Processor performance "squeezing toothpaste": Intel is considering energy efficiency? -Intel processor, process-IT information
Do you still interest areas of PC? If Yes, provide what you want the next generation PC processor performance or energy efficiency? To be honest, for those who are still concerned about the field of computer enthusiasts, more performance, lower power consumption or longer battery life between these topics, often for interesting discussions and debates, and from Intel and AMD that competitive era has already begun.
Today, AMD has apparently had no rhythm, so fans the most discussed development of Intel processors, and represents the future of desktop PC chip bellwether. However, there is no external competitive pressures, Intel seems to be difficult to make faster progress, the most obvious sign is that its 14-nanometer chips had broken the "tick" law extended for three years the time to rest on our laurels. Then, hopes of 10 nm and 7 NM Intel chips do?
Held in New York on Wednesday at a meeting of strategic decisions, Intel Chief Executive Officer bulaien·keziannike (Brian Krzanich) that Intel hopes to be released based on the 10 nm and 7 nm process each generation of chips can provide a stable 10% to 20% performance.
Many PC enthusiasts expressed disappointment about this, because in their opinion these numbers should be incremented, or Zen to be released AMD next generation processor architecture can so remarkable about it? AMD at Computex 2016 days ago once again confirmed that ZEN framework has brought about great improvements in terms of performance, as opposed to the "bulldozer" schema on each clock instruction set implementation rate rose more than 40%.
AMD make·Pei Cooper, Chief Technology Officer at that time masters (Mark Papermaster) that ZEN is unprecedented in history, a large performance increase. AMD Chief Executive Su Da MOM Lisa Su said when he took over this, ZEN highlights of processor architecture is "high performance".
New direction: energy efficiency and to cut costs?
Today's Intel, and AMD seem to have different goals, not only focus on laptops use battery life, and we hope that the PC be able to get higher performance at more affordable prices. Simply put, Intel's top priority is to make chips become smaller, while significantly reducing power consumption and manufacturing costs.
New York meeting, bulaien·keziannike said: "it is to me, I bring the Intel quest each transistor cost and no tile can provide performance, this is the real way, I believe, and will continue to do so. ”
It is clear that Intel is looking in a different direction to lead means, but it does contribute to the achievement of balance between the manufacturing cost and no performance, at least bulaien·kezianni that these changes can keep Intel in chip manufacturing, a leading opponent .
Intel has long made chips, and AMD is different, currently by chipmaker Globalfoundries to manufacturing, some also to Intel's rivals, including Samsung and TSMC.
" If you know anything about the other competitors, know what they say 10 nm for performance improvement is small, they would choose 7 nm to improve performance. And our plan in these technology nodes, each generation can 10% to 20% more consistent performance improvement . "Kezianni said.
Since 40 years, Intel has followed Moore's law as a guide to transistor development, its core ideas as a technological innovation. According to Moore's law, the number of transistors on integrated power 1 time every two years, and lower costs. But now more and more difficult to maintain, the strategy appears stalled, to design the chips more economical, hoping more advanced PC technology, while ensuring that costs more affordable.
Accurate to say, Intel seems to have started to consider new strategies that promote a technology process for three years, instead of every two years. But note that the fact that technological progress closer to the physical limits of semiconductor Silicon 10nm, making more difficult, the costs and risks are higher, Intel is now considering on the billions of dollars invested in the new plant, so spend more time is also reasonable.
Of course, Intel has confirmed that the first 10 nanometer chip called Cannonlake, will release later in 2017.
Technology process isn't important?
Tirias Research Chief Analyst jimu·maigeleige (Jim McGregor), Intel bosses hold different views on the performance improvements in chip. He said Intel had said "rival technology performance improvements will not much progress" on the right, but with the maturity of the technology process performance also improved significantly, only on the basis of advanced technology to support could stimulate the growth of the performance.
If proved, can do it directly with Intel's example. Originally Intel had just stepped into the 14-nanometer technology, performance and 22 nm, but as the technology matures, now 14-nanometer chip performance has been further uplifted. On the development of technology processes, Intel's competitors are struggling to pass, no surprise Samsung has the potential to be the first manufacturer that can be put into operation 10 nm, but TSMC said mass production in 10-nanometer chips by the end of this year.
