The Experts below are selected from a list of 123 Experts worldwide ranked by ideXlab platform
Christos Kozyrakis - One of the best experts on this subject based on the ideXlab platform.
-
dynamic management of turbomode in modern multi core chips
High-Performance Computer Architecture, 2014Co-Authors: Christos KozyrakisAbstract:Dynamic overclocking of CPUs, or TurboMode, is a feature recently introduced on all x86 multi-core chips. It leverages thermal and Power Headroom from idle execution resources to overclock active cores to increase performance. TurboMode can accelerate CPU-bound applications at the cost of additional Power consumption. Nevertheless, naive use of TurboMode can significantly increase Power consumption without increasing performance. Thus far, there is no strategy for managing TurboMode to optimize its use across all workloads and efficiency metrics. This paper analyzes the impact of TurboMode on a wide range of efficiency metrics (performance, Power, cost, and combined metrics such as QPS=W and ED2) for representative server workloads on various hardware configurations. We determine that TurboMode is generally beneficial for performance (up to +24%), cost efficiency (QPS/$ up to +8%), energy-delay product (ED, up to +47%), and energy-delay-squared product (ED2, up to +68%). However, TurboMode is inefficient for workloads that exhibit interference for shared resources. We use this information to build and validate a model that predicts the optimal TurboMode setting for each efficiency metric. We then implement autoturbo, a background daemon that dynamically manages TurboMode in real time without any hardware changes. We demonstrate that autoturbo improves QPS=$, ED, and ED2 by 8%, 47%, and 68% respectively over not using TurboMode. At the same time, autoturbo virtually eliminates all the large drops in those same metrics (−12%, −25%, −25% for QPS/$, ED, and ED2) that occur when TurboMode is used naively (always on).
-
HPCA - Dynamic management of TurboMode in modern multi-core chips
2014 IEEE 20th International Symposium on High Performance Computer Architecture (HPCA), 2014Co-Authors: Christos KozyrakisAbstract:Dynamic overclocking of CPUs, or TurboMode, is a feature recently introduced on all x86 multi-core chips. It leverages thermal and Power Headroom from idle execution resources to overclock active cores to increase performance. TurboMode can accelerate CPU-bound applications at the cost of additional Power consumption. Nevertheless, naive use of TurboMode can significantly increase Power consumption without increasing performance. Thus far, there is no strategy for managing TurboMode to optimize its use across all workloads and efficiency metrics. This paper analyzes the impact of TurboMode on a wide range of efficiency metrics (performance, Power, cost, and combined metrics such as QPS=W and ED2) for representative server workloads on various hardware configurations. We determine that TurboMode is generally beneficial for performance (up to +24%), cost efficiency (QPS/$ up to +8%), energy-delay product (ED, up to +47%), and energy-delay-squared product (ED2, up to +68%). However, TurboMode is inefficient for workloads that exhibit interference for shared resources. We use this information to build and validate a model that predicts the optimal TurboMode setting for each efficiency metric. We then implement autoturbo, a background daemon that dynamically manages TurboMode in real time without any hardware changes. We demonstrate that autoturbo improves QPS=$, ED, and ED2 by 8%, 47%, and 68% respectively over not using TurboMode. At the same time, autoturbo virtually eliminates all the large drops in those same metrics (−12%, −25%, −25% for QPS/$, ED, and ED2) that occur when TurboMode is used naively (always on).
N J Dimopoulos - One of the best experts on this subject based on the ideXlab platform.
-
execution phase prediction based on phase precursors and locality
International Workshop on Energy Efficient Supercomputing, 2017Co-Authors: Saman Khoshbakht, N J DimopoulosAbstract:This paper focuses on different methods developed to detect the upcoming execution phase of a workload with regards to Power demands. By controlling the state of the processor in Power demanding phases, the operating system can maintain a relatively steady Power pattern in the workload, leading to higher Power Headroom in the system. We compared two main approaches in phase prediction. Firstly, we show that by detecting the precursors leading to an upcoming phase, the system can speculate the next phase with high accuracy. Additionally, we compared this method with another approach which relies on the assumption of phase locality, expecting the current dominant phase to continue in the near future. Our results show that by detecting the precursors we can detect 81% of the upcoming phases with lower processor frequency switching overhead compared to most of the proposed locality-based methods.
