David Irwin
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
David Irwin has a total of 97 co-authors. Here are the top co-authors by number of shared publications:
- Prashant Shenoy
University of Massachusetts Amherst, Amherst, Massachusetts, USA
40 shared publications - Noman Bashir
Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
23 shared publications - Walid A. Hanafy
University of Massachusetts Amherst, Amherst, USA
18 shared publications - Qianlin Liang
University of Massachusetts Amherst, Amherst, MA, USA
7 shared publications - Ramesh K. Sitaraman
University of Massachusetts Amherst, Amherst, MA, USA
6 shared publications - Abel Souza
University of California, Santa Cruz, CA, USA
5 shared publications
Latest Publications
- Open Access
- 22 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 349-362
Go with the Flow: Analyzing the Carbon Footprint of Green Streaming
- 22 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 349-362
Open AccessGo with the Flow: Analyzing the Carbon Footprint of Green Streaming
- Open Access
- 22 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 390-394
City-Scale Decarbonization of Residential Heating under Transformer Constraints
- 22 June 2026
- Published by Association for Computing Machinery (ACM)
- p. 390-394
Open AccessCity-Scale Decarbonization of Residential Heating under Transformer Constraints
- Open Access
- 3 May 2026
- Published by Association for Computing Machinery (ACM)
- p. 25-36
CarbonShare: Carbon-Fair Allocation for Shared Clusters
- 3 May 2026
- Published by Association for Computing Machinery (ACM)
- p. 25-36
Open AccessCarbonShare: Carbon-Fair Allocation for Shared Clusters
- Open Access
- 3 May 2026
- Published by Association for Computing Machinery (ACM)
- p. 333-340
To Offload or Not To Offload: Model-driven Comparison of Edge-native and On-device Processing In the Era of Accelerators
Computational offloading is a promising approach to overcome client device resource constraints by moving application computations to remote servers. With the advent of specialized hardware accelerators, client devices can now perform fast local processing of tasks such as machine learning inference, reducing the need for offloading. However, edge servers with accelerators also offer faster offloading performance than was previously possible. In this paper, we present an analytic and experimental comparison of on-device processing and edge offloading across accelerator, multi-tenant, and workload scenarios to understand when to use local processing versus offloading. We present models that leverage queuing theory to derive explainable closed-form equations for end-to-end latencies, yielding quantitative performance crossover predictions to guide adaptive offloading. We validate our models across various settings and show that they achieve a mean absolute percentage error of 2.2% compared to observed latencies. We further use these models to develop a resource manager for adaptive offloading and demonstrate its effectiveness in dynamic multi-tenant edge environments.
- 3 May 2026
- Published by Association for Computing Machinery (ACM)
- p. 333-340
Open AccessTo Offload or Not To Offload: Model-driven Comparison of Edge-native and On-device Processing In the Era of Accelerators
Computational offloading is a promising approach to overcome client device resource constraints by moving application computations to remote servers. With the advent of specialized hardware accelerators, client devices can now perform fast local processing of tasks such as machine learning inference, reducing the need for offloading. However, edge servers with accelerators also offer faster offloading performance than was previously possible. In this paper, we present an analytic and experimental comparison of on-device processing and edge offloading across accelerator, multi-tenant, and workload scenarios to understand when to use local processing versus offloading. We present models that leverage queuing theory to derive explainable closed-form equations for end-to-end latencies, yielding quantitative performance crossover predictions to guide adaptive offloading. We validate our models across various settings and show that they achieve a mean absolute percentage error of 2.2% compared to observed latencies. We further use these models to develop a resource manager for adaptive offloading and demonstrate its effectiveness in dynamic multi-tenant edge environments.
- 24 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 75 (8) , 2715-2729
Untangling the Carbon-Cost Tradeoffs and Stampede Effect Challenges in Cloud Computing
As society explores new computing applications, data centers have seen a sharp rise in capacity and energy use, driving up data centers’ carbon footprint and raising concerns about their sustainability. In response, efforts to reduce emissions now complement the traditional goals of cutting energy costs and boosting performance. Given the spatiotemporal variability of the grid carbon intensity, researchers are increasingly adopting workload shifting strategies to minimize the operational carbon footprint of cloud workloads. However, shifting from cost- and performance-centric operations to carbon-aware operations introduces performance penalties and cost overheads. Moreover, large-scale spatial and temporal shifts can cause stampede effects, with synchronized migration to the same low-carbon regions or times, thereby straining resources and creating imbalances. In this paper, we quantify the carbon-cost-performance trade-offs of carbon-aware scheduling and evaluate the risk of such stampede effects. We provide real-world examples illustrating these tradeoffs for application and cloud providers. Lastly, we offer practical insights to help mitigate these challenges and guide sustainable scheduling decisions.
