Repository navigation
perf: preserve dictionary encoding in replace - #27449
Closed
Rich-T-kid wants to merge 1 commit into
Closed
Rich-T-kid wants to merge 1 commit into
Rich-T-kid wants to merge 1 commit into
Conversation
Contributor
Author
|
run benchmark tpch tpcds clickbench_partitioned |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing codex/dict-replace-suite-gate (039b1a0) to 73e0b03 (merge-base) diff Run configurationrun benchmark tpch
env:
DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"Results will be posted here when complete File an issue against this benchmark runner |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing codex/dict-replace-suite-gate (039b1a0) to 73e0b03 (merge-base) diff Run configurationrun benchmark tpcds
env:
DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"Results will be posted here when complete File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Instance: Comparing codex/dict-replace-suite-gate (039b1a0) to 73e0b03 (merge-base) diff Run configurationrun benchmark tpch
env:
DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"CPU Details (lscpu)Details
Memory Pool PeaksPeak Base:
Pool accounting vs. process RSS Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it.
Resource Usagetpch — base (merge-base)
tpch — branch
File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Instance: Comparing codex/dict-replace-suite-gate (039b1a0) to 73e0b03 (merge-base) diff Run configurationrun benchmark tpcds
env:
DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"CPU Details (lscpu)Details
Memory Pool PeaksPeak Base:
Pool accounting vs. process RSS Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it.
Resource Usagetpcds — base (merge-base)
tpcds — branch
File an issue against this benchmark runner |
|
🤖 Benchmark running (GKE) | trigger CPU Details (lscpu)Comparing codex/dict-replace-suite-gate (039b1a0) to 73e0b03 (merge-base) diff Run configurationrun benchmark clickbench_partitioned
env:
DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"Results will be posted here when complete File an issue against this benchmark runner |
|
🤖 Benchmark completed (GKE) | trigger Instance: Comparing codex/dict-replace-suite-gate (039b1a0) to 73e0b03 (merge-base) diff Run configurationrun benchmark clickbench_partitioned
env:
DATAFUSION_EXECUTION_PARQUET_ENABLE_RLE_TO_DICTIONARY: "true"
DATAFUSION_RUNTIME_MEMORY_LIMIT: "32G"CPU Details (lscpu)Details
Memory Pool PeaksPeak Base:
Pool accounting vs. process RSS Max pool peak is the largest reservation any single query in the run reached; peak RSS covers the whole invocation, including data loading and allocator retention, and the two high-water marks need not coincide in time. The gap is therefore an upper bound on what the pool did not account for, not a measurement of it.
Resource Usageclickbench_partitioned — base (merge-base)
clickbench_partitioned — branch
File an issue against this benchmark runner |
Contributor
Author
|
Hosted benchmark result: this does not provide evidence for the scalar replace optimization. None of the required suites calls scalar replace. ClickBench Q28 calls REGEXP_REPLACE, which is a different UDF and cannot execute this patch:
Because no benchmark query enters the changed path, the isolated classifications cannot be attributed to dictionary-preserving replace. There is no clear evidence of an end-to-end speedup in the required suites, so I am closing the draft. |
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
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.
Which issue does this PR close?\n\n- Part of #20935.\n\n## Rationale for this change\n\nreplace currently materializes dictionary-encoded strings before applying scalar replacements. Low-cardinality inputs therefore repeat the same replacement work for every key occurrence.\n\nThis is a draft performance experiment. It is gated on the existing hosted TPC-H, TPC-DS, and partitioned ClickBench suites and will be closed if they do not show an attributable speedup.\n\n## What changes are included in this PR?\n\n- Preserve the haystack dictionary when the replacement arguments are scalar.\n- Apply replacement to dictionary values and reuse the input keys.\n- Compact oversized retained dictionaries before replacement.\n- Fall back to decoded arrays for array-valued replacement arguments.\n- Add coercion, null, scalar, fallback, and retained-dictionary tests.\n\nNo benchmark files or custom benchmark queries are added or changed.\n\n## What is the testing strategy?\n\n- cargo test -p datafusion-functions string::replace --lib\n- cargo clippy -p datafusion-functions --lib --tests -- -D warnings\n- cargo fmt --all -- --check\n- Full repository pre-push hook\n- Hosted TPC-H, TPC-DS, and partitioned ClickBench runs on the exact PR head\n\n## Are there any user-facing changes?\n\nLogical values are unchanged. Dictionary input with scalar replacement arguments now returns a dictionary array with the same key type.