Adding kernel to add parse_uri support for protocol extraction #1502
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
Signed-off-by: Mike Wilson <[email protected]>
hyperbolic2346
force-pushed
the
mwilson/parse-url-protocol
branch
from
19a69ff to
04f94c4
Compare
|
build |
hyperbolic2346
commented
Oct 13, 2023
jlowe
reviewed
Oct 13, 2023
3 tasks
|
build |
Signed-off-by: Mike Wilson <[email protected]>
JNI bindings have been added. |
|
build |
Signed-off-by: Mike Wilson <[email protected]>
|
build |
jlowe
reviewed
Oct 16, 2023
ttnghia
reviewed
Oct 17, 2023
ttnghia
reviewed
Oct 17, 2023
Signed-off-by: Mike Wilson <[email protected]>
Signed-off-by: Mike Wilson <[email protected]>
Signed-off-by: Mike Wilson <[email protected]>
Signed-off-by: Mike Wilson <[email protected]>
|
build |
jlowe
reviewed
Oct 17, 2023
Co-authored-by: Jason Lowe <[email protected]>
Co-authored-by: Jason Lowe <[email protected]>
ttnghia
previously approved these changes
Oct 17, 2023
Signed-off-by: Mike Wilson <[email protected]>
|
build |
jlowe
approved these changes
Oct 17, 2023
rapids-bot bot
pushed a commit
to rapidsai/cudf
that referenced
this pull request
Oct 24, 2023
Ran across this issue during the review of NVIDIA/spark-rapids-jni#1502 and since the code was modeled after this code, I am pushing the fix here as well. Authors: - Mike Wilson (https://github.com/hyperbolic2346) Approvers: - David Wendt (https://github.com/davidwendt) - Nghia Truong (https://github.com/ttnghia) URL: #14290
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Outdated suggestions cannot be applied.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.
This adds support for parsing protocols from a uri on the GPU. This change models the kernel after
url_decode, which dedicates a warp to a string. This is a better approach than a thread per byte due to removing all thelower_boundcalls that hit global memory to figure out which row a byte belongs. This kernel copies the string from global memory into shared memory a block at a time for processing.There is a potential optimization here by having the first kernel build a starting offset for each string. With that and the number of bytes the second kernel could turn into a series of
cudaMemcpyAsynccalls. This protocol kernel even starts at byte 0, so the starting offset is known. I don't know how this will fare as support for other extractions like HOST and PATH may add further complications. I am not sure if this will be a monolithic kernel or if it will be best to have a kernel per parse type yet.This will evolve over time, but I don't know which direction yet so I have not done this optimization.
closes #1501