RA2: Predicting Simulation Execution Time for Cloud-Based Design Space Explorations

Execution time - area tradeoff in gausing residual load decoder: Integrated exploration of chaining based schedule and allocation in HLS for hardware accelerators

Facta universitatis - series Electronics and Energetics ◽

10.2298/fuee1402235s ◽

2014 ◽

Vol 27 (2) ◽

pp. 235-249 ◽

Cited By ~ 1

Author(s):

Anirban Sengupta ◽

Reza Sedaghat ◽

Vipul Mishra

Keyword(s):

Execution Time ◽

Design Space Exploration ◽

Design Space ◽

Space Exploration ◽

Integrated Design ◽

Hardware Accelerators ◽

Average Improvement ◽

Genetic Algorithm Approach ◽

High Level ◽

The Cost

Design space exploration is an indispensable segment of High Level Synthesis (HLS) design of hardware accelerators. This paper presents a novel technique for Area-Execution time tradeoff using residual load decoding heuristics in genetic algorithms (GA) for integrated design space exploration (DSE) of scheduling and allocation. This approach is also able to resolve issues encountered during DSE of data paths for hardware accelerators, such as accuracy of the solution found, as well as the total exploration time during the process. The integrated solution found by the proposed approach satisfies the user specified constraints of hardware area and total execution time (not just latency), while at the same time offers a twofold unified solution of chaining based schedule and allocation. The cost function proposed in the genetic algorithm approach takes into account the functional units, multiplexers and demultiplexers needed during implementation. The proposed exploration system (ExpSys) was tested on a large number of benchmarks drawn from the literature for assessment of its efficiency. Results indicate an average improvement in Quality of Results (QoR) greater than 26% when compared to a recent well known GA based exploration method.

Download Full-text

Regression-Based Prediction for Task-Based Program Performance

Journal of Circuits System and Computers ◽

10.1142/s0218126619500609 ◽

2019 ◽

Vol 28 (04) ◽

pp. 1950060 ◽

Cited By ~ 1

Author(s):

Isil Oz ◽

Muhammad Khurram Bhatti ◽

Konstantin Popov ◽

Mats Brorsson

Keyword(s):

Execution Time ◽

Regression Models ◽

Input Data ◽

Design Space ◽

Programming Model ◽

Parallel Execution ◽

Program Performance ◽

Multiple Factors ◽

Parallel Tasks ◽

Processor Cores

As multicore systems evolve by increasing the number of parallel execution units, parallel programming models have been released to exploit parallelism in the applications. Task-based programming model uses task abstractions to specify parallel tasks and schedules tasks onto processors at runtime. In order to increase the efficiency and get the highest performance, it is required to identify which runtime configuration is needed and how processor cores must be shared among tasks. Exploring design space for all possible scheduling and runtime options, especially for large input data, becomes infeasible and requires statistical modeling. Regression-based modeling determines the effects of multiple factors on a response variable, and makes predictions based on statistical analysis. In this work, we propose a regression-based modeling approach to predict the task-based program performance for different scheduling parameters with variable data size. We execute a set of task-based programs by varying the runtime parameters, and conduct a systematic measurement for influencing factors on execution time. Our approach uses executions with different configurations for a set of input data, and derives different regression models to predict execution time for larger input data. Our results show that regression models provide accurate predictions for validation inputs with mean error rate as low as 6.3%, and 14% on average among four task-based programs.

Download Full-text