A day in the life of the fastest supercomputer
Frontier, the fastest supercomputer, supports diverse research like climate modeling and astrophysics, executing an exaflop. Access is competitive, aiding in open-source AI model development and enhancing scientific understanding.
Read original article
Frontier, the world's fastest supercomputer located in Oak Ridge, Tennessee, is revolutionizing scientific research by enabling unprecedented simulations across various fields, from climate science to astrophysics. With nearly 50,000 processors and a peak power consumption of 27 megawatts, Frontier can execute an exaflop, or 10^18 floating point operations per second, making it a vital tool for researchers globally. In 2023, it supported 1,744 users from 18 countries, leading to an anticipated 500 published papers in 2024. The supercomputer's architecture, featuring 9,408 nodes with GPUs, allows for complex simulations that capture both large-scale patterns and minute details, such as the interactions of proteins or the evolution of galaxies. Researchers are leveraging Frontier to create high-resolution climate models and to simulate biological processes at atomic levels. However, access to Frontier is competitive, with only about 25% of research proposals being approved. The supercomputer also plays a crucial role in bridging the gap between academic and industry AI research, providing resources for developing open-source large language models. This access is essential for ensuring that academic research can keep pace with industry advancements, particularly in addressing societal needs and biases in AI technologies.
- Frontier is the fastest supercomputer, capable of executing an exaflop.
- It supports diverse research, including climate modeling and astrophysics.
- Access to Frontier is competitive, with a 25% approval rate for proposals.
- The supercomputer aids in developing open-source AI models to counter industry dominance.
- It enables high-resolution simulations that enhance scientific understanding across disciplines.
Related
AI is exhausting the power grid
Tech firms, including Microsoft, face a power crisis due to AI's energy demands straining the grid and increasing emissions. Fusion power exploration aims to combat fossil fuel reliance, but current operations heavily impact the environment.
Tenstorrent Unveils High-End Wormhole AI Processors, Featuring RISC-V
Tenstorrent launches Wormhole AI chips on RISC-V, emphasizing cost-effectiveness and scalability. Wormhole n150 offers 262 TFLOPS, n300 doubles power with 24 GB GDDR6. Priced from $999, undercutting NVIDIA. New workstations from $1,500.
XAI's Memphis Supercluster has gone live, with up to 100,000 Nvidia H100 GPUs
Elon Musk launches xAI's Memphis Supercluster with 100,000 Nvidia H100 GPUs for AI training, aiming for advancements by December. Online status unclear, SemiAnalysis estimates 32,000 GPUs operational. Plans for 150MW data center expansion pending utility agreements. xAI partners with Dell and Supermicro, targeting full operation by fall 2025. Musk's humorous launch time noted.
First USGS supercomputer decommissioning makes way for successors
The USGS decommissioned its first supercomputer, Yeti, after a decade, and introduced Hovenweep, its most powerful system, to enhance research in environmental management alongside two other supercomputers.
Tenstorrent's Blackhole chips boast 768 RISC-V cores and almost as many FLOPS
Tenstorrent launched Blackhole AI accelerators with 768 RISC-V cores, achieving 24 petaFLOPS performance. The TT-Metalium model supports AI frameworks, enhancing deployment and usability compared to Nvidia's systems.
10 commentsFunny story about Bronson Messer (quoted in the article):
On my first trip to Oak Ridge we went on a tour of “The Machine”. Afterwards we were hanging out on the observation deck and got introduced to something like 10 people.
Everyone at Oak Ridge is just Tom, Bob, etc. No titles or any of that stuff - I’m not sure I’ve ever heard anyone refer to themselves or anyone else as “Doctor”.
Anyway, the guy to my right asks me a question about ML frameworks or something (don’t even remember it specifically). Then he says “Sorry, I’m sure that seems like a really basic question, I’m still learning this stuff. I’m a nuclear astrophysicist by training”.
Then someone yells out “AND a three-time Jeopardy champion”! Everyone laughs.
You guessed it, guy was Bronson.
Place is wild.
Is it really realistic to assume that this is the "fastest supercomputer"? What are estimated sizes for supercomputers used by OpenAI, Microsoft, Google etc?
Strangely enough, the Nature piece only mentions possible secret military supercomputers, but not ones used by AI companies.
Two things I’ve always wondered since I’m not an expert.
1. Obviously, applications must be written to run effectively to distribute the load across the supercomputer. I wonder how often this prevents useful things from being considered to run on the supercomputer.
2. It always seems like getting access to run anything on the supercomputer is very competitive or even artificially limited? A shame this isn’t open to more people. That much processing resources seems like it should go much further to be utilized for more things.
Related
AI is exhausting the power grid
Tech firms, including Microsoft, face a power crisis due to AI's energy demands straining the grid and increasing emissions. Fusion power exploration aims to combat fossil fuel reliance, but current operations heavily impact the environment.
Tenstorrent Unveils High-End Wormhole AI Processors, Featuring RISC-V
Tenstorrent launches Wormhole AI chips on RISC-V, emphasizing cost-effectiveness and scalability. Wormhole n150 offers 262 TFLOPS, n300 doubles power with 24 GB GDDR6. Priced from $999, undercutting NVIDIA. New workstations from $1,500.
XAI's Memphis Supercluster has gone live, with up to 100,000 Nvidia H100 GPUs
Elon Musk launches xAI's Memphis Supercluster with 100,000 Nvidia H100 GPUs for AI training, aiming for advancements by December. Online status unclear, SemiAnalysis estimates 32,000 GPUs operational. Plans for 150MW data center expansion pending utility agreements. xAI partners with Dell and Supermicro, targeting full operation by fall 2025. Musk's humorous launch time noted.
First USGS supercomputer decommissioning makes way for successors
The USGS decommissioned its first supercomputer, Yeti, after a decade, and introduced Hovenweep, its most powerful system, to enhance research in environmental management alongside two other supercomputers.
Tenstorrent's Blackhole chips boast 768 RISC-V cores and almost as many FLOPS
Tenstorrent launched Blackhole AI accelerators with 768 RISC-V cores, achieving 24 petaFLOPS performance. The TT-Metalium model supports AI frameworks, enhancing deployment and usability compared to Nvidia's systems.