SOSA: Self-Optimizing Learning with Self-Adaptive Control for Hierarchical System-on-Chip Management
SOSA: Self-Optimizing Learning with Self-Adaptive Control for Hierarchical System-on-Chip Management
复制标题
SOSA：通过自适应控制实现分层片上系统管理的自我优化学习
DOI：
10.1145/3352460.3358312
复制
发表时间：
2019
期刊：
影响因子：
--
通讯作者：
Herkersdorf, Andreas
中科院分区：
文献类型：
--
作者：
Donyanavard, Bryan;Mück, Tiago;Rahmani, Amir M.;Dutt, Nikil;Sadighi, Armin;Maurer, Florian;Herkersdorf, Andreas
众核系统的资源管理策略要求在应用程序之间共享资源，如电源、处理核心和内存带宽，以实现系统目标。系统目标需要考虑系统约束（例如，功率包络）和用户需求（例如，响应时间、能源效率）。现有的方法使用数学、控制理论和机器学习来进行资源管理。它们都依赖于静态系统模型，需要系统动态的先验知识，因此过于僵化，以适应新出现的工作负载或不断变化的系统dynamics. SOSA，跨层的硬件/软件层次资源管理器。低级控制器优化旋钮配置以满足潜在冲突的目标（例如，最大化吞吐量和最小化能量）。SOSA通过使用基于规则的强化学习在运行时从头开始构建子系统模型，为众核系统和不可预测的动态工作负载实现了这一点。SOSA雇用一名高级管理员，以响应由于运行条件而变化的系统目标，例如，由于热事件而从最大化性能切换到最小化功率。SOSA的主管将系统目标转化为低层次的目标（例如，每秒核心处理（IPS））以便通过协调多个旋钮（例如，核心工作频率、任务分配），以实现目标。软件主管允许的灵活性，而硬件学习者允许快速，高效的optimization.We评估SOSA的基于模拟的实现，并证明SOSA的能力，管理多个相互作用的资源存在冲突的目标，其效率配置旋钮，和适应性，在面对不可预测的工作负载。SOSA在多核片上系统上执行机器学习内核和微基准测试的组合，从未经训练的模型开始，SOSA以小于1%的误差实现目标性能，在面对工作负载干扰时保持性能，并在运行时自动适应不断变化的约束。我们还展示了资源管理器的FPGA上的硬件实现。
Resource management strategies for many-core systems dictate the sharing of resources among applications such as power, processing cores, and memory bandwidth in order to achieve system goals. System goals require consideration of both system constraints (e.g., power envelope) and user demands (e.g., response time, energy-efficiency). Existing approaches use heuristics, control theory, and machine learning for resource management. They all depend on static system models, requiring a priori knowledge of system dynamics, and are therefore too rigid to adapt to emerging workloads or changing system dynamics.We present SOSA, a cross-layer hardware/software hierarchical resource manager. Low-level controllers optimize knob configurations to meet potentially conflicting objectives (e.g., maximize throughput and minimize energy). SOSA accomplishes this for many-core systems and unpredictable dynamic workloads by using rule-based reinforcement learning to build subsystem models from scratch at runtime. SOSA employs a high-level supervisor to respond to changing system goals due to operating condition, e.g., switch from maximizing performance to minimizing power due to a thermal event. SOSA's supervisor translates the system goal into low-level objectives (e.g., core instructions-per-second (IPS)) in order to control subsystems by coordinating numerous knobs (e.g., core operating frequency, task distribution) towards achieving the goal. The software supervisor allows for flexibility, while the hardware learners allow quick and efficient optimization.We evaluate a simulation-based implementation of SOSA and demonstrate SOSA's ability to manage multiple interacting resources in the presence of conflicting objectives, its efficiency in configuring knobs, and adaptability in the face of unpredictable workloads. Executing a combination of machine-learning kernels and microbenchmarks on a multicore system-on-a-chip, SOSA achieves target performance with less than 1% error starting with an untrained model, maintains the performance in the face of workload disturbance, and automatically adapts to changing constraints at runtime. We also demonstrate the resource manager with a hardware implementation on an FPGA.
登录
查看更多内容
用于多核异构系统中电源和能源管理的自适应反射中间件
DOI：
10.1049/pbpc022e_ch7
发表时间：
2019
期刊：
Many-Core Computing: Hardware and Software
影响因子：
--
作者：
T. Muck;A. Rahmani;N. Dutt
通讯作者：
N. Dutt
用于 MPSoC 参数优化的低开销监控环互连
DOI：
10.1109/ddecs.2012.6219023
发表时间：
2012
期刊：
2012 IEEE 15th International Symposium on Design and Diagnostics of Electronic Circuits & Systems (DDECS)
影响因子：
--
作者：
Abdelmajid Bouajila;Abdallah Lakhtel;J. Zeppenfeld;W. Stechele;A. Herkersdorf
通讯作者：
A. Herkersdorf
MAMUT：多智能体强化学习，实现高效实时多用户视频转码
DOI：
10.23919/date.2019.8715256
发表时间：
2019
期刊：
2019 Design, Automation & Test in Europe Conference & Exhibition (DATE)
影响因子：
--
作者：
Luis Costero;Arman Iranfar;Marina Zapater;Francisco D. Igual;Katzalin Olcoz;David Atienza Alonso
通讯作者：
David Atienza Alonso
异构多核的响应式鲁棒 MIMO 控制设计方法
DOI：
10.1109/tmscs.2018.2808524
发表时间：
2018
期刊：
IEEE Transactions on Multi-Scale Computing Systems
影响因子：
--
作者：
Muck, Tiago;Donyanavard, Bryan;Moazzemi, Kasra;Rahmani, Amir M.;Jantsch, Axel;Dutt, Nikil
通讯作者：
Dutt, Nikil
黑暗硅时代多核系统的可靠性感知运行时电源管理
DOI：
10.1109/tvlsi.2016.2591798
发表时间：
2017
影响因子：
2.8
作者：
A. Rahmani;M. Haghbayan;A. Miele;P. Liljeberg;A. Jantsch;H. Tenhunen
通讯作者：
H. Tenhunen