Rich-T-kid opened a new pull request, #27340: URL: https://github.com/apache/datafusion/pull/27340
## Which issue does this PR close? - Follow-up to the dictionary benchmarks in #27252. ## Rationale for this change High-fanout hash joins currently resolve and compare dictionary string values for every collision candidate. Translating each side once per probe batch lets the hot loop compare dense integer IDs instead. ## What changes are included in this PR? Adds a cached translation path for single-column string and binary dictionary joins when candidate fanout is high enough to amortize setup. HJ Q28 covers independently encoded dictionaries. ## What is the testing strategy for this PR? Local HJ Q28 improved from 148.6 ms to 107.4 ms (-27.8%). Plain-string Q23 remained within noise at roughly 135-136 ms. Hosted TPC-H, TPC-DS, and partitioned ClickBench results are pending. ## Are there any user-facing changes? No. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected] --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
