Digital Event Horizon
NVIDIA Accelerates Generative AI Research at UC San Diego with DGX B200 System
The University of California San Diego's Hao AI Lab has recently received an NVIDIA DGX B200 system, a powerful artificial intelligence computing platform that will accelerate the lab's research in large language model inference. With this powerful computing platform, the lab is poised to make significant advancements in LLM inference, low-latency LLM serving, and disaggregated inference.
The University of California San Diego's Hao AI Lab has received an NVIDIA DGX B200 system to accelerate its research in large language model inference. The system is one of the most powerful AI platforms from NVIDIA, enabling faster prototyping and experimentation than previous-generation hardware. The lab is utilizing the DGX B200 for two key research projects: FastVideo and Lmgame benchmark. The Hao AI Lab is exploring new ways to achieve low-latency LLM serving using disaggregated inference, which separates prefill and decode jobs onto different GPUs.
The University of California San Diego's Hao AI Lab has recently received an NVIDIA DGX B200 system, a powerful artificial intelligence computing platform that will accelerate the lab's research in large language model inference. This move marks an exciting development in the field of generative AI, with the potential to significantly enhance the performance and efficiency of LLMs.
According to Dr. Hao Zhang, assistant professor in the Halıcıoğlu Data Science Institute and department of computer science and engineering at UC San Diego, the DGX B200 system is one of the most powerful AI systems from NVIDIA to date. "DGX B200 is one of the most powerful AI systems from NVIDIA to date, which means that its performance is among the best in the world," Dr. Zhang stated. "It enables us to prototype and experiment much faster than using previous-generation hardware."
The Hao AI Lab has already begun utilizing the DGX B200 system for two key research projects: FastVideo and Lmgame benchmark. FastVideo focuses on training a family of video generation models to produce a five-second video based on a given text prompt — in just five seconds. The research phase of FastVideo taps into NVIDIA H200 GPUs in addition to the DGX B200 system.
Lmgame, on the other hand, is a benchmarking suite that puts LLMs to the test using popular online games including Tetris and Super Mario Bros. Users can test one model at a time or put two models up against each other to measure their performance. The illustrated workflow of Hao AI Lab’s Lmgame-Bench project shows how the researchers are using the DGX B200 system to accelerate this benchmarking process.
Another ongoing project at Hao AI Labs explores new ways to achieve low-latency LLM serving, pushing large language models toward real-time responsiveness. This research is being conducted using the disaggregated inference method, which separates the prefill and decode jobs onto different GPUs to maximize goodput.
"Previously, if you put these two jobs on a GPU, they would compete with each other for resources, which could make it slow from a user perspective," Dr. Chen stated. "Now, if I split the jobs onto two different sets of GPUs — one doing prefill, which is compute intensive, and the other doing decode, which is more memory intensive — we can fundamentally eliminate the interference between the two jobs, making both jobs run faster."
This process is called prefill/decode disaggregation, or separating the prefill from decode to get greater goodput. Increasing goodput and using the disaggregated inference method enables the continuous scaling of workloads without compromising on low-latency or high-quality model responses.
NVIDIA Dynamo — an open-source framework designed to accelerate and scale generative AI models at the highest efficiency levels with the lowest cost — enables scaling disaggregated inference. In addition to these projects, cross-departmental collaborations, such as in healthcare and biology, are underway at UC San Diego to further optimize an array of research projects using the NVIDIA DGX B200, as researchers continue exploring how AI platforms can accelerate innovation.
In conclusion, the recent acquisition of the NVIDIA DGX B200 system by the Hao AI Lab marks a significant development in the field of generative AI. With this powerful computing platform, the lab is poised to make significant advancements in LLM inference, low-latency LLM serving, and disaggregated inference. As researchers continue to explore the potential of AI platforms in accelerating innovation, it will be exciting to see how these developments unfold.
Related Information:
https://www.digitaleventhorizon.com/articles/NVIDIA-Accelerates-Generative-AI-Research-at-UC-San-Diego-with-DGX-B200-System-deh.shtml
https://blogs.nvidia.com/blog/ucsd-generative-ai-research-dgx-b200/
Published: Wed Dec 17 11:37:23 2025 by llama3.2 3B Q4_K_M
