DGEMM Using Tensor Cores, and Its Accurate and Reproducible Versions
Daichi Mukunoki, Katsuhisa Ozaki, Takeshi Ogita, Toshiyuki Imamura
Topics & Concepts
Computer scienceMatrix multiplicationFLOPSDouble-precision floating-point formatFloating pointParallel computingMultiplication (music)Tensor (intrinsic definition)GraphicsMatrix (chemical analysis)Graphics processing unitComputational scienceAlgorithmRange (aeronautics)Computer graphics (images)MathematicsPhysicsPure mathematicsQuantum mechanicsQuantumComposite materialCombinatoricsMaterials scienceParallel Computing and Optimization TechniquesNumerical Methods and AlgorithmsAdvanced Data Storage Technologies