andishgar commented on PR #47586: URL: https://github.com/apache/arrow/pull/47586#issuecomment-3411289319
@rok I removed all the code related to the conversion from a column-major `Tensor` to a `SparseCOOTensor` and instead reused the logic from `StridedTensor` for this conversion. My reasons are: 1. It uses `std::vector` for large allocations instead of `arrow::AllocateBuffer`. 2. It doesn’t use the result of sorting, which wastes a lot of CPU cycles. I know that the current approach may reduce the benefits of CPU cache usage. However, since the existing logic is problematic, I think it’s better to use the `StridedTensor` logic as a temporary solution until a better approach is implemented. Given all this, are you okay with such a change? -- 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]
