wjones127 commented on code in PR #15101:
URL: https://github.com/apache/arrow/pull/15101#discussion_r1060902137
##########
cpp/src/parquet/arrow/reader_writer_benchmark.cc:
##########
@@ -197,6 +197,52 @@ BENCHMARK_TEMPLATE2(BM_WriteColumn, true, DoubleType);
BENCHMARK_TEMPLATE2(BM_WriteColumn, false, BooleanType);
BENCHMARK_TEMPLATE2(BM_WriteColumn, true, BooleanType);
+int32_t kInfiniteUniqueValues = -1;
+
+std::shared_ptr<::arrow::Table> RandomStringTable(int64_t length, int64_t
unique_values,
+ int64_t null_percentage) {
+ std::shared_ptr<::arrow::DataType> type = ::arrow::utf8();
+ std::shared_ptr<::arrow::Array> arr;
+ ::arrow::random::RandomArrayGenerator generator(500);
+ double null_probability = static_cast<double>(null_percentage) / 100.0;
+ if (unique_values == kInfiniteUniqueValues) {
+ arr = generator.String(length, /*min_length=*/3, /*max_length=*/32,
+ /*null_probability=*/null_probability);
+ } else {
+ arr = generator.StringWithRepeats(length, /*unique=*/unique_values,
+ /*min_length=*/3, /*max_length=*/32,
+ /*null_probability=*/null_probability);
Review Comment:
I was thinking it would depending on encoding (dictionary vs PLAIN), and
IIUC we automatically switch from RLE to plain when we hit a certain threshold
of unique values. Though perhaps it's sufficient to just benchmark 32 unique
values and infinite, rather than several different cardinalities. I will also
make sure to enforce the encoding.
--
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]
