[GLUTEN-3594] [VL] Allow users to set bloom filter configurations #3610
Conversation
|
Run Gluten Clickhouse CI |
1 similar comment
|
Run Gluten Clickhouse CI |
| @@ -1267,4 +1272,12 @@ object GlutenConfig { | |||
| + "partial aggregation may be early abandoned.") | |||
| .intConf | |||
| .createOptional | |||
|
|
|||
| val COLUMNAR_VELOX_BLOOM_FILTER_MAX_NUM_BITS = | |||
| buildConf("spark.gluten.sql.columnar.backend.velox.bloomFilter.maxNumBits") | |||
There was a problem hiding this comment.
Add a new config as spark.bloom_filter.max_num_bits has different behavior from spark conf RUNTIME_BLOOM_FILTER_MAX_NUM_BITS.
https://github.com/facebookincubator/velox/blob/73d4279a14744cf4d038d3a967a49dcddbad9d39/velox/core/QueryConfig.h#L632C5-L632C24
| @@ -62,6 +62,9 @@ const std::string kAbandonPartialAggregationMinPct = | |||
| "spark.gluten.sql.columnar.backend.velox.abandonPartialAggregationMinPct"; | |||
| const std::string kAbandonPartialAggregationMinRows = | |||
| "spark.gluten.sql.columnar.backend.velox.abandonPartialAggregationMinRows"; | |||
| const std::string kBloomFilterExpectedNumItems = "spark.sql.optimizer.runtime.bloomFilter.expectedNumItems"; | |||
| const std::string kBloomFilterNumBits = "spark.sql.optimizer.runtime.bloomFilter.numBits"; | |||
There was a problem hiding this comment.
How about incorporating the two parameters mentioned above with the "spark.gluten" prefix? For instance, we could use "spark.gluten.sql.optimizer.runtime.bloomFilter.expectedNumItems" and "spark.gluten.sql.optimizer.runtime.bloomFilter.numBits".
There was a problem hiding this comment.
Ok to me, let me add new config entries for them.
|
Run Gluten Clickhouse CI |
1 similar comment
|
Run Gluten Clickhouse CI |
| getConfigValue(confMap_, kBloomFilterExpectedNumItems, "1000000"); | ||
| configs[velox::core::QueryConfig::kSparkBloomFilterNumBits] = | ||
| getConfigValue(confMap_, kBloomFilterNumBits, "8388608"); | ||
| configs[velox::core::QueryConfig::kSparkBloomFilterMaxNumBits] = |
There was a problem hiding this comment.
It seems the default value of kSparkBloomFilterMaxNumBits in vanilla spark is 67108864L here . Why here is 4194304?
There was a problem hiding this comment.
This is because velox has below requirement: https://github.com/facebookincubator/velox/blob/73d4279a14744cf4d038d3a967a49dcddbad9d39/velox/core/QueryConfig.h#L632C5-L632C24
|
Please also update the document, https://github.com/oap-project/gluten/blob/main/docs/Configuration.md |
Added, thanks. |
|
Run Gluten Clickhouse CI |
|
@zhli1142015 Can you help to rebase again? Thanks. |
zhli1142015
force-pushed
the
bloom_filter_config
branch
from
8eae37e
to
ee2b7d6
Compare
|
Run Gluten Clickhouse CI |
Rebased, thanks. |
|
===== Performance report for TPCH SF2000 with Velox backend, for reference only ====
|
What changes were proposed in this pull request?
Fixes: #3594
How was this patch tested?
(Please explain how this patch was tested. E.g. unit tests, integration tests, manual tests)
(If this patch involves UI changes, please attach a screenshot; otherwise, remove this)