美国芯片巨头英特尔近日宣布了一项重大突破,成功研发出全球最大的神经拟态计算系统——Hala Point。该系统旨在推动未来类脑人工智能的研究,以应对AI在效率和可持续性方面的挑战。Hala Point搭载了1152个基于Intel 4(7nm)工艺的Loihi 2处理器,具备11.5亿个神经元和1280亿个突触,每秒可执行超过380万亿个8位突触操作和240万亿个神经元运算,展现了强大的计算能力。
据钛媒体报道,当应用于仿生尖峰神经网络模型时,Hala Point在处理相同数量神经元任务时,其速度竟然可以达到人脑的200倍,即20000%的增益。这一成就为人工智能的未来开辟了新的可能,尤其是在需要高效能和低能耗的复杂计算场景下。
Hala Point的首次部署选在了美国桑迪亚国家实验室,这里将为该系统的潜力提供一个理想的实验平台。英特尔的这一创新不仅标志着神经拟态计算领域的重大进步,也为人工智能的快速发展和应用提供了强大的硬件支持,预示着未来AI技术将更加接近人脑的智能水平。
英语如下:
**News Title:** “Intel Unveils World’s Largest Neuromorphic System, Hala Point, Outperforming Human Brain by 200 Times, Paving the Way for AI Revolution”
**Keywords:** Intel, Hala Point, Neuromorphic System
**News Content:**
Title: Intel Launches Hala Point, the World’s Largest Neuromorphic System, with 200 Times the Speed of Human Brain
American chip giant Intel recently announced a major breakthrough with the development of the globe’s largest neuromorphic computing system, known as Hala Point. Aimed at advancing brain-inspired AI research to address efficiency and sustainability challenges in AI, this system is equipped with 1152 Intel 4 (7nm) process-based Loihi 2 processors, featuring 11.5 billion neurons and 128 billion synapses. It can perform over 380 trillion 8-bit synaptic operations and 240 trillion neuron computations per second, demonstrating its immense computational prowess.
As reported by TMT News, when applied to bio-inspired spiking neural network models, Hala Point can process tasks involving the same number of neurons at a staggering 200 times the speed of the human brain, representing a 20,000% increase. This achievement opens up new possibilities for the future of AI, particularly in scenarios that require high performance and low power consumption for complex computations.
The initial deployment of Hala Point is set to take place at the Sandia National Laboratories in the United States, which will provide an ideal testing ground for the system’s potential. Intel’s innovation not only signifies a significant leap in neuromorphic computing but also offers robust hardware support for the rapid advancement and application of AI, indicating that future AI technologies will more closely resemble the cognitive capabilities of the human brain.
【来源】https://www.tmtpost.com/7044204.html
Views: 1