simd_quantized_matmul

Function simd_quantized_matmul 

Source
pub fn simd_quantized_matmul(
    a: &QuantizedMatrix,
    a_params: &QuantizationParams,
    b: &QuantizedMatrix,
    b_params: &QuantizationParams,
) -> LinalgResult<Array2<f32>>
Expand description

SIMD-accelerated quantized matrix-matrix multiplication

Performs matrix-matrix multiplication where both matrices are in quantized form. The result is returned as f32.

§Arguments

  • a - First quantized matrix
  • a_params - Quantization parameters for the first matrix
  • b - Second quantized matrix
  • b_params - Quantization parameters for the second matrix

§Returns

  • Result matrix of the multiplication