US Department of Energy requested suppliers of information to create a supercomputer with a capacity of more than 10 exaflops. The installation is planned to be launched by 2030.
According to the document, the ministry is interested in deploying one or more supercomputers capable of solving high-precision tasks five to ten times faster than current systems.
The department expects that by 2030 they will be able to assemble a plant with a capacity of 10-20 exaflops. After 2030, system performance is planned to be increased to 100 exaflops, which they intend to implement “thanks to hardware and software acceleration mechanisms.”
It is expected that the power consumption of such a plant will be 20-60 MW.
The ministry also announced a desire to move from “monolithic installations” to modular systems. This will contribute to the faster implementation of innovations in the field of hardware and software.
As a result, the upgrade cycle of supercomputer components can be reduced from four to five years to 12 to 24 months, according to the department.
The level of information requested by the Ministry of Energy from suppliers covers promising areas for 2025-2030:
- types of processors, memory and storage;
- connection options;
- system-on-a-chip combinations;
- advanced technical processes;
- throughput expectations;
- potential node configuration and so on.
It also indicates that future installations will require an appropriate software stack designed for a wide range of tasks in the field of large-scale modeling, machine learning and data analysis. They must be fault-tolerant to minimize manual labor.
Vendors must submit their proposals by the end of July 2022.
Recall that in May, the American supercomputer Frontier took first place in the Top500 ranking. This is the first rig to hit a peak of 1.1 exaflops in the Linmark test.
In June, American startup Cerebras trained the “largest artificial intelligence model” on a single device, setting a world record.
In May 2021, scientists unveiled an AI supercomputer that will help build the largest ever 3D map of the visible universe to study dark energy.
Subscribe to Cryplogger news in Telegram: Cryplogger AI – all the news from the world of AI!
Found a mistake in the text? Select it and press CTRL+ENTER
US Department of Energy requested suppliers of information to create a supercomputer with a capacity of more than 10 exaflops. The installation is planned to be launched by 2030.
According to the document, the ministry is interested in deploying one or more supercomputers capable of solving high-precision tasks five to ten times faster than current systems.
The department expects that by 2030 they will be able to assemble a plant with a capacity of 10-20 exaflops. After 2030, system performance is planned to be increased to 100 exaflops, which they intend to implement “thanks to hardware and software acceleration mechanisms.”
It is expected that the power consumption of such a plant will be 20-60 MW.
The ministry also announced a desire to move from “monolithic installations” to modular systems. This will contribute to the faster implementation of innovations in the field of hardware and software.
As a result, the upgrade cycle of supercomputer components can be reduced from four to five years to 12 to 24 months, according to the department.
The level of information requested by the Ministry of Energy from suppliers covers promising areas for 2025-2030:
- types of processors, memory and storage;
- connection options;
- system-on-a-chip combinations;
- advanced technical processes;
- throughput expectations;
- potential node configuration and so on.
It also indicates that future installations will require an appropriate software stack designed for a wide range of tasks in the field of large-scale modeling, machine learning and data analysis. They must be fault-tolerant to minimize manual labor.
Vendors must submit their proposals by the end of July 2022.
Recall that in May, the American supercomputer Frontier took first place in the Top500 ranking. This is the first rig to hit a peak of 1.1 exaflops in the Linmark test.
In June, American startup Cerebras trained the “largest artificial intelligence model” on a single device, setting a world record.
In May 2021, scientists unveiled an AI supercomputer that will help build the largest ever 3D map of the visible universe to study dark energy.
Subscribe to Cryplogger news in Telegram: Cryplogger AI – all the news from the world of AI!
Found a mistake in the text? Select it and press CTRL+ENTER
