Dynamic translation of tree-instructions into VLIWs

We describe an approach and its implementation for the dynamic translation of programs
represented as sequences of tree-instructions into very-long instruction words (VLIWs)
suitable for execution on wide-issue processors. Tree-instructions are characterized by not
reflecting the organization of any processor implementation, whereas VLIWs incorporate
implementation-specific features. Tree-instructions are dynamically translated, during
instruction cache reloading/accessing, to fit processor implementations constraints. Such
a translation is performed without the need for checking dependencies between primitive
instructions because tree-instructions consist of operations executable simultaneously.
This approach allows executing the same object-code and delivers scalable performance
on implementations with varying resources, including superscalar and VLIW-based ones,
thus solving one of the most critical objections to the adoption of the VLIW paradigm in
general-purpose architectures.

By: Jaime H. Moreno

Published in: RC20505 in 1996

LIMITED DISTRIBUTION NOTICE:

This Research Report is available. This report has been submitted for publication outside of IBM and will probably be copyrighted if accepted for publication. It has been issued as a Research Report for early dissemination of its contents. In view of the transfer of copyright to the outside publisher, its distribution outside of IBM prior to publication should be limited to peer communications and specific requests. After outside publication, requests should be filled only by reprints or legally obtained copies of the article (e.g., payment of royalties). I have read and understand this notice and am a member of the scientific community outside or inside of IBM seeking a single copy only.

8193.ps.gz

Questions about this service can be mailed to reports@us.ibm.com .