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

Fixing parquet coalescing of reads #12808

Merged
merged 2 commits into from
Feb 22, 2023
Merged

Fixing parquet coalescing of reads #12808

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
bf55bd4
Moving read out a layer so it is only called once so coalescing into …
hyperbolic2346 Feb 21, 2023
b408354
Merge branch 'branch-23.04' into mwilson/parquet_reader_single_read
hyperbolic2346 Feb 22, 2023
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
22 changes: 11 additions & 11 deletions cpp/src/io/parquet/reader_impl_preprocess.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -685,7 +685,6 @@ void reader::impl::load_and_decompress_data(std::vector<row_group_info> const& r
auto const row_group_start = rg.start_row;
auto const row_group_source = rg.source_index;
auto const row_group_rows = std::min<int>(remaining_rows, row_group.num_rows);
auto const io_chunk_idx = chunks.size();

// generate ColumnChunkDesc objects for everything to be decoded (all input columns)
for (size_t i = 0; i < num_input_columns; ++i) {
Expand Down Expand Up @@ -733,18 +732,19 @@ void reader::impl::load_and_decompress_data(std::vector<row_group_info> const& r
total_decompressed_size += col_meta.total_uncompressed_size;
}
}
// Read compressed chunk data to device memory
read_rowgroup_tasks.push_back(read_column_chunks_async(_sources,
raw_page_data,
chunks,
io_chunk_idx,
chunks.size(),
column_chunk_offsets,
chunk_source_map,
_stream));

remaining_rows -= row_group.num_rows;
}

// Read compressed chunk data to device memory
read_rowgroup_tasks.push_back(read_column_chunks_async(_sources,
raw_page_data,
chunks,
0,
ttnghia marked this conversation as resolved.
Show resolved Hide resolved
chunks.size(),
column_chunk_offsets,
chunk_source_map,
_stream));

for (auto& task : read_rowgroup_tasks) {
task.wait();
}
Expand Down