[opt](hash_join) Merging all build blocks at once will cause performance issue#37471
Merged
mrhhsg merged 1 commit intoapache:masterfrom Jul 10, 2024
Merged
[opt](hash_join) Merging all build blocks at once will cause performance issue#37471mrhhsg merged 1 commit intoapache:masterfrom
mrhhsg merged 1 commit intoapache:masterfrom
Conversation
|
Thank you for your contribution to Apache Doris. Since 2024-03-18, the Document has been moved to doris-website. |
Member
Author
|
run buildall |
![github-actions[bot]](https://avatars.githubusercontent.com/in/15368?s=60&v=4){kind=small}
| @@ -100,30 +100,33 @@ size_t PartitionedHashJoinSinkLocalState::revocable_mem_size(RuntimeState* state | |||
| } | |||
|
|
|||
| Status PartitionedHashJoinSinkLocalState::_revoke_unpartitioned_block(RuntimeState* state) { | |||
Contributor
There was a problem hiding this comment.
warning: function '_revoke_unpartitioned_block' has cognitive complexity of 52 (threshold 50) [readability-function-cognitive-complexity]
Status PartitionedHashJoinSinkLocalState::_revoke_unpartitioned_block(RuntimeState* state) {
^Additional context
be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:108: +1, including nesting penalty of 0, nesting level increased to 1
if (inner_sink_state_) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:113: +1, including nesting penalty of 0, nesting level increased to 1
if (build_block.rows() <= 1) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:119: +1, including nesting penalty of 0, nesting level increased to 1
if (build_block.columns() > num_slots) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:123: nesting level increased to 1
auto spill_func = [build_block = std::move(build_block), state, this]() mutable {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:130: nesting level increased to 2
[](std::vector<uint32_t>& indices) { indices.reserve(reserved_size); });
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:132: nesting level increased to 2
auto flush_rows = [&state, this](std::unique_ptr<vectorized::MutableBlock>& partition_block,
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:137: +3, including nesting penalty of 2, nesting level increased to 3
if (!status.ok()) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:149: +2, including nesting penalty of 1, nesting level increased to 2
while (offset < total_rows) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:153: +3, including nesting penalty of 2, nesting level increased to 3
for (size_t i = 0; i != build_block.columns(); ++i) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:167: +3, including nesting penalty of 2, nesting level increased to 3
for (size_t i = 0; i != sub_block.rows(); ++i) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:171: +3, including nesting penalty of 2, nesting level increased to 3
for (uint32_t partition_idx = 0; partition_idx != p._partition_count; ++partition_idx) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:177: +4, including nesting penalty of 3, nesting level increased to 4
if (UNLIKELY(!partition_block)) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:185: +4, including nesting penalty of 3, nesting level increased to 4
if (!st.ok()) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:195: +4, including nesting penalty of 3, nesting level increased to 4
if (partition_block->rows() >= reserved_size || is_last_block) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:195: +1
if (partition_block->rows() >= reserved_size || is_last_block) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:196: +5, including nesting penalty of 4, nesting level increased to 5
if (!flush_rows(partition_block, spilling_stream)) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:208: nesting level increased to 1
auto exception_catch_func = [spill_func, this]() mutable {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:209: nesting level increased to 2
auto status = [&]() {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:210: +3, including nesting penalty of 2, nesting level increased to 3
RETURN_IF_CATCH_EXCEPTION(spill_func());
^be/src/common/exception.h:89: expanded from macro 'RETURN_IF_CATCH_EXCEPTION'
do { \
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:210: +4, including nesting penalty of 3, nesting level increased to 4
RETURN_IF_CATCH_EXCEPTION(spill_func());
^be/src/common/exception.h:94: expanded from macro 'RETURN_IF_CATCH_EXCEPTION'
} catch (const doris::Exception& e) { \
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:210: +5, including nesting penalty of 4, nesting level increased to 5
RETURN_IF_CATCH_EXCEPTION(spill_func());
^be/src/common/exception.h:95: expanded from macro 'RETURN_IF_CATCH_EXCEPTION'
if (e.code() == doris::ErrorCode::MEM_ALLOC_FAILED) { \
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:214: +2, including nesting penalty of 1, nesting level increased to 2
if (!status.ok()) {
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:228: +1, including nesting penalty of 0, nesting level increased to 1
DBUG_EXECUTE_IF(
^be/src/util/debug_points.h:36: expanded from macro 'DBUG_EXECUTE_IF'
if (UNLIKELY(config::enable_debug_points)) { \
^be/src/pipeline/exec/partitioned_hash_join_sink_operator.cpp:228: +2, including nesting penalty of 1, nesting level increased to 2
DBUG_EXECUTE_IF(
^be/src/util/debug_points.h:38: expanded from macro 'DBUG_EXECUTE_IF'
if (dp) { \
^
TPC-H: Total hot run time: 40248 ms |
TPC-DS: Total hot run time: 174236 ms |
ClickBench: Total hot run time: 30.31 s |
Member
Author
|
run buildall |
TPC-H: Total hot run time: 39759 ms |
TPC-DS: Total hot run time: 173918 ms |
ClickBench: Total hot run time: 29.82 s |
Member
Author
|
run buildall |
TPC-H: Total hot run time: 39702 ms |
TPC-DS: Total hot run time: 173814 ms |
ClickBench: Total hot run time: 30.66 s |
Member
Author
|
run buildall |
TPC-H: Total hot run time: 40510 ms |
TPC-DS: Total hot run time: 174220 ms |
ClickBench: Total hot run time: 30.4 s |
BiteTheDDDDt
approved these changes
Jul 10, 2024
Contributor
|
PR approved by at least one committer and no changes requested. |
github-actions
bot
added
approved
Contributor
|
PR approved by anyone and no changes requested. |
Mryange
approved these changes
Jul 10, 2024
seawinde
pushed a commit
to seawinde/doris
that referenced
this pull request
Jul 17, 2024
…nce issue (apache#37471) ## Proposed changes For better performance, blocks should be merged as soon as possible in the sink operator, allowing parallel execution with the upstream pipeline.
dataroaring
pushed a commit
that referenced
this pull request
Jul 17, 2024
…nce issue (#37471) ## Proposed changes For better performance, blocks should be merged as soon as possible in the sink operator, allowing parallel execution with the upstream pipeline.
Closed
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
5 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.
Proposed changes
For better performance, blocks should be merged as soon as possible in the sink operator, allowing parallel execution with the upstream pipeline.