viiccwen opened a new pull request, #1116:
URL: https://github.com/apache/mahout/pull/1116
### Related Issues
closes #1113
### Changes
- [ ] Bug fix
- [x] New feature
- [ ] Refactoring
- [ ] Documentation
- [x] Test
- [ ] CI/CD pipeline
- [ ] Other
### Why
Python CUDA tensors with shape `[batch_size, features]` and dtype
`torch.float32` now use the existing core batch float32 GPU-pointer path
### How
#### Python bindings
- Extended `QdpEngine::_encode_from_cuda_tensor` to support 2D CUDA float32
amplitude inputs
- Wired the Python CUDA path to `encode_batch_from_gpu_ptr_f32_with_stream`
- Kept stream interop consistent with the existing single-sample float32
CUDA path
- Preserved the existing float64 / basis / other method dispatch behavior
#### Tests
- Replaced the old Python test that expected 2D CUDA float32 amplitude to
fail
- Added coverage for successful 2D CUDA float32 amplitude encoding
- Added coverage for engine precision behavior:
- default `float32` engine returns `complex64`
- `float64` engine still returns `complex128`
## 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: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
