Electricity and Control April 2024

INDUSTRY 4.0 + IIOT : PRODUCTS + SERVICES

Energy-hungry data processing microchips raise environmental concerns

As reflected in its increasing stock market value, Nvidia’s H100 AI GPUs (graphics processing units) are taking the tech world by storm, but their reign comes at the price of a hefty energy bill. According to Stocklytics. com, these power-hungry processors are projected to consume some 13 797 GWh in 2024, surpassing the annual energy consumption of entire nations like Georgia and Costa Rica. These findings bring up concerns about the environ mental impact and sustainability of AI advances. Stocklytics Financial Analyst Edith Reads commented on the analysis: “AI, which often requires running com putations on gigabytes of data, needs enormous com puting power compared with ordinary workloads. And Nvidia’s cutting-edge H100 AI GPUs are leading the way in escalating energy consumption. Each H100 GPU, run ning at 61% annual utilisation, consumes roughly 3 740 kilowatt hours (kWh) of electricity annually. This is equiv alent to the average American household.” Reads added though: “While this figure might seem alarming, GPU ef ficiency may improve in the near future, offering a poten tial path towards more sustainable computing.” Venturing into the $30 billion tailored chip market As a leading player in AI chip design, Nvidia is broaden ing its scope by venturing into custom chip development for cloud computing and AI applications. The firm is now looking to tap into the growing custom chip sector, pro jected to reach $10 billion this year and double by 2025. The broader custom chip market hit around $30 billion in 2023, accounting for 5% of chip sales annually. Based in Santa Clara, California, Nvidia targets the changing needs of tech giants such as OpenAI, Microsoft, Alphabet, and Meta Platforms. The company is establishing a division focused on developing custom chips, including powerful artificial intelligence

[Photo credit: american-public-power-association-TF-DL_2L1JM-unsplash]

The soaring energy consumption of AI GPUs raises concerns about the environmental impact and sustainability of AI advances. (AI) processors, for cloud computing firms and other enterprises. Currently holding 80% of the high-end AI chip market share, the position has driven Nvidia’s stock market value up by 40% so far this year to $1.7 trillion after a more than threefold increase in 2023. The energy challenge of powering AI chips As Nvidia’s aspirations grow, concerns are emerging regarding the impact of the escalating energy require ments linked to its cutting-edge chip technologies. According to Paul Churnock, Microsoft’s Principal Electrical Engineer of Datacentre Technical Governance and Strategy, the installation of millions of Nvidia H100 GPUs will consume more energy than all households in Phoenix, Arizona by the end of 2024. Successfully navigating these challenges and fos tering innovation will shape the future landscape of AI computing and beyond. Amazon’s recently unveiled Arm-based Graviton4 and Trainium2 chips hold promise for efficiency gains. □ and as required, so they no longer all have to be clocked at the same rate. The clock rate per core can be defined for real-time transmission and user-mode applications. It is also possible to operate individual cores permanently and in real-time in a so-called ‘turbo mode’. This results in several application benefits, including up to 50% more computing power for one or more processor cores, and the possibility of using more cost-effective CPUs. The permitted power consumption and temperature of each processor core (and of the overall system) is mon itored by TwinCAT Core Boost, to ensure reliable opera tion even when turbo mode is used. TwinCAT Core Boost can be used with all Industrial PCs with Intel ® Core™ I processors from the 11th generation onwards.

Turbo processors advance computing performance

TwinCAT 3 consistently supports modern multicore pro cessor technology. The multi-thread capability enables the application to be distributed across several cores. Supplemented by TwinCAT Core Boost, the computing

performance of individu al real-time or user-mode cores can now be in creased by up to 50% to gain the maximum perfor mance out of the system and adapt it optimally to specific requirements. With TwinCAT Core Boost, the clock frequency of the processor cores can be configured individually

With TwinCAT Core Boost, individual processor cores can be operated as required and in turbo mode.

For more information visit: www.beckhoff.com

8 Electricity + Control APRIL 2024