-
Notifications
You must be signed in to change notification settings - Fork 310
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Update calls to cuDF Series ctors, bug fix to cugraph.subgraph()
for handling non-renumbered Graphs
#1901
Update calls to cuDF Series ctors, bug fix to cugraph.subgraph()
for handling non-renumbered Graphs
#1901
Conversation
…that has not been renumbered is handled correctly by cugraph.subgraph()
Codecov Report
@@ Coverage Diff @@
## branch-21.12 #1901 +/- ##
===============================================
Coverage ? 70.13%
===============================================
Files ? 143
Lines ? 8812
Branches ? 0
===============================================
Hits ? 6180
Misses ? 2632
Partials ? 0 Continue to review full report at Codecov.
|
…ted from Buffers. Also cleaned up various cython imports and some minor test cleanup.
…ore cleanup to cython imports (still testing).
cugraph.subgraph()
cugraph.subgraph()
for handling non-renumbered Graphs
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This solves the problem and hopefully will unblock CI. One question added. I'm OK either way... although it might be better if we resolve the question and make the changes if the answer is it would be better with the change.
from libc.stdint cimport uintptr_t | ||
|
||
|
||
cdef move_device_buffer_to_series(unique_ptr[device_buffer] device_buffer_unique_ptr, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Question... This implementation generates a column and converts it into a Series. In many of the algorithms we generate a Series using this function to create a column in a data frame.
Would it be better to generate a column and leave it as a column? Perhaps we should have a move_device_buffer_to_series
and a move_device_buffer_to_column
method?
Or am I reading too much into this?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Thanks @ChuckHastings , I like that idea and I'm testing out your suggestion now. In the interest of unblocking our other PRs by fixing the CI failures here, I'll ask @BradReesWork to merge this now and then I'll create a follow-up PR that implements your suggestion.
@gpucibot merge |
…eries objects (#1915) This is a follow-up PR to address [this feedback](#1901 (comment)) from #1901 This adds a separate cython helper for moving a buffer to a column to allow for a slightly more efficient call where column objects can be used. Authors: - Rick Ratzel (https://github.com/rlratzel) Approvers: - Brad Rees (https://github.com/BradReesWork) - Chuck Hastings (https://github.com/ChuckHastings) - Joseph Nke (https://github.com/jnke2016) URL: #1915
Series
objects are constructed fromBuffer
types.ci/gpu/build.sh
to add the conda envbin
dir toPATH
, since it was not finding the correctcmake
during a local run.cugraph.subgraph()
.pytest.ini
for users that don't have the required extra pytest plugins installed, left comments in place to for users to re-enable.closes #1908
closes #1899
Note: this PR was originally intended to be two PRs: one for the
subgraph
bug fix, and another for thecudf.Series
updates. I accidentally pushed the Series updates to the wrong branch, hence this combined PR. I can separate them if reviewing is too difficult.