Enable prefetching in cudf.pandas.install() #16439
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
bdice
added
bug
bdice
commented
Jul 30, 2024
Comment on lines
+48
to
+51
| if (data_ptr == nullptr) { | ||
| // Do not call chars_size if the data_ptr is nullptr. | ||
| return; | ||
| } |
There was a problem hiding this comment.
This was a bug. We can't call chars_size on string column views that are not fully initialized like this one:
cudf/cpp/src/column/column_factories.cu
Lines 60 to 62 in 8def2ec
Errors from calling chars_size() looked like this:
terminate called after throwing an instance of 'cudf::logic_error'
what(): CUDF failure at: /home/coder/cudf/cpp/src/strings/strings_column_view.cpp:34: strings column has no children
Aborted (core dumped)
bdice
commented
Jul 30, 2024
Comment on lines
+54
to
+67
| // Don't try to prefetch nullptrs or empty data. Sometimes libcudf has column | ||
| // views that use nullptrs with a nonzero size as an optimization. | ||
| if (ptr == nullptr) { | ||
| if (prefetch_config::instance().debug) { | ||
| std::cerr << "Skipping prefetch of nullptr" << std::endl; | ||
| } | ||
| return cudaSuccess; | ||
| } | ||
| if (size == 0) { | ||
| if (prefetch_config::instance().debug) { | ||
| std::cerr << "Skipping prefetch of size 0" << std::endl; | ||
| } | ||
| return cudaSuccess; | ||
| } |
There was a problem hiding this comment.
I'm not really sure if this is necessary or not, but it seems like it would improve our library safety to avoid calling the cudaMemPrefetchAsync API unless we know we have a non-null pointer and non-zero size.
bdice
requested review from
isVoid,
charlesbluca,
davidwendt and
pmattione-nvidia
and removed request for
a team
galipremsagar
approved these changes
Jul 30, 2024
davidwendt
approved these changes
Jul 30, 2024
vyasr
approved these changes
Jul 30, 2024
|
Verified that this works in local testing. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Description
This PR enables
cudf.pandasmanaged memory prefetching incudf.pandas.install(), to ensure that prefetching is enabled for all methods of enablingcudf.pandas.I also fixed a bug in libcudf's prefetching logic, where it tried to compute the number of characters in a strings column view even if the string column view's data is
nullptr. This errors, so we must avoid thechars_size()call and stop the prefetch attempt early.Checklist