Reading https://www.science.org/doi/10.1126/science.adh1174 https://www.science.org/doi/10.1126/science.adh1174, that’s “a 22 times lower time metric of latency

Reading https://www.science.org/doi/10.1126/science.adh1174 https://www.science.org/doi/10.1126/science.adh1174, that’s “a 22 times lower time metric of latency […] relative to a graphics processing unit (GPU) that uses a comparable 12-nanometer technology process”