Effective Compiler Support For Predicated Execution Using The Hyperblock

Abstract

Predicated execution is an effective technique for dealing with conditional branches in application programs. However, there are several problems associated with conventional compiler support for predicated execution. First, all paths of control are combined into a single path regardless of their execution frequency and size with conventional if-conversion techniques. Second, speculative execution is difficult to combine with predicated execution. In this paper, we propose the use of a new structure, referred to as the hyperblock, to overcome these problems. The hyperblock is an efficient structure to utilize predicated execution for both compiletime optimization and scheduling. Preliminary experimental results show that the hyperblock is highly effective for a wide range of superscalar and VLIW processors. 45 0-8186-3175-9/92 $3.00 0 1992 IEEE

A Generator for Time-Predictable Code

“…Their motivation is to make these loops suitable for software pipelining, a technique to speed up loop executions by interleaving iterations, which requires a linear sequence of instructions without control-flow changes. Their algorithm provides the basis for more elaborate techniques like hyperblock formation [18], which does not if-convert the whole loop body but only a subset of basic blocks. The blocks are merged into singleentry multiple-exit regions of predicated instructions, the socalled hyperblocks.…”

Hardware realization of an FPGA processor — Operating system call offload and experiences

“…It is often used in SIMD engines such as graphics processors to allow efficient software pipelinening of loops [24]. Mahlke et al show that the use of hyperblock scheduling on a fully predicated instruction set results in an average speedup of 72% on an 8 issue processor [25] Allen et al have demonstrated how control dependence in the instruction stream can be translated into data dependence using an if-conversion pass in the compiler back end [26].…”

Exploiting compiler-generated schedules for energy savings in high-performance processors

“…An S-Region could be a basic block or a group of basic blocks such as loops, hyperblocks [11], superblocks [19] or subroutines. There are several criteria that must be considered before a region can selected for issue in the static mode: 4.…”