You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@mahout.apache.org by "Howardisme (via GitHub)" <gi...@apache.org> on 2026/03/26 10:24:34 UTC
[PR] Fix/gpu f32 alignment [mahout]
Howardisme opened a new pull request, #1219:
URL: https://github.com/apache/mahout/pull/1219
### Related Issues
Closes #1218
### Changes
- [x] Bug fix
- [ ] New feature
- [ ] Refactoring
- [ ] Documentation
- [ ] Test
- [ ] CI/CD pipeline
- [ ] Other
### Why
To fix the alignment problem related to amplitude_encode_batch_kernel_f32 when sample size is odd.
### How
By adding a per-sample float2 alignment check in amplitude_encode_batch_kernel_f32, then using:
- vectorized float2 loads on aligned sample bases
- scalar __ldg loads on unaligned sample bases
This follows the same aligned-fast-path + scalar-fallback pattern used in l2_norm_batch_kernel_f32, keeps output semantics unchanged, and prevents CUDA_ERROR_MISALIGNED_ADDRESS when sample_size is odd and num_samples > 1.
## Checklist
- [ ] Added or updated unit tests for all changes
- [ ] Added or updated documentation for all changes
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@mahout.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Fix/gpu f32 alignment [mahout]
Posted by "ryankert01 (via GitHub)" <gi...@apache.org>.
ryankert01 commented on PR #1219:
URL: https://github.com/apache/mahout/pull/1219#issuecomment-4133601814
please help resolve conflicts
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@mahout.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Fix/gpu f32 alignment [mahout]
Posted by "Howardisme (via GitHub)" <gi...@apache.org>.
Howardisme commented on PR #1219:
URL: https://github.com/apache/mahout/pull/1219#issuecomment-4134385332
sure, conflicts solved!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@mahout.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Fix/gpu f32 alignment [mahout]
Posted by "Howardisme (via GitHub)" <gi...@apache.org>.
Howardisme commented on PR #1219:
URL: https://github.com/apache/mahout/pull/1219#issuecomment-4134594680
sorry, seems like there is some problem in my local version. The bug is already fixed. I'll close this pr. sorry again ;(
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@mahout.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [PR] Fix/gpu f32 alignment [mahout]
Posted by "Howardisme (via GitHub)" <gi...@apache.org>.
Howardisme closed pull request #1219: Fix/gpu f32 alignment
URL: https://github.com/apache/mahout/pull/1219
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: commits-unsubscribe@mahout.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org