Moore's Law is generally taken to mean that the number of transistors on a chip -- and by extension, processing power -- doubles every two years. In reality, Gordon Moore, the computer scientist who originated Moore's Law in 1965, was talking about the economic costs of chip production and not the scientific achievements behind advances in chip design.
摩尔认为,芯片产量的成本将在未来10年内每年减半,但之后可能无法可持续[来源:希金斯]。然后,可以从经济上而不是科学地达到摩尔定律的极限。
几位著名的计算机专家认为,摩尔的法律不能持续二十年以上[来源:IEEE]。为什么摩尔的律法注定要失败?因为随着晶体管变小,芯片的生产变得更加昂贵。
一位分析师预测,到2014年,晶体管的大小将是20纳米,但芯片大小的任何进一步降低都会太昂贵,对于群众生产而言[来源::Nuttall]。
For comparison, as of summer 2009, only Samsung and Intel have invested in making 22-nanometer chips.
生产这些芯片的工厂耗资数十亿美元。GlobalFoundries的Fab 2工厂将于2012年开始在纽约开始生产,建造成本为42亿美元。很少有公司拥有这种资源,英特尔表示,一家公司必须拥有90亿美元的年收入才能在尖端的芯片市场上竞争[来源:Nuttall]。
同样的上述分析师认为,在投资新的,更昂贵,更小的芯片设计之前,公司将尝试从当前的技术中获得最大收益[来源:Isuppli]。So while the end of Moore's Law may limit the rate at which we add transistors to chips, that does not necessarily mean that other innovations will prevent the creation of faster, more advanced computers.