Skip to content

[MINOR] Use cudf::packed_size to estimate the memory usage of a table.#967

Merged
rapids-bot[bot] merged 9 commits intorapidsai:mainfrom
nirandaperera:minor_cudf_packedsize
Apr 14, 2026
Merged

[MINOR] Use cudf::packed_size to estimate the memory usage of a table.#967
rapids-bot[bot] merged 9 commits intorapidsai:mainfrom
nirandaperera:minor_cudf_packedsize

Conversation

@nirandaperera
Copy link
Copy Markdown
Contributor

@nirandaperera nirandaperera commented Apr 13, 2026

This change enables estimated_memory_usage to return exact packed size. Then we wouldn't need to pad the reservations.

Closes #847

Signed-off-by: niranda perera <niranda.perera@gmail.com>
@nirandaperera nirandaperera requested a review from a team as a code owner April 13, 2026 23:38
@nirandaperera nirandaperera added improvement Improves an existing functionality non-breaking Introduces a non-breaking change labels Apr 13, 2026
Signed-off-by: niranda perera <niranda.perera@gmail.com>
Signed-off-by: niranda perera <niranda.perera@gmail.com>
Comment thread cpp/src/integrations/cudf/utils.cpp Outdated
Signed-off-by: niranda perera <niranda.perera@gmail.com>
…dsmpf into minor_cudf_packedsize

Signed-off-by: niranda perera <niranda.perera@gmail.com>
@nirandaperera
Copy link
Copy Markdown
Contributor Author

@madsbk I ended up keeping estimated_memory_usage and introduced a new util that inlines cudf::packed_size.
The reason was, we use estimated_memory_usage before reserving memory for a cudf table. I think it would be better representative than packed_size for tables. WDYT?

Signed-off-by: niranda perera <niranda.perera@gmail.com>
@nirandaperera
Copy link
Copy Markdown
Contributor Author

/merge

@rapids-bot rapids-bot bot merged commit 5ea50bb into rapidsai:main Apr 14, 2026
66 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

improvement Improves an existing functionality non-breaking Introduces a non-breaking change

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Use cudf::packed_size() for reserving memory for packing

2 participants