Browsing by Subject "performance evaluation"

Now showing 1 - 5 of 5
  • Thumbnail Image
    Item
    Hardware and Compiler-Directed Cache Coherence in Large-Scale Multiprocessors
    (1997) Choi, Lynn; Yew, Pen-Chung
    In this paper, we study a hardware-supported, compiler-directed (HSCD) cache coherence scheme, which can be implemented on a large-scale multiprocessor using off-the-shelf microprocessors, such as the Cray T3D. The scheme can be adapted to various cache organizations, including multi-word cache lines and byte-addressable architectures. Several system related issues, including critical sections, inter-thread communication, and task migration have also been addressed. The cost of the required hardware support is minimal and proportional to the cache size. The necessary compiler algorithms, including intra- and interprocedural array data flow analysis, have been implemented on the Polaris parallelizing compiler [33]. From our simulation study using the Perfect Club benchmarks [5], we found that in spite of the conservative analysis made by the compiler, the performance of the proposed HSCD scheme can be comparable to that of a full-map hardware directory scheme. Given its comparable performance and reduced hardware cost, the proposed scheme can be a viable alternative for large-scale multiprocessors such as the Cray T3D, which rely on users to maintain data coherence.
  • Thumbnail Image
    Item
    Performance Evaluation of Media Losses in the Continuous Media Toolkit
    (1997) Wijesekera, Duminda; Parikh, Shwetal; Varadarajan, Srivatsan; Srivastava, Jaideep; Nerode, Anil
    Rapid growth of multimedia systems, and accordingly research in this area requires fast prototyping environments. The Berkeley Continuous Media Toolkit (GMT} is a popular environment that satisfies this need. Form a human user's perspective, in order for multimedia demonstrations to be comprehensible, the number of audio or video frames dropped and the timing delays in the ones that are displayed, need to be kept to a minimum. Therefore, it is important to know the frame dropping characteristics of CMT. In a series of experiments we rnonir.ored tlHi variation of thPse parameters with respect processor and network loads. It was obsprvrd that loads affen ap,gregau· frame drops at lower rates and consecutive frame drops at higher rates. Because at a higher rates a liarge number of consecutive frames are dropped, the ones that are played appear in a more timely manner. As a solution to observed problems, we present some QoS based approaches to control drop and delay parameters.
  • Thumbnail Image
    Item
    Performance Evaluation of Synchronization Losses in the Continuous Media Toolkit
    (1997) Wijesekera, Duminda; Parikh, Shwetal; Varadarajan, Srivatsan; Srivastava, Jaideep; Nerode, Anil; Foresti, Mark
    This paper presents a performance analysis of synchronization services provided by the Berkeley Continuous Media Toolkit (CMT). The quality of audio-video synchronization is measured against processor and network loads for both remote and local clients. The metrics of analysis are the perceptible and tolerable human perceptual limits reported by Steinmetz, and another metric designed to measure synchronization of lossy media streams. It is shown that according to Steinmetz' metric CMT provides imperceptible audio-video mis-synchronization for about 10 seconds, and tolerable synchronization for about 13 seconds from the start of the clips for local clients under low processor loads. It is also shown that under high loads, synchronization is achieved at the cost of losing media frames.
  • Thumbnail Image
    Item
    Performance Study of a Concurrent Multithreaded Processor
    (1997) Tsai, Jenn-Yuan; Jiang, Zhenzhen; Ness, Eric; Yew, Pen-Chung
    The performance of a concurrent multithreaded architectural model, called superthreading [15), is studied in this paper. It tries to integrate optimizing compilation techniques and run-time hardware support to exploit both thread-level and instruction-level parallelism, as opposed to exploit only instruction-level parallelism in existing superscalars. The superthreaded architecture uses a thread pipelining execution model to enhance the overlapping between threads, and to facilitate data dependence enforcement between threads through compiler-directed, hardwaresupported, thread-level control speculation and run-time data dependence checking. We also evaluate the performance of the superthreaded processor through a detailed trace-driven simulator. Our results show that the superthreaded execution model can obtain good performance by exploiting both thread-level and isntruction-level parallelism in programs. We also study the design parameters of its main system components, such as the size of the memory buffer, the bandwidth requirement of the communication links between thread processing units, and the bandwidth requirement of the shared data cache.
  • Thumbnail Image
    Item
    Validity and Reliability Evidence for an Experimental Performance Evaluation Instrument for Educational Speech-Language Pathologists
    (2018-02) Rentmeester Disher, Jill
    Performance evaluation for educators is intended to measure, develop, and support professional practices, and, in turn, improve student outcomes. To date, however, very little research exists to support the performance evaluation practices for non-classroom educators (Holdheide, Goe, Croft, & Reschly, 2010), such as educational speech-language pathologists (SLPs). Validity and reliability evidence for an experimental performance evaluation instrument specifically designed for SLPs was examined in this study. Study data were from 111 SLPs in a mid-size urban district who were evaluated one during an academic school year. The performance of the 111 SLPs was also described, so that any potential bias in the instrument could be examined. Results showed a restricted range of performance in which most SLPs were rated as proficient or exemplary on performance evaluation items. Some preliminary indications of bias were present, such that SLPs serving birth-five students, students with combined communication disorders, or students with moderate-severe disabilities were rated 6-10 points lower in total score (out of a maximum score of 108) compared to their colleagues who did not serve those populations. Construct validity analyses showed that the instrument's items were only loosely related to each other, although exploratory factor analyses did suggest an underlying structure of four domains. Face validity was gauged through optional perception surveys of the 111 SLPs in which the majority of survey participants felt the instrument items represented effective SLP practices. Finally, performance evaluation items demonstrating the highest evidence of reliability were related to an SLP’s planning of intervention; items with the lowest evidence of reliability were related to an SLP’s management of the session and rapport with students. Despite some limitations, it was concluded the SLP performance evaluation instrument showed initial evidence of being able to evaluate SLPs fairly, accurately, and with perceived credibility from the district SLPs.