CORD: Co-design of Resource Allocation and Deadline Decomposition with Generative Profiling

TL;DR

This work addresses real-time DAG scheduling on multicore platforms where shared-resource contention causes time-varying execution times. It introduces Cord, a co-design that uses generative resource profiling based on a multimarginal Schrödinger bridge to create conditional, time-varying execution models Θ_{τ|β} and to jointly allocate per-subtask resources and deadlines. The core contributions are (i) an efficient MSB-based generative profiling pipeline that learns ξ(t)|β from limited measurements, and (ii) the Cord algorithm that iteratively reassigns budgets and deadlines to maximize execution progress and schedulability, using the multi-phase model to guide decisions. Empirical results on PARSEC/SPLASH-2x benchmarks show substantial schedulability gains over state-of-the-art deadline decomposition, while the generative profiler reduces required profiling data by orders of magnitude, enabling scalable resource-aware real-time scheduling on multicore systems.

Abstract

As multicore hardware is becoming increasingly common in real-time systems, traditional scheduling techniques that assume a single worst-case execution time for a task are no longer adequate, since they ignore the impact of shared resources on execution time. When tasks execute concurrently on different cores, their execution times often vary substantially with their allocated budgets of shared resources, such as cache and memory bandwidth. Even under a specific resource allocation, the resource use pattern of a task also changes with time during a job execution. It is therefore important to consider the relationship between multicore resources and execution time in task modeling and scheduling algorithm design. In this paper, we propose a much more precise execution model for DAG-based real-time tasks that captures the time-varying resource use characteristics of a task under different budgets of shared resources. We present a generative resource profiling algorithm that efficiently predicts, from limited measurement data, the resource profile of a task at any time during its execution under a given resource budget. The generative profiles can then be used to construct the execution models for tasks, using which one can make informed resource allocation decisions. We further introduce a multicore resource allocation and deadline decomposition co-design technique for DAG-based tasks that leverages the generated execution models to jointly allocate resources and deadlines to subtasks, to maximize resource efficiency and schedulability. Our evaluation results show that our generative profiling algorithm achieves high accuracy while being efficient, and that our co-allocation technique substantially improves schedulability compared to a state-of-the-art deadline decomposition method.

Figures (5)

• Figure 1: Workflow of our generative profiling and schedule-resource co-design algorithms.
• Figure 2: Maximum-likelihood synthetic profile (blue), mean synthetic profile (red), mean empirical profile (black), and all empirical profiles (grey) for the benchmarks $\mathsf{dedup,\: fft,\: canneal,\:}$ and $\mathsf{radiosity}$ when $\beta=(13,\:15)^\top$.
• Figure 3: Fraction of schedulable tasksets on $m = 4$ cores for different $p$ values
• Figure 4: Schedulability ($m = 10$, $p = 0.5$)
• Figure 5: Cord with generative profiles.

Theorems & Definitions (2)

• Remark
• Remark