Where Do the Joules Go? Diagnosing Inference Energy Consumption
This work addresses energy efficiency for ML practitioners and datacenter operators by providing a diagnostic framework, though it is incremental as it builds on existing measurement studies.
The paper tackles the problem of understanding and diagnosing energy consumption in ML inference by conducting a large-scale measurement study across 46 generative AI models, 7 tasks, and 1,858 configurations, revealing order-of-magnitude variations such as 25x energy differences due to LLM task types and over 100x differences for video vs. image generation.
Energy is now a critical ML computing resource. While measuring energy consumption and observing trends is a valuable first step, accurately understanding and diagnosing why those differences occur is crucial for optimization. To that end, we begin by presenting a large-scale measurement study of inference time and energy across the generative AI landscape with 46 models, 7 tasks, and 1,858 different configurations on NVIDIA H100 and B200 GPUs. Our empirical findings span order-of-magnitude variations: LLM task type can lead to 25$\times$ energy differences, video generation sometimes consumes more than 100$\times$ the energy of images, and GPU utilization differences can result in 3--5$\times$ energy differences. Based on our observations, we present a framework for reasoning about the underlying mechanisms that govern time and energy consumption. The essence is that time and energy are determined by latent metrics like memory and utilization, which are in turn affected by various factors across the algorithm, software, and hardware layers. Our framework also extends directly to throughput per watt, a critical metric for power-constrained datacenters.