Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Politics : Formerly About Advanced Micro Devices -- Ignore unavailable to you. Want to Upgrade?

To: grok who wrote (59988)	5/29/1999 4:47:00 PM
From: grok	Respond to of 1574386

Yet again more. At this link you can see what they are working on: crhc.uiuc.edu Here is one sample: "Advanced Compiler Technology and Architecture Techniques for Instruction-Level Parallel Processing W.-M Hwu*, D. Lavery, J. Gyllenhaal, S. Hwu, D. August, T. Johnson, B. Deitrich, C.-H. A. Hsieh, D. Connors, B.-C. Cheng, J. Braun, L.-C. Hsu, L.-C. Wu, K. Safford, M. Conte, M. Merten, M. Trommer Hewlett-Packard Company, Intel Corporation The objective of the project is to develop IMPACT, a new compiler foundation for high performance instruction-level processing computer systems. The project is motivated by the fact that existing commercial compilers have not been able to absorb the advanced research results due to the limitations of their software architectures and their internal program representation. It is also motivated by the the lack of effective compilers for parallel systems. The compiler technology studied under this project involves predicated compilation model, machine description technology, control and data speculation compilation models, software pipelining techniques, region compilation, memory dependence handling, debugging parallelized code, height reduction techniques, inlining, and profiling. A large number of experimental computer architecture research projects are also being supported by IMPACT, including predication architectures, control and data speculation architectures, wide issue architectures, and run-time optimization architectures. The project has contributed to the technology base of several important products at the sponsoring companies." Guess who the sponsoring companies are? Hint: you can read their names above. (I do like their modesty as shown in the last line.) There have been several posters on this thread who claim that compilers cannot do {fill something in here} and often cloud up the issue by throwing in VLIW or even dataflow. But the real truth is that modern compiler technology when coupled with a properly designed new instruction set will soon obliterate the older notations of what is possible.