Skip to content

Matmul int4 micro-kernels for QSI8D32 (LHS) x QSI4C32 (RHS) -> F32

Gian Marco Iodice requested to merge qsi8d32_qsi4c32_mm into main
  • The LHS matrix is Quantized (Q) Symmetric (S) Signed 8-bit (I8) with per-block quantization (D32) quantization parameters
  • The RHS matrix is Quantized (Q) Symmetric (S) Signed 4-bit (I4) with per-block quantization(C32) F16 scale factors,
  • The destination is F32
  • Implement micro-kernels to perform the matrix multiplication
  • Implement a micro-kernel to pack the LHS and RHS matrices
  • Added unit tests

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 Anitha Raj

Merge request reports