Skip to content
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

Support spilling to disk directly via cuFile/GDS #1313

Merged
merged 6 commits into from
Dec 17, 2020

Conversation

rongou
Copy link
Collaborator

@rongou rongou commented Dec 8, 2020

This adds an option to spill to disk directly via GDS. Default is off and shouldn't have any effect on existing behavior.

Initial test on my local desktop shows that it performs similarly to spilling to host memory, with a single Samsung 970 EVO Plus 2TB NVMe M.2 drive. Need to do more benchmarking in real deployment environments.

Depends on rapidsai/cudf#6940.

Signed-off-by: Rong Ou <rong.ou@gmail.com>
@rongou rongou added performance A performance related task/issue shuffle things that impact the shuffle plugin labels Dec 8, 2020
Copy link
Collaborator

@revans2 revans2 left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

My only comment on this is that either we need to hide the config, or we need to document it better and explain it is an experimental feature and point to more information about GDS if we want others to try it out.

@rongou
Copy link
Collaborator Author

rongou commented Dec 9, 2020

@revans2 added wording to the config to specify it's an experimental feature, and link to gds docs.

revans2
revans2 previously approved these changes Dec 9, 2020
@jlowe
Copy link
Member

jlowe commented Dec 10, 2020

build

@rongou
Copy link
Collaborator Author

rongou commented Dec 11, 2020

build

@pxLi
Copy link
Collaborator

pxLi commented Dec 15, 2020

build

1 similar comment
@pxLi
Copy link
Collaborator

pxLi commented Dec 15, 2020

build

@rongou rongou requested a review from abellina December 15, 2020 04:19
@rongou rongou added the feature request New feature or request label Dec 16, 2020
@abellina abellina merged commit 2e0fe6a into NVIDIA:branch-0.4 Dec 17, 2020
@sameerz sameerz added this to the Dec 7 - Dec 18 milestone Dec 26, 2020
@rongou rongou mentioned this pull request Jan 5, 2021
11 tasks
@rongou rongou deleted the gds-store branch May 11, 2021 17:50
nartal1 pushed a commit to nartal1/spark-rapids that referenced this pull request Jun 9, 2021
* support spilling to GDS

Signed-off-by: Rong Ou <rong.ou@gmail.com>

* address review comments

* review feedback

Signed-off-by: Rong Ou <rong.ou@gmail.com>
nartal1 pushed a commit to nartal1/spark-rapids that referenced this pull request Jun 9, 2021
* support spilling to GDS

Signed-off-by: Rong Ou <rong.ou@gmail.com>

* address review comments

* review feedback

Signed-off-by: Rong Ou <rong.ou@gmail.com>
tgravescs pushed a commit to tgravescs/spark-rapids that referenced this pull request Nov 30, 2023
…p ci] [bot] (NVIDIA#1313)

* Update submodule cudf to 9c559c94fcaa1525d7c95faf94e5486fcd992ef1

Signed-off-by: spark-rapids automation <70000568+nvauto@users.noreply.github.com>

* Update submodule cudf to b7994bc16b1b1743b0743860b4f02ac4da8245d5

Signed-off-by: spark-rapids automation <70000568+nvauto@users.noreply.github.com>

* Update submodule cudf to 15cc5011902f1026e04662e725b880f48d38ba8d

Signed-off-by: spark-rapids automation <70000568+nvauto@users.noreply.github.com>

---------

Signed-off-by: spark-rapids automation <70000568+nvauto@users.noreply.github.com>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
feature request New feature or request performance A performance related task/issue shuffle things that impact the shuffle plugin
Projects
None yet
Development

Successfully merging this pull request may close these issues.

6 participants