http://www.eecs.berkeley.edu/Pubs/TechRpts/2008/EECS-2008-23.pdf
Intel/MS가 공동 투자한 버클리의 병렬 컴퓨팅 랩에서, 올해 발간한 향후 연구 아젠더 아티클. 그 중에서 특히 눈길을 끄는 대목은:
This switch to multicore breaks the 50-year tradition of how to program computers. Since the dawn of computing, many have tried to replace a single large processor with many small processors, but with little success. For example, there has been a nearly 100% failure rate of parallel computing startups: Convex, Encore, MasPar, nCUBE, Kendall Square Research, Sequent, Inmos, and Thinking Machines all bet on parallelism and are no longer in the marketplace.
이 목록들을 보고 있으니, 겁이 덜컹 나긴 한다. ^^
