Use AVX2 for bitmap intersection stores

[perfloop-agent]

Description

Use AVX2 to materialize bitmap-container intersections on supported amd64 CPUs.

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Performance improvement
• Code refactoring
• Documentation update
• Test improvements
• Build/CI changes

Changes Made

What was changed?

• Added an AVX2 four-word AND/store helper for bitmap results.
• Added focused dense, sparse, and disjoint intersection benchmarks and dispatch tests.

Why was it changed?

• Bitmap results count the intersection first, then materialize every bitmap word one at a time.

How was it changed?

• Reuse the existing runtime AVX2 check and keep scalar generic and appengine fallbacks.
• Keep the existing array-result path for low-cardinality intersections.

Testing

• go test ./...
• go test -tags appengine .

Checks: 2 passed.

Formatting

Ran gofmt -s -l on changed Go files.

Fuzzing

• go test -tags=gofuzz -run=TestGenerateSmatCorpus
• The documented FuzzSmat command completed without a fuzz target in this checkout.

Performance Impact

The vector store applies only when the existing cardinality decision keeps a bitmap result. Unsupported CPUs and low-cardinality results keep their existing paths.

Workload: roaring_and_bitmap_dense_intersection

Metric	Before	After	Change
ns/op	3897	2926	24.9% lower

Workload: roaring_and_bitmap_sparse_intersection

Metric	Before	After	Change
ns/op	3880	2933	24.4% lower

Breaking Changes

None.

Related Issues

None.

Additional Notes

None.

---

Generated by Perfloop. Measurements and checks.