This white paper analyzes how Intel hardware optimizations improve the performance of IBM Watson NLP workloads on CPUs. While GPUs traditionally deliver faster AI processing, they are costly and in short supply. Intel Xeon Scalable processors, equipped with Deep Learning Boost and optimized libraries like oneDNN, enhance CPU efficiency for AI workloads. Experiments comparing base and Intel-optimized builds of Watson NLP showed significant gains, with up to 35% higher function throughput across tasks like entity extraction and sentiment analysis. These improvements enable faster, more scalable NLP applications without additional strain on CPU or memory resources.

Join for free to read