One of the main drawbacks that significantly impacts the performance of dynamically reconfigurable systems (like FPGAs), is their high reconfiguration overhead. Configuration prefetching is one method to reduce this penalty by overlapping FPGA reconfigurations with useful computations. In this paper we propose a speculative approach that schedules prefetches at design time and simultaneously performs HW/SW partitioning, in order to minimize the expected execution time of an application. Our method prefetches and executes in hardware those configurations that provide the highest performance improvement. The algorithm takes into consideration profiling information (such as branch probabilities and execution time distributions), correlated with the application characteristics. Compared to the previous state-of-art, we reduce the reconfiguration penalty with 34% on average, and with up to 59% for particular case studies.

This paper presents an approach to system-level optimization of error detection implementation in the context of fault-tolerant realtime distributed embedded systems used for safety-critical applications. An application is modeled as a set of processes communicating by messages. Processes are mapped on computation nodes connected to the communication infrastructure. To provide resiliency against transient faults, efficient error detection and recovery techniques have to be employed. Our main focus in this paper is on the efficient implementation of the error detection mechanisms. We have developed techniques to optimize the hardware/software implementation of error detection, in order to minimize the global worst-case schedule length, while meeting the imposed hardware cost constraints and tolerating multiple transient faults. We present two design optimization algorithms which are able to find feasible solutions given a limited amount of resources: the first one assumes that, when implemented in hardware, error detection is deployed on static reconfigurable FPGAs, while the second one considers partial dynamic reconfiguration capabilities of the FPGAs.

Abstract-Modern systems demand high performance, as well as high degrees of flexibility and adaptability. Many current applications exhibit a dynamic and nonstationary behavior, having certain characteristics in one phase of their execution, that will change as the applications enter new phases, in a manner unpredictable at design-time. In order to meet the performance requirements of such systems, it is important to have on-line optimization algorithms, coupled with adaptive hardware platforms, that together can adjust to the run-time conditions. We propose an optimization technique that minimizes the expected execution time of an application by dynamically scheduling hardware prefetches. We use a piecewise linear predictor in order to capture correlations and predict the hardware modules to be reached. Experiments show that the proposed algorithm outperforms the previous state-of-art in reducing the expected execution time by up to 27% on average.

No abstract

This paper presents an approach to minimize the average program execution time by optimizing the hardware/software implementation of error detection. We leverage the advantages of partial dynamic reconfiguration of FPGAs in order to speculatively place in hardware those error detection components that will provide the highest reduction of execution time. Our optimization algorithm uses frequency information from a counter-based execution profile of the program. Starting from a control flow graph representation, we build the interval structure and the control dependence graph, which we then use to guide our error detection optimization algorithm.