Hamamu: Specializing FPGAs for ML Applications by Adding Hard Matrix Multiplier Blocks | IEEE Conference Publication | IEEE Xplore