IMPACT Papers with Impact!

• (1992) The Superblock: An Effective Technique for VLIW and Superscalar Compilation

• (1991) IMPACT: an architectural framework for multiple-instruction-issue processors

The IMPACT Group has become a major source of advanced compiler technology for the U.S. microprocessor industry. Our work on superblocks and hyperblocks has become part of the technology base of new compilers in major corporations such as Sun Microsystems, Intel, Hewlett-Packard, IBM, AMD, Lucent, and Motorola, all of whom have licensed our IMPACT compiler for technology transfer. And when Intel and HP announced their IA-64 (Itanium) product line in 1997, they publicly acknowledged their use of IMPACT in deriving critical early-performance results for the new architecture. More recently, the superblock techniques have been incorporated into GCC.

About Us - The Past

The Instruction-Level Parallel Processing Era

The IMPACT research group was established in 1987, when the rapid increase in hardware execution resources had created pressing needs for instruction-level parallelizing compilers. Many well-known researchers, using the output of the then mainstream compilers, announced very pessimistic projections of the amount of instruction-level parallelism available to future microprocessors. If not addressed, this shortfall in compiler technology could seriously limit the long term growth of microprocessor performance. Wen-mei Hwu and his students rose to the challenge by constructing a revolutionary compiler infrastructure called IMPACT. And we became known as the IMPACT Group.

IMPACT (Illinois Microarchitecture Project Utilizing Advanced Compiler Technology) used engineering prototypes to show that compilers can generate efficient code with far more parallelism than anyone ever thought possible. Wen-mei and his students published seminal papers on the "superblock" and "hyperblock" structures, which enable a compiler to parallelize code across complex control structures through a clever combination of code replication and code predication. When combined with aggressive function inlining and memory dependence analysis, algorithms based on superblocks and hyperblocks increased parallelism in the output code and maintained efficiency.

Our group's work with instruction-level parallelism, also known as Explicitly Parallel Instruction Computing (EPIC), spawned generations of new research in program analysis, recognizing that the compiler must have comprehensive knowledge about program control and memory access to make the sweeping transformations necessary for high performance. During this era, we published many papers on various types of analysis, including highly accurate and highly efficient predicate analysis and pointer analysis. This work, combined with the application knowledge gained from years of work in ILP, has led naturally to the IMPACT Group's current research, in which we are defining sweeping program transformations for ultra-efficient computing.

A series of national awards recognized the tremendous technical contributions by the grad students of the IMPACT Group. Check the 2004 ACM/IEEE International Symposium on Computer Architecture paper by Sias et al. for a critical evaluation of the technical contributions of the IMPACT Group during the instruction-level parallel processing era.