Route CUDA pruning io through host memory#8862
Merged
Merged
Conversation
onursatici
force-pushed
the
os/cuda-reads
branch
from
July 20, 2026 15:46
f7216f0 to
323aaaf
Compare
Contributor
Polar Signals Profiling ResultsLatest Run
Previous Runs (3)
Powered by Polar Signals Cloud |
Contributor
Benchmarks: Vortex queries 📖Verdict: No clear signal (low confidence) How to read Verdict and Engines
datafusion / vortex-file-compressed (0.759x ✅, 2↑ 0↓)
datafusion / parquet (0.768x ✅, 2↑ 0↓)
duckdb / vortex-file-compressed (0.772x ✅, 2↑ 0↓)
duckdb / parquet (0.878x ✅, 1↑ 0↓)
No file size changes detected. |
robert3005
approved these changes
Jul 20, 2026
onursatici
force-pushed
the
os/cuda-reads
branch
from
July 21, 2026 10:29
323aaaf to
9946560
Compare
Contributor
Author
|
will merge after #8876 |
Comment on lines
+73
to
+85
| impl Clone for VortexOpenOptions { | ||
| fn clone(&self) -> Self { | ||
| Self { | ||
| session: self.session.clone(), | ||
| segment_cache: self.segment_cache.clone(), | ||
| initial_read_size: self.initial_read_size, | ||
| file_size: self.file_size, | ||
| dtype: self.dtype.clone(), | ||
| footer: self.footer.clone(), | ||
| initial_read_segments: RwLock::new(self.initial_read_segments.read().clone()), | ||
| metrics_registry: self.metrics_registry.clone(), | ||
| labels: self.labels.clone(), | ||
| cache_layout_reader: self.cache_layout_reader, |
Contributor
There was a problem hiding this comment.
can we not default this?
Contributor
Author
There was a problem hiding this comment.
yea I think this is not great, I will file a follow up, I have another idea
joseph-isaacs
approved these changes
Jul 21, 2026
Signed-off-by: Onur Satici <onur@spiraldb.com>
onursatici
force-pushed
the
os/cuda-reads
branch
from
July 21, 2026 12:50
9946560 to
6e5baed
Compare
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Rationale for this change
For zonemap segments we should be using the host memory in cuda reads, but for the data segments we should be using the pinned buffer pool reader that ships segments to the gpu.
If we read every segment with the cuda reader then we would get the zonemap segments shipped to the device by default, and to use them in pruning we have to get them back to the host
What changes are included in this PR?
Cuda reads now use a routing vortex read at implementation that picks zonemap segments and routes them to the host reader, whereas data segments route to the gpu reader