Module vectorized_scan 

SIMD-Accelerated Vectorized Scan Engine (Recommendation 2)

Problem

Current scan implementation is row-at-a-time:

for entry in self.data.iter() {  // DashMap iteration - pointer chasing
    if let Some(v) = entry.value().read_at(snapshot_ts, current_txn_id)
        && let Some(value) = &v.value  // Option unwrapping
    {
        results.push((key.clone(), value.clone())); // Heap allocs
    }
}

This has:

• DashMap iterator overhead (~20ns per entry)
• MVCC version chain traversal per row
• Clone allocations in hot path

Solution

Vectorized execution: process 1024+ rows in a batch, amortizing overhead.

Performance Analysis

Vectorized execution model:

• Batch size B = 1024 rows
• Per-batch overhead: O(1) iterator setup + O(B) SIMD operations
• Amortized cost: (setup + B×simd_op) / B ≈ simd_op for large B

SIMD comparison (AVX-512):

Scalar: 100 cycles × 1000 rows = 100,000 cycles
SIMD: 100 cycles × (1000/16) = 6,250 cycles (16x speedup)

Memory bandwidth calculation:

• Current: 300 bytes/row × 1M rows = 300 MB, random access
• Proposed: 60 bytes/row × 1M rows = 60 MB, sequential scan
• RAM bandwidth: ~50 GB/s → theoretical 833M rows/sec

Expected Improvement

10-20x scan throughput improvement

Structs

Int64Comparison

Comparison predicate for i64 columns

SimdVisibilityFilter

SIMD-accelerated MVCC visibility filter

SoaBatch

Structure of Arrays (SoA) batch for optimal SIMD visibility filtering

SoaScanIterator

High-performance SoA scan iterator with SIMD + late materialization

SoaScanStats

Statistics for SoA scan performance monitoring

StreamingScanIterator

Streaming scan iterator with SIMD visibility filtering

VectorBatch

A batch of rows for vectorized processing

VectorizedScanConfig

Configuration for vectorized scans

VectorizedScanStats

Statistics for vectorized scan operations

VersionedSlice

Versioned slice for zero-copy access

Enums

ColumnVector

Typed column vector for SIMD-friendly access

ComparisonOp

Comparison operators

ValueHandle

Handle for late value materialization

Constants

DEFAULT_BATCH_SIZE

Default batch size for vectorized operations 1024 rows fits comfortably in L2 cache (~256KB with 256-byte rows)

MAX_BATCH_SIZE

Maximum batch size (above this, memory pressure increases)

MIN_BATCH_SIZE

Minimum batch size (below this, scalar is faster)

Traits

SoaSource

Trait for sources that provide SoA batches

VectorPredicate

Predicate for vectorized filtering

Functions

simd_compare_i64_gt