.
I want to see what happens if these three lines look like
const uint_t mask_u = ~MASKS[x_max + 1];
const uint_t mask_l = MASKS[x_max];
auto lambda = [&, mask_u, mask_l](const int_t i, double &val_re, double &val_im) -> void {
i.e., the only change from the original is to add , mask_u, mask_l to the capture so that those are captured by value instead of reference. And likewise for the other places with the proposed change.
Comment on lines
-243
to
244
|
|
||
| const size_t N = qubits.size(); | ||
| size_t N = qubits.size(); | ||
| auto func = [&](const areg_t<2> &inds, |
Member
There was a problem hiding this comment.
Here would be
const size_t N = qubits.size();
auto func = [&, N](const areg_t<2> &inds,
Member
@diehlpk, thank you for trying that. I added a new suggestion above. |
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.
Summary
Compiling qiskit-aer with CUDA support and gcc 12/13 resulted in some internal compiler error, see #2227.
The error was resolved by removing the
constexpression in from of the lambda functions. The issue is that the aboveconstdeclared variables are passed to the lambda function using&as the default capture of the lambda function. Passing a constant value by reference to a lambda function causes issues by gcc 12/13.Details and comments
Removing the
constexpression allows the code to compile with gcc 12/13 using CUDA 12.4/12.5. I have not yet checked if the issue us specific to Grace Grace and Grace Hopper which is ARM.