Skip to content

feat(cudf): Restore Spark xxhash64 and runtime Bloom filters #102

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
sperlingxx wants to merge 2 commits into
base: dev
Choose a base branch
from
Open

feat(cudf): Restore Spark xxhash64 and runtime Bloom filters #102

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
48615d8
feat(cudf): support Spark xxhash64 and runtime bloom filters
sperlingxx Aug 24, 2026
668f0bc
fix(cudf): reject multi-column Spark xxhash64
sperlingxx Aug 25, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
14 changes: 13 additions & 1 deletion velox/experimental/cudf/exec/CudfAggregation.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -206,6 +206,12 @@ std::vector<ResolvedAggregateInfo> resolveAggregateInfos(
const auto isDecimalAggregate = aggregate.rawInputTypes.size() == 1 &&
aggregate.rawInputTypes[0]->isDecimal();

std::vector<core::TypedExprPtr> extraInputs;
if (aggregate.call->inputs().size() > 1) {
extraInputs.assign(
aggregate.call->inputs().begin() + 1, aggregate.call->inputs().end());
}

params.emplace_back(
companionStep,
aggregate.call->name(),
Expand All @@ -215,7 +221,8 @@ std::vector<ResolvedAggregateInfo> resolveAggregateInfos(
isCountFunctionName(aggregate.call->name())
? std::make_optional(getCountInputKind(aggregate, constants[i]))
: std::nullopt,
isDecimalAggregate);
isDecimalAggregate,
std::move(extraInputs));
}
return params;
}
Expand Down Expand Up @@ -279,6 +286,11 @@ AggregationInputChannels buildAggregationInputChannels(
} else if (
auto constant =
dynamic_cast<const core::ConstantTypedExpr*>(arg.get())) {
if (!aggInputs.empty()) {
// Extra constant arguments (bloom_filter_agg estimatedNumItems and
// numBits) are not aggregation input columns.
continue;
}
result.constants[i] = constant->toConstantVector(operatorCtx.pool());
aggInputs.push_back(fallbackChannel);
} else {
Expand Down
5 changes: 5 additions & 0 deletions velox/experimental/cudf/exec/CudfAggregation.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,11 +19,13 @@
#include "velox/experimental/cudf/expression/ExpressionEvaluator.h"
#include "velox/experimental/cudf/vector/CudfVector.h"

#include "velox/core/ITypedExpr.h"
#include "velox/exec/Operator.h"
#include "velox/expression/FunctionSignature.h"

#include <optional>
#include <string_view>
#include <vector>

namespace facebook::velox::cudf_velox {

Expand Down Expand Up @@ -67,6 +69,9 @@ struct ResolvedAggregateInfo {
// Routing keys off the function family, not the physical batch type (which is
// VARBINARY/STRING on intermediate and final steps).
bool isDecimalAggregate;
// Extra call arguments after the first aggregation input, e.g. constant
// estimatedNumItems/numBits for bloom_filter_agg.
std::vector<core::TypedExprPtr> extraInputs;
};

// Parse aggregate inputs from the aggregation node and resolve companion steps,
Expand Down
Loading
Loading