| D. Talla and L. K. John, "Cost-effective hardware acceleration of multimedia applications," Proc. IEEE Int. Conference on Computer Design, pp. 415-424, Sep. 2001. |
....Multimedia applications are increasingly becoming a dominant workload in modern computing as the Internet and mobile communication technologies are growing. These multimedia workloads are characterized by structured and regular computations on large data sets with small element widths [1]. Current industry support for multimedia application appears in multiple forms: ASICs, application specific processors (ASPs) multimedia extensions to general purpose processors, and multimedia co processors [2] ASICs while highly efficient for particular tasks, are not flexible enough to ....
.... performance improvement with architectural extensions such as Intel s MMX SSE, Sun Microsystems VIS, AMD s 3DNow , Motolora s AltiVec, DEC s MVI, HP s MAX2, and MIPS MDMX [3] 4] There have been analysis of media workloads on general purpose processors to identify their major features [1][2] 5] 6] 7] 8] 9] 10] It has been identified that these workloads have significant fine coarsegrained data parallelism, high instruction reference locality, high memory network bandwidth, continuousmedia data types, and extensive data reorganization. Many of the applications also need real time ....
D. Talla and L. K. John, "Cost-effective hardware acceleration of multimedia applications," In Proceedings of the IEEE International Conference on Computer Design '01, pp. 415-424, 2001.
No context found.
D. Talla and L. K. John, "Cost-effective hardware acceleration of multimedia applications," Proc. IEEE Int. Conference on Computer Design, pp. 415-424, Sep. 2001.
....situation. The speedup of a 2 way MediaBreeze enhanced processor with slip based prefetching is observed to be up to 28x over a 2 way SIMD GPP. We investigated the cost of incorporating the MediaBreeze hardware into a state of the art general purpose superscalar processor with SIMD extensions in [19]. Our results indicate that the newly added MediaBreeze hardware requires less than 10 of the SIMD execution units chip area and 0.3 overall chip area, and consumes less than 1 of the total processor power. This is achieved without elongating the critical path of the processor. 4. Summary and ....
D. Talla and L. K. John, "Cost-effective hardware acceleration of multimedia applications," Proc. Int. Conf. on Computer Design, pp. 415-424, Sep. 2001.
....1.5 Contributions This dissertation makes several contributions to the characterization of multimedia workloads, detection of bottlenecks, and explicit hardware support for accelerating media applications on SIMD GPPs. These are also described in more detail in [95] 96] 97] 98] and [99]. The summary of the contributions is listed below. 1. I perform a quantitative study of the execution characteristics of commercial multimedia applications on a state of the art superscalar processor. Memory access behavior, cache and branch behavior, and resource 16 usage are studied. It is ....
D. Talla and L. K. John, "Cost-effective hardware acceleration of multimedia applications," Proc. IEEE Int. Conf. on Computer Design, Sep. 2001, to appear.
No context found.
D. Talla, L. John, "Cost-effective Hardware Acceleration of Multimedia Applications," Proceedings of the International Conference on Computer Design, pp. 427-439, 2001.
No context found.
D. Talla, and L. K. John. "Cost-effective hardware acceleration of multimedia applications ". International Conference on Computer Design, Pages 415-424, September 2001.
Online articles have much greater impact More about CiteSeer.IST Add search form to your site Submit documents Feedback
CiteSeer.IST - Copyright Penn State and NEC