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

Address poor performance of Parquet string decoding #15304

Merged
merged 4 commits into from
Mar 19, 2024
Merged

Address poor performance of Parquet string decoding #15304

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
0dc15ec
switch between char and string parallel within a page
etseidl Mar 14, 2024
5b58e26
Merge branch 'rapidsai:branch-24.04' into lumpy_strings
etseidl Mar 14, 2024
bd87566
Merge branch 'branch-24.04' into lumpy_strings
etseidl Mar 18, 2024
9e80e84
Merge branch 'branch-24.04' into lumpy_strings
ttnghia Mar 19, 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
13 changes: 5 additions & 8 deletions cpp/src/io/parquet/page_string_decode.cu
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1045,12 +1045,6 @@ CUDF_KERNEL void __launch_bounds__(decode_block_size)
//
if (!has_repetition) { dst_pos -= s->first_row; }

// need to do this before we branch on src_pos/dst_pos so we don't deadlock
// choose a character parallel string copy when the average string is longer than a warp
using cudf::detail::warp_size;
auto const use_char_ll =
s->page.num_valids > 0 && (s->page.str_bytes / s->page.num_valids) >= warp_size;

if (me < warp_size) {
for (int i = 0; i < decode_block_size - out_thread0; i += warp_size) {
dst_pos = sb->nz_idx[rolling_index<rolling_buf_size>(src_pos + i)];
Expand All @@ -1061,10 +1055,13 @@ CUDF_KERNEL void __launch_bounds__(decode_block_size)
: cuda::std::pair<char const*, size_t>{nullptr, 0};

__shared__ cub::WarpScan<size_type>::TempStorage temp_storage;
size_type offset;
cub::WarpScan<size_type>(temp_storage).ExclusiveSum(len, offset);
size_type offset, warp_total;
cub::WarpScan<size_type>(temp_storage).ExclusiveSum(len, offset, warp_total);
offset += last_offset;

// choose a character parallel string copy when the average string is longer than a warp
auto const use_char_ll = warp_total / warp_size >= warp_size;

if (use_char_ll) {
__shared__ __align__(8) uint8_t const* pointers[warp_size];
__shared__ __align__(4) size_type offsets[warp_size];
Expand Down
Loading