Occupancy improvement for Hash table build #15700
Conversation
|
I think the approach of specializing the type dispatcher is very cumbersome and will lead to a lot of code replication. Currently, I have the conditional dispatch working for |
PointKernel
added
non-breaking
|
/ok to test |
|
/ok to test |
|
@tgujar I've updated the docs to unblock CI. Have you noticed any performance regressions for other use cases? It seems that it improves the performance for mixed join but the performance drops significantly in other cases using row hasher. |
| id_to_type<type_id::DECIMAL128>, | ||
| id_to_type<type_id::DECIMAL64>, | ||
| id_to_type<type_id::DECIMAL32>, |
There was a problem hiding this comment.
I don't think decimal types are complex type. They are just a wrapper around some integer type.
There was a problem hiding this comment.
Equality operator for Decimal will perform scaling which uses exponentiation.
I see a reduction in register usage if I comment out decimal types in #15502. I think we can still decide on the types excluded in the branches later on
|
/ok to test |
|
@tgujar Could you take a look at the failing tests? |
|
/ok to test |
|
/ok to test |
| * @throw cudf::logic_error if the input tables were preprocessed to transform any nested children | ||
| * columns into integer columns but `PhysicalElementComparator` is not | ||
| * @throw cudf::logic_error if the input tables were preprocessed to transform any nested | ||
| * children columns into integer columns but `PhysicalElementComparator` is not |
There was a problem hiding this comment.
It appears that a significant number of changes to this file are due to reformatting comments.
Would it be possible to undo those changes? This particular change is certainly not desirable.
There was a problem hiding this comment.
Makes sense, will fix. This was caused because of the clang-format extension on vscode
|
This PR needs to be rebased on branch-24.08. |
|
Specializing both the comparator and the hasher drops the register usage to 54 instead of the expected 46 for the mixed semi join case. Investigating why the register pressure is different from commenting out the code paths. |
Description
Prototype implementation for: #15502
Checklist