Open jonkeane opened 3 years ago
Arrow and dplyr are the same, but arrow and duckdb have slight differences in promo_revenue:
# query 14 scale factor 0.01 Loading required package: DBI [1] 0 ================================================== Query: 14 ================================================== ✔ No differences arrow vs duckdb promo_revenue - arrow[1, ] 16.28156 + duckdb[1, ] 15.48655 `arrow$promo_revenue`: 16.3 `duckdb$promo_revenue`: 15.5 # query 14 scale factor 0.1 jkeane@het arrowbench % ARROWBENCH_LOCAL_DIR="~/repos/ab_store/het" Rscript inst/tpch-answer-gen.R Loading required package: DBI [1] 0 ================================================== Query: 14 ================================================== ✔ No differences arrow vs duckdb promo_revenue - arrow[1, ] 16.67804 + duckdb[1, ] 16.28386 `arrow$promo_revenue`: 16.7 `duckdb$promo_revenue`: 16.3
And these are probably related to the fact that the query is slightly off:
https://github.com/ursacomputing/arrowbench/blob/e8bcce4ed54168623ad0a4205da1187d61f1a9dd/R/tpch-queries.R#L499
This date should be 1995-09-01
Arrow and dplyr are the same, but arrow and duckdb have slight differences in promo_revenue: