Intel's Gaudi2 Technology Surpasses NVIDIA in Language Model Inference
站长之家
67
Intel's Gaudi2 technology excels in large-scale language model inference, rivaling NVIDIA's AI accelerators. Gaudi2 matches the decoding performance of NVIDIA's H100 system and even surpasses NVIDIA's A100. According to public cloud pricing, Gaudi2 also offers better cost-effectiveness for both training and inference compared to NVIDIA's A100 and H100. Intel plans to launch Gaudi3 technology in 2024, which will provide enhanced processing power and network bandwidth, further boosting performance.
© Copyright AIbase Base 2024, Click to View Source - https://www.aibase.com/news/4703