Skip to content

Matmul int4 micro-kernels for QAI8DX (LHS) x QSI4C32 (RHS) -> F32

Gian Marco Iodice requested to merge qsi4c32_matmul into main
  • The LHS matrix is Quantized (Q) Asymmetric (A) Signed 8-bit (I8) with per-row (DX) quantization parameters
  • The RHS matrix is quantized (Q) Symmetric (S) Signed 4-bit (I4) with per-block quantization
  • The destination is F32
  • Implement micro-kernels to perform the matrix multiplication
  • Implement a micro-kernel to pack the RHS matrix

Signed-off-by: Gian Marco Iodice gianmarco.iodice@arm.com

Signed-off-by: Anitha Raj anitha.raj@arm.com

Signed-off-by: Viet-Hoa Do viet-hoa.do@arm.com

Edited by Jakub Sujak

Merge request reports