next up previous contents
Next: 7.1 Trace Scheduling Up: Very Large Instruction Word Previous: 6 The Transmeta Crusoe   Contents

7 Scheduling Algorithms for VLIW

The difficulty of programming VLIW processors by hand should be evident even from the simple Defoe programming examples. One reason programming VLIWs is more difficult than writing code for a super scalar processor is that the program for a super scalar processor is inherently sequential and it is left to the hardware to extract parallelism from the sequential program. On the other hand, when generating code for a VLIW processor, the assembly language programmer or the compiler is faced with the task of extracting parallelism from a sequential algorithm and scheduling independent operations concurrently. For this reason, instruction scheduling algorithms are critical to the performance of a VLIW processor. We next describe three important scheduling algorithms starting with the classic trace scheduling algorithm.



Subsections

Binu K. Mathew