- 24 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computers
- Vol. 75 (8) , 2715-2729
Untangling the Carbon-Cost Tradeoffs and Stampede Effect Challenges in Cloud Computing
As society explores new computing applications, data centers have seen a sharp rise in capacity and energy use, driving up data centers’ carbon footprint and raising concerns about their sustainability. In response, efforts to reduce emissions now complement the traditional goals of cutting energy costs and boosting performance. Given the spatiotemporal variability of the grid carbon intensity, researchers are increasingly adopting workload shifting strategies to minimize the operational carbon footprint of cloud workloads. However, shifting from cost- and performance-centric operations to carbon-aware operations introduces performance penalties and cost overheads. Moreover, large-scale spatial and temporal shifts can cause stampede effects, with synchronized migration to the same low-carbon regions or times, thereby straining resources and creating imbalances. In this paper, we quantify the carbon-cost-performance trade-offs of carbon-aware scheduling and evaluate the risk of such stampede effects. We provide real-world examples illustrating these tradeoffs for application and cloud providers. Lastly, we offer practical insights to help mitigate these challenges and guide sustainable scheduling decisions.
- Open Access
Collaborative Processing for Multi-Tenant Inference on Memory-Constrained Edge TPUs
IoT applications are increasingly relying on on-device AI accelerators to ensure high performance, especially in limited connectivity and safety-critical scenarios. However, the limited on-chip memory of these accelerators forces inference runtimes to swap model segments between host and accelerator memory, substantially inflating latency. While collaborative processing by partitioning the model processing between CPU and accelerator resources can reduce accelerator memory pressure and latency, naive partitioning may worsen end-to-end latency by either shifting excessive computation to the CPU or failing to sufficiently curb swapping, a problem that is further amplified in multi-tenant and dynamic environments. To address these issues, we present SwapLess, a system for adaptive, multi-tenant TPU-CPU collaborative inference for memory-constrained Edge TPUs. SwapLess utilizes an analytic queueing model that captures partition-dependent CPU/TPU service times as well as inter- and intra-model swapping overheads across different workload mixes and request rates. Using this model, SwapLess continuously adjusts both the partition point and CPU core allocation online to minimize end-to-end response time with low decision overhead. An implementation on Edge TPU-equipped platforms demonstrates that SwapLess reduces mean latency by up to 63.8% for single-tenant workloads and up to 77.4% for multi-tenant workloads relative to the default Edge TPU compiler.
Open AccessCollaborative Processing for Multi-Tenant Inference on Memory-Constrained Edge TPUs
IoT applications are increasingly relying on on-device AI accelerators to ensure high performance, especially in limited connectivity and safety-critical scenarios. However, the limited on-chip memory of these accelerators forces inference runtimes to swap model segments between host and accelerator memory, substantially inflating latency. While collaborative processing by partitioning the model processing between CPU and accelerator resources can reduce accelerator memory pressure and latency, naive partitioning may worsen end-to-end latency by either shifting excessive computation to the CPU or failing to sufficiently curb swapping, a problem that is further amplified in multi-tenant and dynamic environments. To address these issues, we present SwapLess, a system for adaptive, multi-tenant TPU-CPU collaborative inference for memory-constrained Edge TPUs. SwapLess utilizes an analytic queueing model that captures partition-dependent CPU/TPU service times as well as inter- and intra-model swapping overheads across different workload mixes and request rates. Using this model, SwapLess continuously adjusts both the partition point and CPU core allocation online to minimize end-to-end response time with low decision overhead. An implementation on Edge TPU-equipped platforms demonstrates that SwapLess reduces mean latency by up to 63.8% for single-tenant workloads and up to 77.4% for multi-tenant workloads relative to the default Edge TPU compiler.
- Open Access
- 3 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-16
EcoLearn: Optimizing the Carbon Footprint of Federated Learning
- 3 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-16
Open AccessEcoLearn: Optimizing the Carbon Footprint of Federated Learning
- Open Access
- 3 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-17
LLM-Driven Auto Configuration for Transient IoT Device Collaboration
- 3 December 2025
- Published by Association for Computing Machinery (ACM)
- p. 1-17
Open AccessLLM-Driven Auto Configuration for Transient IoT Device Collaboration
- Open Access
- 19 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 416-429
FailLite: Failure-Resilient Model Serving for Resource-Constrained Edge Environments
- 19 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 416-429
Open AccessFailLite: Failure-Resilient Model Serving for Resource-Constrained Edge Environments
- Open Access
- 19 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 762-775
PowerTrip: Exploiting Federated Heterogeneous Datacenter Power for Distributed ML Training
- 19 November 2025
- Published by Association for Computing Machinery (ACM)
- p. 762-775
Open AccessPowerTrip: Exploiting Federated Heterogeneous Datacenter Power for Distributed ML Training