Expand description
SIMD utility functions for execution engine
This module provides fallback implementations for SIMD operations used in the execution engine for high-performance metrics calculations.
Functionsยง
- abs_vec
- Element-wise absolute value
- add_vec
- Add two vectors element-wise
- cosine_
similarity - Compute cosine similarity between two vectors
- divide_
vec - Divide two vectors element-wise
- divide_
vec_ inplace - In-place division of target vector by divisor vector
- dot_
product - Compute dot product of two vectors
- euclidean_
distance - Compute Euclidean distance between two vectors
- exp_vec
- Element-wise exponential
- l1_norm
- Compute L1 norm (Manhattan norm) of a vector
- l2_norm
- Compute L2 norm (Euclidean norm) of a vector
- log_vec
- Element-wise natural logarithm
- mean_
vec - Compute mean of a vector of f32 values
- min_
max_ vec - Find minimum and maximum values in a vector
- multiply_
vec - Multiply two vectors element-wise
- multiply_
vec_ inplace - In-place multiplication of target vector by multiplier vector
- normalize_
vec - Normalize a vector to unit length
- scale_
vec - Scale a vector by a scalar value
- sqrt_
vec - Element-wise square root
- sum_vec
- Compute sum of a vector of f32 values
- variance_
vec - Compute variance of a vector of f32 values