Access the full text.
Sign up today, get DeepDyve free for 14 days.
References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.
A VLIW ARCHITECTURE FOR OPTIMAL EXECUTION OF BRANCH-INTENSIVE LOOPS Bogong Suf, Wei Zhao$, and Zhizhong Tang Lkpt.dc! mpwrscimx T~inglnm Univmity Beijii loo084 Chim Stanley Habib CcmpucrsckmcDcp. Tbz City univcnity EMnil: ThcGradmleSdxoldUdwnityChtu of New York, 33 W. 42 St., New Yak, kLu.l~UNYvM.cUNY.Ew FAX: 212-642-1902 NY lo036 Tel: 212-642-2201 ABSTRACT We propose a VLIW architectural model for optimal execution of branch-intensive loops as well as a new single-chip architecture URPR-2 for digital signal and image processing based on this model. In this architecture the instructions belonging to different iterations and different paths can be executed simultaneously. Instruction level parallelism can be exploited in a wider scope as multi-branching can be processed in each machine cycle. A mechanism called the pipeline control blackboard(PCBB) is also proposed to support conditional branches. The URPR-2 can not only execute loops with basic block at high speed but also can run loops with conditional branches at a cost of reduced time and space occupancy. Many approaches have been developed to solve the branch problem at compile-time. They may obtain very good results for those loops with sparse branches, sometimes their parallelism can reach 90 operations per cycle[Nicolau84]. But they are poor for the branchintensive loops,
ACM SIGMICRO Newsletter – Association for Computing Machinery
Published: Dec 10, 1992
Read and print from thousands of top scholarly journals.
Already have an account? Log in
Bookmark this article. You can see your Bookmarks on your DeepDyve Library.
To save an article, log in first, or sign up for a DeepDyve account if you don’t already have one.
Copy and paste the desired citation format or use the link below to download a file formatted for EndNote
Access the full text.
Sign up today, get DeepDyve free for 14 days.
All DeepDyve websites use cookies to improve your online experience. They were placed on your computer when you launched this website. You can change your cookie settings through your browser.