Struct SimdOps

pub struct SimdOps;

Safe SIMD-accelerated operations for performance-critical paths Uses the wide crate for portable, safe SIMD operations

This struct provides vectorized math operations for numerical processing without any unsafe code. It automatically uses SIMD when beneficial and falls back to scalar operations for small arrays.

Performance considerations

• Uses safe wide f64x4 operations for parallel processing
• Portable across all platforms (not limited to x86_64)
• NaN-safe operations with hardware NaN propagation
• Avoids heap allocations for small to medium-sized arrays
• Handles edge cases like empty arrays and single-element arrays efficiently

Implementations

impl SimdOps

pub fn sum_f64(values: &[f64]) -> f64

Apply safe SIMD-accelerated sum to an array of f64 values Uses wide f64x4 operations for maximum performance and safety Optimized with different strategies based on array size

pub fn decimal_to_f64(values: &[Decimal]) -> Vec<f64>

Convert an array of decimal values to f64 for SIMD operations

pub fn sum_decimal(values: &[Decimal]) -> Decimal

Sum an array of Decimal values using safe SIMD acceleration

pub fn sum_price_level_sizes<T>( levels: &[T], extract_size: impl Fn(&T) -> Decimal, ) -> Decimal

Sum the size values from an array of PriceLevel structs using SIMD acceleration This avoids creating a temporary Vec for the size values

pub fn min_f64(values: &[f64]) -> f64

Apply safe SIMD-accelerated min to an array of f64 values Uses wide f64x4 operations with NaN-safe comparisons

pub fn max_f64(values: &[f64]) -> f64

Apply safe SIMD-accelerated max to an array of f64 values Uses wide f64x4 operations with NaN-safe comparisons

pub fn min_decimal(values: &[Decimal]) -> Decimal

Calculate the min value of an array of Decimal values using SIMD acceleration

pub fn max_decimal(values: &[Decimal]) -> Decimal

Calculate the max value of an array of Decimal values using SIMD acceleration

pub fn dot_product_f64(a: &[f64], b: &[f64]) -> f64

Calculate the dot product of two f64 arrays using safe SIMD acceleration

pub fn mean_f64(values: &[f64]) -> f64

Calculate the mean (average) of an array of f64 values using SIMD acceleration

pub fn mean_decimal(values: &[Decimal]) -> Decimal

Calculate the mean (average) of an array of Decimal values using SIMD acceleration

pub fn variance_f64(values: &[f64]) -> f64

Calculate the variance of an array of f64 values using SIMD acceleration with Welford’s algorithm This is the population variance (divides by n, not n-1) - more numerically stable than naive method

pub fn std_dev_f64(values: &[f64]) -> f64

Calculate the standard deviation of an array of f64 values using SIMD acceleration This is the population standard deviation (divides by n, not n-1)

pub fn variance_decimal(values: &[Decimal]) -> Decimal

Calculate the variance of an array of Decimal values using SIMD acceleration This is the population variance (divides by n, not n-1)

pub fn std_dev_decimal(values: &[Decimal]) -> Decimal

Calculate the standard deviation of an array of Decimal values using SIMD acceleration This is the population standard deviation (divides by n, not n-1)