Tensor Cores are mixed-precision matrix multiplication units, that act
as co-processors in the latest generation of GPUs. This special-purpose
hardware was designed to speed up training and inference in deep
learning, but it can be used for other purposes. In this talk, I will
show that tensor cores are a truly disruptive technology for
signal-processing tasks like correlations and beam forming. I will
present the Tensor-Core Correlator, that is 5–20 faster and more energy
efficient than our legacy correlator that runs on regular GPU cores.