.Ted Hisokawa.Aug 23, 2024 14:53.NVIDIA designers to introduce improvements in the Blackwell platform, fluid cooling, and AI-driven potato chip design at Hot Chips 2024.
At the upcoming Hot Chips 2024 seminar, NVIDIA engineers are actually readied to found groundbreaking advancements aimed at improving information facility performance and power performance. The celebration, booked for August 25-27 at Stanford University, will certainly include four talks through elderly NVIDIA developers, depending on to NVIDIA Blog.NVIDIA Blackwell Platform.NVIDIA's presentations are going to highlight the NVIDIA Blackwell platform, which includes multiple potato chips, devices, as well as NVIDIA CUDA software program to drive the future generation of artificial intelligence apps. The NVIDIA GB200 NVL72, a multi-node, liquid-cooled service, will definitely also be showcased for its capability to hook up 72 Blackwell GPUs and also 36 Kindness CPUs, setting a new standard for AI system style. Also, the NVLink interconnect innovation is going to be actually reviewed for its own high throughput and low-latency capabilities, important for generative AI.Fluid Cooling Advancements.A notable portion of the event will definitely concentrate on liquid air conditioning technologies. Ali Heydari, director of records center air conditioning and facilities at NVIDIA, will definitely provide layouts for hybrid-cooled records facilities. These designs target to retrofit existing air-cooled data facilities with liquid-cooling systems, delivering a much more reliable and space-saving remedy. Heydari's staff is also collaborating along with the U.S. Department of Power on the COOLERCHIPS program to build sophisticated cooling innovations. Using the NVIDIA Omniverse platform, they are actually producing digital doubles to design energy usage and also cooling down efficiency.AI-Driven Chip Concept.NVIDIA is additionally leveraging AI to improve semiconductor concept. Smudge Ren, director of concept computerization study at NVIDIA, will certainly go over how AI styles boost layout top quality and efficiency by automating taxing activities. These designs feature prophecy as well as optimization resources that aid engineers study and improve styles rapidly. AI substances powered by large language versions (LLMs) are actually being built to autonomously full activities, communicate with designers, as well as learn from a data bank of human as well as agent experiences. Ren is going to share examples of artificial intelligence functions in time document evaluation, tissue cluster optimization, and also code generation.These presentations highlight NVIDIA's commitment to pushing the borders of information center computer with cutting-edge modern technologies in AI and cooling down solutions. The business's efforts aim to deliver exceptional performance, efficiency, and also optimization all over various sectors.Image source: Shutterstock.