FPGA implementation of large-scale matrix inversion using single, double and custom floating-point precision | IEEE Conference Publication | IEEE Xplore