Nvidia to Directly Control AI Data Center Grid Loads
TECHWORLD ·
✦ AI Summary
Nvidia said on the 16th that it would participate in the flexible-load interconnection program with "NVIDIA DSX Flex."
DSX Flex flexibly adjusts AI factory power use in line with grid conditions and supports more efficient grid operations.
Nvidia demonstrated automated load reduction with Emerald AI in the Silicon Valley Power environment, and the system protected AI workload performance while responding to hundreds of power demand adjustment request signals.
Nvidia said it has unveiled technology that links AI factory power consumption to grid conditions and will push to expand AI data center use of the power grid. On the 16th, Nvidia announced that it would participate in the flexible-load interconnection program with "NVIDIA DSX Flex." "NVIDIA DSX Flex" supports flexible adjustments to AI factory power use and helps the grid operate more efficiently.
Silicon Valley Power operates the flexible-load interconnection program. The program aims to help ensure grid flexibility for AI factories and includes functions that support adjusting AI data center power consumption according to grid conditions.
Nvidia worked with Emerald AI in the Silicon Valley Power environment. Nvidia and Emerald AI demonstrated automated load reduction in that environment.
The system protected AI workload performance. It also successfully responded to hundreds of power demand adjustment request signals from Silicon Valley Power.
Emerald AI, a Nvidia partner, plans to apply NVIDIA DSX Flex to its grid-responsive power management software, Conductor. Conductor is software that dynamically adjusts an AI factory's energy consumption based on real-time grid signals and hybrid energy sources.
Emerald AI plans to combine DSX Flex so it can automatically manage power use for each AI workload according to grid conditions. The combination works by linking AI factory power use to grid conditions.
DSX Flex receives load reduction requests, demand response (DR) events, and electricity price signals, and responds automatically according to predefined workload priorities. As a result, lower-priority AI tasks can be temporarily paused or reduce power use during periods of high electricity demand.
When grid conditions stabilize, DSX Flex can return lower-priority AI tasks to normal levels. By contrast, it protects higher-priority AI workloads so they continue running.
The core of the technology is adjusting AI factory power demand to match grid conditions, and it showed the potential for AI factories to be used not just as large-scale power-consuming facilities but also as flexible power resources. The technology also demonstrated the potential for AI factories to contribute to the grid.
When grid load relief is needed, AI factory power demand is reduced, and when power supply conditions improve, operations return to normal. This approach may help improve grid operating efficiency and could also help secure additional power capacity for expanding AI infrastructure, the company said.
Nvidia said DSX Flex can automatically control AI factory power consumption while maintaining continuity for higher-priority AI workloads.
Source: TECHWORLD · Park Gyu-chan
Original: https://www.epnc.co.kr/news/articleView.html?idxno=407021
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.