Cool. Are there APL implementations which take good advantage of GPUs or TPUs etc for optimizing matrix & vector operations?