China
Africa
Explore chapters and articles related to this topic
Integration of Graphics Processing Cores with Microprocessors
Published in Tomasz Wojcicki, Krzysztof Iniewski, VLSI: Circuits for Emerging Applications, 2017
Deepak C. Sekar, Chinnakrishnan Ballapuram
HSA: As the demand to use GPU for purposes other than graphics has been increasing, the GPU architecture is also evolving to support both graphics and compute engine. Also, there is an increased need of software support to use GPUs as compute engines in parallel with other asymmetric cores in the system. HSA provides an ecosystem to build a powerful system from combining simple, efficient, unique, and disparate processors. AMD plans to support coherent and unified memory for CPU and GPU and then provide GPU context switching in future products. The HSA road map is to move from physical integration, architectural integration, and finally to system integration. The architectural integration supports unified address space, pageable system memory for GPU and then fully coherent memory between CPU and GPU. The system integration provides preemption, context switching, and quality of service. AMD’s plan is to treat GPU as a first-class core and give equal privileges as CPU cores with HSA. Currently, AMD is positioned well with their commitment and support to hardware, architecture, and OS tools, and applications for HSA framework and ecosystem.
Harmony Search
Published in Nazmul Siddique, Hojjat Adeli, Nature-Inspired Computing, 2017
Niu et al. (2014a,b) applied the HSA with a new pitch adjustment rule to a dynamic EED problem. The pitch adjustment rule is based on the perturbation information and the mean value of the HM. The new HSA is simple and helps in enhancing solution quality and convergence speed. Kherfane et al. (2014) applied the HSA to a power network system with injected renewable energy. To validate the robustness of the proposed approach, the algorithm is tested on the IEEE 30-bus system with six generating units. Three case studies were investigated: minimization of fuel cost, emission of gas, and integration of renewable energy into the network. Comparison of the results with recent global optimization methods shows the superiority of the HSA.
A robust non-iterative algorithm for multi-body dynamics and vehicle–structure interaction analysis
Published in Vehicle System Dynamics, 2022
Arya Datta, Dimitris C. Rizos, Yu Qian, Robert Mullen
The second type of methods uses the substructure approach where the vehicle and the structure are considered as two different systems. Usually, they are solved separately and the coupling is achieved through enforcing compatibility (displacement, velocity or acceleration) and dynamic force equilibrium in an iterative manner within each solution step, [27,35]. The Hertz contact model has been used by authors [3,26,28,30] to model the wheel–rail contact and achieve the solution of the systems in an iterative procedure. The drawback of this type of methods is the slow rate of convergence of the solution, which may even diverge in some cases, thus, requiring significant computation effort. Therefore, the computational advantage of solving smaller substructures may be negated when compared to solving using the monolithic approach. Amongst the substructure approaches, Zhai et al. [36] developed a hybrid explicit-implicit numerical integration scheme to simulate the train–track–bridge interactions. The train, track and the bridge were considered different subsystems. Zhai method [37] was effectively used for dynamic train–track interaction analysis and Newmark-β was used for dynamic analysis of the bridge. The wheel–rail dynamic interactions were studied using spatially dynamic wheel–rail coupling model [38]. The track and the bridge were coupled using track–bridge interaction model. Zhu et al. [39] demonstrated a hybrid solution approach (HSA) to simulate train–track–bridge interactions. The train–track was considered as one system and the solution was obtained using the monolithic approach. Thus, the structural matrices of the train–track system needed to be updated at each time step, depending on the position of the vehicle. The bridge was considered as another subsystem and the coupling of the train–track–bridge system was achieved through an iterative procedure. HSA is shown to be more computationally efficient compared to the monolithic approach. Although, the HSA showed better computational stability compared to the substructure approach it was less computationally efficient than the substructure approach, when using the same time step.