-
E2SC@SC - Execution Phase Prediction Based on Phase Precursors and Locality
Proceedings of the 5th International Workshop on Energy Efficient Supercomputing, 2017Co-Authors: Saman Khoshbakht, N J DimopoulosAbstract:This paper focuses on different methods developed to detect the upcoming execution phase of a workload with regards to Power demands. By controlling the state of the processor in Power demanding phases, the operating system can maintain a relatively steady Power pattern in the workload, leading to higher Power Headroom in the system. We compared two main approaches in phase prediction. Firstly, we show that by detecting the precursors leading to an upcoming phase, the system can speculate the next phase with high accuracy. Additionally, we compared this method with another approach which relies on the assumption of phase locality, expecting the current dominant phase to continue in the near future. Our results show that by detecting the precursors we can detect 81% of the upcoming phases with lower processor frequency switching overhead compared to most of the proposed locality-based methods.
Jani Puttonen - One of the best experts on this subject based on the ideXlab platform.
-
VTC Fall - Using LTE Power Headroom Report for Coverage Optimization
2011 IEEE Vehicular Technology Conference (VTC Fall), 2011Co-Authors: Jussi Turkka, Jani PuttonenAbstract:This paper describes how LTE Rel' 8 Power Headroom report (PHR) can be used to detect outage problems in LTE networks. The paper outlines the proposed minimization of drive tests use case for PH triggered measurements, alleviates the problems in this proposal, and describes improvements which can help operators to gather more coverage and outage related information from their networks. The article points out, that the earlier proposal can lead to unnecessary signaling, it does not work always as supposed, and the existing measurements can be used to gather the same information. The study is conducted by simulating two different outage problem scenarios with a fully dynamic LTE system simulator.
Linzhen Xie - One of the best experts on this subject based on the ideXlab platform.
-
GLOBECOM - Power synergy to enhance DCI reliability for OFDM-based mobile system optimization
2014 IEEE Global Communications Conference, 2014Co-Authors: Min Chen, Anpeng Huang, Linzhen XieAbstract:In Orthogonal Frequency Division Multiplexing (OFDM)-based mobile networks (e.g., Long Term Evolution Advanced), Downlink Control Information (DCI) in a Physical Downlink Control Channel (PDCCH) plays a unique role in carrying control signaling and scheduling information, which enables the flexibility and diversity of radio resource utilization. Due to system configuration limitations, Power dimension is the only potential to enhance DCI reliability. Based on this motivation, we apply the synergy concept to Power dimension: called Power Synergy, which can deal with negative effects caused by Cell-specific Reference Signal (CRS) Power boosting, and also allow flexibly Power lending between OFDM symbols in the physical control region. Under the constraints of the Block Error Rate (BLER) threshold 102 given for PDCCHs and the CRS full-cell coverage requirement, performance tests reveal there are significant profits (for example, 10-dB gain available for enhancing DCI reliability in the worst case) from the Power Headroom in the physical control region. This valuable gain is helpful to accommodate emerging mobile broadband services (e.g., mobile health) in LTE/LTE-A networks.
Yixiang Gao - One of the best experts on this subject based on the ideXlab platform.
-
Adaptive Power Point Tracking Control of PV System for Primary Frequency Regulation of AC Microgrid with High PV Integration
IEEE Transactions on Power Systems, 2021Co-Authors: Zhiping Cheng, Shuyuan Zhang, Dong Lianghui, Yixiang GaoAbstract:With the increasing integration of PV generation in AC microgrids, it is challenging to keep the stability of system frequency due to the intermittent and stochastic nature of PVs. Thus, in order to reduce the investment and maintenance costs of storage systems, the electric utility has shown increasing interest in calling on PVs to provide frequency regulation services. In this paper, a novel sliding mode control (SMC) based adaptive Power point tracking (APPT) control strategy is proposed to provide bi-directional primary frequency regulation (BPFR) of an AC microgrid. In this strategy, in order to restrain the variation of frequency, the sliding mode surface is adaptively regulated to provide an adaptive Power reserve. Thus, the PVs will have a Power Headroom to regulate the frequency both up and down through releasing the reserved Power or increasing the reserves of the PV systems. Furthermore, the sliding mode surface is adaptively regulated only based on the locally measured frequency of the microgrid without the requirements of communication, detailed PV model, irradiance sensors as well as MPP estimators. Thus, the proposed control strategy shows the advantages of easy to implement and reduces the investment and maintenance costs of installing storage systems or irradiance sensors.