3 ms·
Small addtion: matrix multiplications (and other operations implemented through BLAS) do use the M1's AMX matrix co-processor (through the Apple Accelerate fram
by microtonal 5y ago
Small addtion: matrix multiplications (and other operations implemented through BLAS) do use the M1's AMX matrix co-processor (through the Apple Accelerate framework).