عنوان اصلی نسخه لاتین: A Hybrid Real-Time Scheduling Approach for Large-Scale Multicore Platforms
عنوان اصلی فارسی : یک روش زمانبندی بلادرنگ ترکیبی برای پلت فرمهای چند هسته ای
مرتبط با رشته های: فناوری اطلاعات و کامپیوتر
فرمت فایل ترجمه شده: ورد آفیس(امکان ویرایش)
تعداد صفحات فایل ترجمه شده: 19
جهت دانلود رایگان مقاله انگلیسی این مقاله اینجا کلیک کنید
ترجمه ی سلیس و روان مقاله آماده ی خرید می باشد.
_______________________________________
چکیده ترجمه:
در
این مقاله قصد داریم روشی را برای زمانبندی وظایف بلادرنگ بر روی پلت
فرمهای چند هسته ای و بزرگ مقیاس، و با استفاده از کشهای سلسله مراتبی
اشتراکی ارائه دهیم. در این روش، ۱ پلت فرم چند هسته ای در داخل کلاستر یا
خوشهها بخش بندی شده است. وظایت نیز به صورت پویا به این خوشهها تخصیص
داده شده و در داخل هر خوشه با استفاده از الگوریتم زمانبندی EDF انحصاری
زمانبندی میشوند. نشان دادهایم که این روش ترکیبی در بخش بندی و زمابندی
میتواند نسبت به پلت فرمهای بزرگ مقیاس عملکرد بهتری داشته باشد. همچنین
اندازهی مناسبی را برای خوشه در نظر گرفتهایم تا بتوانیم به بهترین
کارائی ممکن دست پیدا کنیم، البته با این شرط که مشخصه های ۱ مجموعه وظیفه
را بتوان پشتیبانی کرد.
واژگان کلیدی: زمانبندی بلادرنگ، EDF انحصاری، خوشه بندی
جهت دانلود محصول اینجا کلیک نمایید
قسمتی از مقاله انگلیسی
1 Introduction
Multicore architectures, which include several processors on a single chip, are being widely touted as a solution to the ìthermal roadblockî imposed by single-core designs. Most chip makers have released dual-core chips, and a few designs with more than two cores have been released as well. For instance, both Intel and AMD have released four -core chips, Sun recently released its eight-core Nia- gara chip, and Intel is expected to release chips with 80 cores within ve years [6]. Azul, a compan y that builds Java appliances, has created 48-core chips that are used in systems with up to 768 total cores [1]. These appli- ances are used to process lar ge numbers of transactions with soft real-time requirements. To summarize, lar ge- scale multicore platforms with tens or even hundreds of cores per chip may become a reality fairly soon and ap- plications with (soft) real-time constraints will lik ely be deplo yed on them. In this paper , we consider the issue of how to efciently schedule soft real-time workloads on such lar ge platforms.
Multicore architectures, which include several processors on a single chip, are being widely touted as a solution to the ìthermal roadblockî imposed by single-core designs. Most chip makers have released dual-core chips, and a few designs with more than two cores have been released as well. For instance, both Intel and AMD have released four -core chips, Sun recently released its eight-core Nia- gara chip, and Intel is expected to release chips with 80 cores within ve years [6]. Azul, a compan y that builds Java appliances, has created 48-core chips that are used in systems with up to 768 total cores [1]. These appli- ances are used to process lar ge numbers of transactions with soft real-time requirements. To summarize, lar ge- scale multicore platforms with tens or even hundreds of cores per chip may become a reality fairly soon and ap- plications with (soft) real-time constraints will lik ely be deplo yed on them. In this paper , we consider the issue of how to efciently schedule soft real-time workloads on such lar ge platforms.