MGMT 1050 Chapter Notes - Chapter 15: Code Morphing Software, Very Long Instruction Word, Instruction Set
![](https://new-preview-html.oneclass.com/xBXoz56OAaVyNBo53RYPQ9n37kpbwJqR/bg1.png)
MGMT 1050 Chapter 15 Notes – Summary
Introduction
CPU features and enhancements
• Instructions are presented to the CPU for execution in 128-bit bundles that include a
group of three instructions plus 5 bits that identify the type of each instruction in the
bundle.
• An assembly language programmer is expected to follow a set of published guidelines
that identify dependencies and allow the parallel execution of each bundle.
• Additionally, bits within each instruction word provide information to the execution unit
that identify potential dependencies and other bottlenecks and help the programmer to
optimize the code for fast execution.
• High-level language EPIC compilers must also create code that satisfies the guidelines.
• A fundamental difference between the Transmeta VLIW and the Intel EPIC architectures
is the placement of responsibility for correct instruction sequencing.
• The VLIW architecture allows any sequence of instructions to enter the CPU for
processing.
• The code morphing software, integral to the architecture, handles proper sequencing.
• The EPIC architecture places the burden on the assembly language programmer or on
the program compiler software.
• This does not suggest that one architecture is superior to the other.
• It simply indicates a different approach to the solution of dependencies.
• Note that the Transmeta VLIW does not allow direct assembly language access to the
CPU.
• All program code must be processed through code-morphing software.
• Each architecture offers an interesting new approach to program execution with
potential benefits.
• We have already introduced you to the fundamental model of a traditional CPU,
represented by an instruction set, registers, and a fetch-execute instruction cycle.
find more resources at oneclass.com
find more resources at oneclass.com