据热点科技报道，人工智能领域的创新公司Stability AI近日发布了一份详尽报告，揭示了英特尔Gaudi 2计算卡在AI图像生成训练方面的显著优势。在对比测试中，Gaudi 2芯片在特定的模型配置下，表现出了比英伟达同类产品更为出色的效率。
报告指出，以一个拥有2亿参数的MMDiT模型为例，当训练深度设定为24，并采用BFloat16的混合精度计算时，Gaudi 2计算卡的峰值性能达到了每秒训练1254张图片。在256容量的配置下，这一速度降至每秒927张，仍远超竞争对手。相比之下，英伟达的H100-80GB计算卡在相同条件下只能实现每秒595张的训练速度，而A100-80GB芯片则为381张。
这意味着，英特尔Gaudi 2在相同内存容量下，相较于英伟达H100，其训练速度提升了55%，展现出强大的计算性能。尤为值得一提的是，Gaudi 2的性能是英伟达A100的2.43倍，这在AI加速计算领域无疑是一次重大突破。
这一结果显示，英特尔的Gaudi 2芯片正逐步在AI训练速度上确立领先地位，对于需要高效能计算的AI开发者和企业来说，可能提供了一个颇具吸引力的新选择。Stability AI的报告无疑为业界提供了一个全新的视角，以审视当前AI硬件的性能对比，也将可能影响未来AI基础设施的选型趋势。
英语如下：
**News Title:** “Intel’s Gaudi 2 Chip Challenges NVIDIA: 55% Faster AI Image Generation Training”
**Keywords:** Intel Gaudi 2, AI Image Generation, Performance Boost
**News Content:**
**Title:** Intel’s Gaudi 2 Chip Outperforms NVIDIA H100 in AI Image Training, Stability AI Reports
According to Hot Tech News, innovative AI company Stability AI recently released a comprehensive report highlighting the exceptional performance of Intel’s Gaudi 2 accelerator card in AI image generation training. In comparative tests, the Gaudi 2 chip demonstrated superior efficiency compared to NVIDIA’s competing offerings.
The report reveals that with a 200 million parameter MMDiT model, configured with a depth of 24 and using BFloat16 mixed-precision computations, the Gaudi 2 reached a peak performance of 1,254 images trained per second. At a 256 capacity setting, this dropped to 927 images per second, still significantly outpacing rivals. In contrast, NVIDIA’s H100-80GB card managed 595 images per second under the same conditions, while the A100-80GB chip achieved 381.
This translates to a 55% faster training speed for Intel’s Gaudi 2 compared to NVIDIA’s H100 at equivalent memory capacity, showcasing its robust computational prowess. Notably, the Gaudi 2’s performance is 2.43 times that of NVIDIA’s A100, marking a major breakthrough in AI accelerated computing.
These findings suggest that Intel’s Gaudi 2 is positioning itself as a leader in AI training speed, offering an attractive new option for AI developers and businesses requiring high-performance computing. Stability AI’s report provides a fresh perspective on the comparative performance of current AI hardware and could influence future trends in AI infrastructure selection.
【来源】https://www.itheat.com/view/45945.html
Views: 1