-
Notifications
You must be signed in to change notification settings - Fork 5.5k
[native]: Enable Velox cuDF #25094
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
base: master
Are you sure you want to change the base?
[native]: Enable Velox cuDF #25094
Conversation
d13712c
to
7794347
Compare
I'm able to build this PR without issues. |
todo: Add build information to README |
221488b
to
2933adb
Compare
This is now complete. |
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 for the doc! One tiny nit of a suggestion.
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.
LGTM! (docs)
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.
LGTM
Co-authored-by: Zoltan Arnold Nagy <[email protected]> Co-authored-by: Steve Burnett <[email protected]>
@czentgr can you take a look? Thanks. |
add_compile_definitions(PRESTO_ENABLE_CUDF) | ||
enable_language(CUDA) | ||
# Determine CUDA_ARCHITECTURES automatically. | ||
cmake_policy(SET CMP0104 NEW) |
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.
Do we need to do this here if it is enabled also by the subproject (it is in the velox cmake as well when this is enabled).
In some environments, the CUDA_ARCHITECTURES and CUDA_COMPILER location must be explicitly set. | ||
The make command will look like: | ||
|
||
`CUDA_ARCHITECTURES=80 CUDA_COMPILER=/usr/local/cuda/bin/nvcc EXTRA_CMAKE_FLAGS=" -DPRESTO_ENABLE_CUDF=ON" make` |
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.
Just FYI. The VELOX CI uses CUDA_ARCHITECTURES=70
because this is the minimum support architecture.
rm -rf build | ||
|
||
# put CUDA binaries on the PATH | ||
ENV PATH=/usr/local/cuda/bin:${PATH} |
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.
Lets see because in the Velox CI this is installed to /usr/local/cuda-12.8
after running the install_cuda
function. So perhaps the location of the binaries changed since?
Edit: Turns out there are multiple symlinks. /usr/local/cuda
is a symlink to /usr/local/cuda-12.8
.
Turns out we missed to make sure this is built in the pipeline and I created facebookincubator/velox#13334 to fix it.
Description
Motivation and Context
Impact
Test Plan
Contributor checklist
Release Notes
Please follow release notes guidelines and fill in the release notes below.
If release note is NOT required, use: