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.

Sign up for GitHub

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

Extend the Parquet writer's dictionary encoding benchmark. #16591

Merged
merged 14 commits into from
Sep 10, 2024
Merged

Extend the Parquet writer's dictionary encoding benchmark. #16591

Changes from 8 commits
Commits
Show all changes
14 commits
Select commit Hold shift + click to select a range
c7b13fd
Initial commit for benchmark for PQ writer dict encoding
mhaseeb123 Aug 15, 2024
f255fba
Updates
mhaseeb123 Aug 17, 2024
340b88b
Merge branch 'branch-24.10' into pq-writer-dict-benchmark
mhaseeb123 Aug 19, 2024
26c3a89
Merge branch 'branch-24.10' of https://github.com/mhaseeb123/cudf int…
mhaseeb123 Aug 19, 2024
c17a4b9
Minor updates
mhaseeb123 Aug 19, 2024
1535fee
Update run length
mhaseeb123 Aug 19, 2024
a29fb5f
Updates
mhaseeb123 Aug 19, 2024
0cc2bd6
Revert erroneous white space change
mhaseeb123 Aug 19, 2024
d8f984d
Merge branch 'branch-24.10' into pq-writer-dict-benchmark
mhaseeb123 Aug 19, 2024
b00850d
Code review updates
mhaseeb123 Sep 6, 2024
507ae3d
Merge branch 'branch-24.10' into pq-writer-dict-benchmark
mhaseeb123 Sep 6, 2024
fef3179
Add 100K cardinality
mhaseeb123 Sep 7, 2024
bfe3b13
Updates from code review
mhaseeb123 Sep 9, 2024
97b766f
Merge branch 'branch-24.10' into pq-writer-dict-benchmark
mhaseeb123 Sep 9, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions cpp/benchmarks/io/parquet/parquet_writer.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -202,8 +202,8 @@ NVBENCH_BENCH_TYPES(BM_parq_write_encode, NVBENCH_TYPE_AXES(d_type_list))
.set_name("parquet_write_encode")
.set_type_axes_names({"data_type"})
.set_min_samples(4)
.add_int64_axis("cardinality", {0, 1000})
.add_int64_axis("run_length", {1, 32});
.add_int64_axis("cardinality", {1, 1000, 10'000, 100'000, 1'000'000})
Copy link
Member Author

Choose a reason for hiding this comment

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

Are there any resource/infrastructure constraints that we need to consider here? We can perhaps remove one or two entries from here if needed.

Copy link
Contributor

Choose a reason for hiding this comment

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

I think we can drop 1. Given the table size, it's effectively the same as 1000.
Also, should 1M just be 0? Not sure if the goal is to have unique elements (AFAIK row count is typically lower than 1M).

.add_int64_axis("run_length", {1, 32, 64});
Copy link
Contributor

Choose a reason for hiding this comment

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

Sorry, missed this the first time around. To me, 32 is already a very high run length. I think we should instead add 4 or 8. I'm fine with 64 if you do see a significant difference in file size and/or performance compared to 32.

Copy link
Member Author

Choose a reason for hiding this comment

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

No significant differences with 64 for the benchmarks we ran either.

Copy link
Contributor

Choose a reason for hiding this comment

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

I vote for 4 or 8, then

Copy link
Member Author

Choose a reason for hiding this comment

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

All done!


NVBENCH_BENCH_TYPES(BM_parq_write_io_compression, NVBENCH_TYPE_AXES(io_list, compression_list))
.set_name("parquet_write_io_compression")
Expand Down
Loading