# Characterization of HBM Power Consumption (ETH Zurich, NYU)

DevFeed: [Characterization of HBM Power Consumption (ETH Zurich, NYU)](<https://devfeed.tech/articles/characterization-of-hbm-power-consumption-eth-zurich-nyu-86829.md>)

Original publisher: [Read original article](<https://semiengineering.com/characterization-of-hbm-power-consumption-eth-zurich-nyu/>)

Author: Technical Paper Link

Published: 2026-10-14T15:27:00Z

Content type: article

Language: en

Sources: [Semiconductor Engineering](<https://devfeed.tech/sources/semiconductor-engineering.md>)

Topics: [ddr5](<https://devfeed.tech/topics/ddr5.md>), [Chip design](<https://devfeed.tech/topics/chip-design.md>)

Tags: [3d-stacked-memory](<https://devfeed.tech/tags/3d-stacked-memory.md>), [ai-accelerators](<https://devfeed.tech/tags/ai-accelerators.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [dram](<https://devfeed.tech/tags/dram.md>), [eth-zurich](<https://devfeed.tech/tags/eth-zurich.md>), [hbm](<https://devfeed.tech/tags/hbm.md>), [hbm2](<https://devfeed.tech/tags/hbm2.md>), [hbm3e](<https://devfeed.tech/tags/hbm3e.md>), [hbm4](<https://devfeed.tech/tags/hbm4.md>), [high-bandwidth-memory](<https://devfeed.tech/tags/high-bandwidth-memory.md>), [jedec](<https://devfeed.tech/tags/jedec.md>), [memory](<https://devfeed.tech/tags/memory.md>), [memory-architecture](<https://devfeed.tech/tags/memory-architecture.md>), [memory-power](<https://devfeed.tech/tags/memory-power.md>), [new-york-university](<https://devfeed.tech/tags/new-york-university.md>), [nvidia-h200](<https://devfeed.tech/tags/nvidia-h200.md>), [nyu](<https://devfeed.tech/tags/nyu.md>), [paper](<https://devfeed.tech/tags/paper.md>), [power](<https://devfeed.tech/tags/power.md>), [power-consumption](<https://devfeed.tech/tags/power-consumption.md>), [power-modeling](<https://devfeed.tech/tags/power-modeling.md>), [power-performance](<https://devfeed.tech/tags/power-performance.md>), [technical-papers](<https://devfeed.tech/tags/technical-papers.md>), [thermal-throttling](<https://devfeed.tech/tags/thermal-throttling.md>), [volatile-memory](<https://devfeed.tech/tags/volatile-memory.md>)

## AI overview

Researchers from ETH Zürich and New York University experimentally characterized power use in HBM2 DRAM chips and developed data on structural and data-pattern-related variation. They report that read operations consume more current than other operations, toggling all internal bank-group bus signals raises power by 29.4% versus keeping them constant, and a designed access pattern used 23.6% more power on average across 36 chips than the highest-power JEDEC-specified measurement loop.

## Source excerpt

Researchers at ETH Zürich and New York University published a technical paper titled "Understanding the Power Consumption of Modern High Bandwidth Memory: Experimental Analysis and Modeling Using Real HBM2 DRAM Chips." Abstract Excerpt: "We perform the first comprehensive experimental characterization of power consumed by real HBM2 DRAM chips. Our extensive characterization yields 1) the empirical... " read more The post Characterization of HBM Power Consumption (ETH Zurich, NYU) appeared first on Semiconductor Engineering.