# Tally: Non-Intrusive Performance Isolation for Concurrent Deep Learning Workloads

arXiv: https://arxiv.org/abs/2410.07381 · HTML: https://mlsanity.com/paper/2410.07381/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 9 Oct 2024.

## Built on by (most cited first, 7 of 7)

- [Prism: Cost-Efficient Multi-LLM Serving via GPU Memory Ballooning](https://mlsanity.com/paper/2505.04021/) (2025)
- [OOCO: Latency-disaggregated Architecture for Online-Offline Co-locate LLM Serving](https://mlsanity.com/paper/2511.21862/) (2025)
- [Towards Effective Orchestration of AI x DB Workloads](https://mlsanity.com/paper/2603.03772/) (2026)
- [VUDA: Enabling Controlled Spatial Sharing of Graphics and Compute on NVIDIA GPUs](https://mlsanity.com/paper/2605.01352/) (2026)
- [Gleam: Adaptive Network-Efficient CUDA API Remoting for Cross-Device GPU Sharing over LANs](https://mlsanity.com/paper/2607.23115/) (2026)
- [Efficient Vision-Language-Action Management and Serving for Robot Factories](https://mlsanity.com/paper/2609.12075/) (2026)
- [KerColle: Unlocking Fine-Grained GPU Concurrency in Vision-Language-Action Models](https://mlsanity.com/paper/2609.22335/) (2026)

## Abstract (arXiv)

GPU underutilization is a significant concern in many production deep learning clusters, leading to prolonged job queues and increased operational expenses. A promising solution to this inefficiency is GPU sharing, which improves resource utilization by allowing multiple workloads to execute concurrently on a single GPU. However, deploying GPU sharing in production settings faces critical obstacles due to the limitations of existing mechanisms, including high integration costs, inadequate performance isolation, and limited application compatibility. To address these issues, we introduce Tally, a non-intrusive GPU sharing mechanism that provides robust performance isolation and comprehensive workload compatibility. The key to Tally's robust performance isolation capability lies in its fine-grained thread-block-level GPU kernel scheduling strategy, which allows the system to effectively mitigate interference caused by workload co-execution. We evaluate Tally on a diverse range of workloads and show that it incurs an average overhead of only 7.2% on the 99th-percentile latency of high-priority inference tasks when executed concurrently with best-effort training workloads, compared to 188.9% overhead exhibited by the state-of-the-art GPU sharing systems like TGS, while achieving over 80% of TGS's system throughput.