A Research Of Program Performance Adapting Simulation In Cloud Environment

With the rapid development of hardware, network and communications technology, the use of heterogeneous resources is inevitable, rational use of heterogeneous resources has become increasingly important and become a research hotspot.For Julia oriented programming language cloud environment task(program) implementation of a performance problem, we propose a time constraint based programs running performance of the adaptive strategy. First, in order to facilitate the implementation of the policy described, and can correctly identify the task, According to a particular issue-oriented(such as the document case models to identify cases) design a set of Julia primitive, describe this task by use Julia primitive; then put forward program performance adapting simulation in cloud environment based on time constraints, primitive and back-end resource nodes can adaptively generate mappings by the strategy, That is all the primitive program and heterogeneous resource library tools can automatic matching merit, Enabling program execution time within the period, the lower bound of the range of proposed user consumption, enabling program execution time allows the user to lower consumption within the time constraints of the proposed user. In order to verify the reasonableness of the policy. In order to verify the reasonableness of the policy for vehicle recognition cases, based verification colored Petri nets and with the CPN TOOL Colored Petri Net tools to build the model, compare multiple options generated during the simulation, we verified the feasibility and effectiveness of adaptive strategy.Due to write parallel programs related to parallel computing, distributed computing, and cloud computing and other related knowledge on heterogeneous resources, Parallel program on the preparation of heterogeneous resources become research difficulties, to this end we cloud-based Julia dynamic language programming model, Create a cloud-based Julia language programming platform. In Julia dynamic cloud-oriented programming language environment. Use of specific methods to achieve high performance, forming a library of tools for problem areas, deployed to the corresponding heterogeneous computing environment(implementation patterns) in accordance with their respective characteristics. Programmers can then arrange Julia programming platform in the cloud program window, write / debug / run the application, the better solution heterogeneous parallel programming resources on the preparation of the issue. Operational phase, based on Julia calibration procedures primitive tool library property, dynamic combination of the right tools(dynamic runtime) on the adaptive strategy, to meet the end user performance and time requirements.