Intel has announced a significant performance acceleration for the protein language model ProtST on its Gaudi 2 AI accelerators. The breakthrough enables researchers to analyze protein sequences more efficiently, potentially speeding up drug discovery and bioinformatics research.
'This optimization showcases the power of Gaudi 2 for scientific computing,' said an Intel spokesperson.
The ProtST model, which is based on protein sequence data, benefits from the Gaudi 2's hardware architecture and software optimizations. Performance gains were achieved through techniques such as mixed-precision training and hardware-aware optimizations.
Intel's efforts aim to provide cost-effective alternatives to GPUs for AI workloads, particularly in specialized domains like genomics.