Nvidia and Emulation for HPC

Nvidia is leaning on emulation to increase the performance of its AI chips in the high-performance computing (HPC) sector. This strategic move comes amid competition where AMD has demonstrated some superiority in recent generations of processors.

Double-precision floating-point computation (FP64) is fundamental for applications ranging from aerospace engineering to scientific research, drug development, and defense. Instead of designing specific chips to natively handle this type of calculation, Nvidia prefers emulation to optimize performance in HPC.

AMD's Reaction

AMD researchers, while recognizing the value of algorithms like the Ozaki scheme, believe that such solutions are not yet mature for widespread use and require further refinement before they can be implemented on a large scale.