← 研踪 PaperTrailarXiv:2610.41672 · v2
Understanding the Energy Cost of Quantized Agents on Shared GPUs
首次提交 2026-10-30 20:40
版本更新 2026-11-03 07:25
雷达观测 2026-11-04 23:40
摘要
Quantization can lower the energy of language-model calls and free memory for serving more agents, but its benefit per completed task remains unclear when agents share a GPU. Measurements of fixed requests or isolated agents leave open how the agents' generated work interact with competition for compute and cache. We present a systematic study of precision and admission limits for two agent models on A100 deployments with 40 or 80 GB of memory, using coding and data-analysis tasks with independently verified outcomes. We replay recorded calls across precisions to isolate execution energy and cap the cache to test the value of freed memory. For both models with four admitted agents where requests fit in the cache, 4-bit weights with 16-bit activations execute the same calls with 10-25% less energy than 8-bit weights and activations. Despite this saving, extra work and fewer completions in live runs raise their energy per verified completion by 49%. Where requests compete for cache on the 40 GB deployment, 4-bit weights cost less, and capping their cache raises energy per verified completion by 34% with 16 agents. These results show how generated work and cache demand change the energy benefit of quantization, providing a practical basis for choosing precision and admission limits together.
作者备注：16 pages, 8 figures. Alex Flesher and Aldric Yang contributed equally
观测到的版本
相邻记录可能跳过中间版本；首次发现 v2 及以上也会标为版本更新。
v2 · 2026-11-03 07:25 · 最新观测版本
Understanding the Energy Cost of Quantized Agents on Shared GPUs
Quantization can lower the energy of language-model calls and free memory for serving more agents, but its benefit per completed task remains unclear when agents share a GPU. Measurements of fixed requests or isolated agents leave open how the agents' generated work interact with competition for compute and cache. We present a systematic study of precision and admission limits for two agent models on A100 deployments with 40 or 80 GB of memory, using coding and data-analysis tasks with independently verified outcomes. We replay recorded calls across precisions to isolate execution energy and cap the cache to test the value of freed memory. For both models with four admitted agents where requests fit in the cache, 4-bit weights with 16-bit activations execute the same calls with 10-25% less energy than 8-bit weights and activations. Despite this saving, extra work and fewer completions in live runs raise their energy per verified completion by 49%. Where requests compete for cache on the 40 GB deployment, 4-bit weights cost less, and capping their cache raises energy per verified completion by 34% with 16 agents. These results show how generated work and cache demand change the energy benefit of quantization, providing a practical basis for choosing precision and admission limits together.
查看此版本解读 →查看此版本原文 ↗