However, every time chip technology process, performance will be steady circulation growth? Who maybe, possibly this ZEN architecture in 40% after upgrade can't do it? After all this time AMD ZEN new schema upgrade is not just 14-nm process as simple, accompanied by improvement of complex 3D transistor structure, these are long-term AMD not fundamentally improve key projects.
In addition, in order to gain a performance advantage 7-nanometer process nodes, most chip manufacturers are betting more advanced manufacturing techniques, such as EUV (extreme ultraviolet lithography) technology, which helps to eliminate the complex problems of the chip manufacturing process, to create a chip more sophisticated. Intel is 7 nm is a major upgrade, process description the process is still a big challenge for all major manufacturers .
In terms of chip technology process, Intel has for many years ahead, unforgettable time and time again there are so many breakthroughs, such as the 2007 made a transistor gate dielectric high-k material, which is the biggest change in transistor technology since 1960, claims to ensure that Moore is extending its 10-year, and 2011 3D transistors Intel brilliance into the 22 nanometer era.
In short, Intel also focused on the development of the semiconductor industry as a whole, believe Intel has its own technology once again, in order to ensure that in the process process leads. Bulaien·kezianni at last week's meeting also said that "at least two process nodes, we know how to stay ahead."
处理器性能“挤牙膏”：英特尔是更考虑注重能效？ - 英特尔,处理器,制程 - IT资讯
你还在热情关注PC领域吗？如果是，你希望下一代PC处理器提供什么样的性能或能效？说实话，对于那些还在关注电脑领域的爱好者，在更强的性能、更低的功耗或更长的使用续航时间这些话题之间，常常进行各种有趣的讨论乃至争论，而且从最初英特尔与AMD那个相互竞争的年代就已经开始了。
今天，AMD显然已经不再节奏上了，所以爱好者讨论最多的英特尔处理器的发展，并认为代表了未来桌面PC芯片的风向标。不过，没有外部竞争的压力，英特尔似乎很难逼自己更快的进步，最明显的标志就在于，其14纳米的芯片竟然打破了的“嘀嗒”定律延续三年时间都在吃老本。那么，被寄予厚望的英特尔10纳米和7纳米芯片呢？
上周三在纽约召开的一次战略决策会议上，英特尔首席执行官布莱恩·科兹安尼克（Brian Krzanich）表示，英特尔希望即将发布基于10纳米和7纳米工艺制程的每一代芯片能够提供稳定10%到20%的性能提升。
对此很多PC爱好者表示失望，因为在他们看来这些数字应该是递增的，要不然AMD即将发布的Zen新一代处理器架构怎么会如此备受瞩目呢？几天前在Computex 2016上AMD再一次确认，ZEN架构在性能方面带来了巨大的改进，相对于“推土机”架构每时钟指令集执行率上涨幅达到了40%以上。
当时 AMD首席技术官马克·佩珀马斯特（Mark Papermaster）表示，ZEN是历史上从未有过的一次性能大提升。而AMD的首席执行官苏大妈Lisa Su也接过此话表示，ZEN处理器架构的最大亮点就是“高性能”。
新方向：削减成本注重能效？
如今的英特尔，似乎与AMD有着不同的目标，不仅更加专注于笔记本电脑的使用续航时间，而且还希望PC能够以更实惠的价格得到更高的性能。简单的说，英特尔的当务之急就是努力让芯片变得更小，同时大幅降低功耗和制造成本。
纽约会议上，布莱恩·科兹安尼克说：“对我来说，我带来的英特尔追求每晶体管的成本和没瓦能够提供的性能，这才是真正的方向，我相信并会继续这么做。”
很显然，英特尔正在寻求通过不同的方向来领先对手的手段，不过这的确有助于实现制造成本与没瓦性能之间的平衡，至少布莱恩·科兹安尼认为这些变化可以让英特尔持续在芯片制造优势上领先对手。
英特尔长期以来自己制造生产芯片，而AMD则有所不同，目前由合作的芯片制造商Globalfoundries改工生产，部分还交给英特尔的其他竞争对手，包括三星和台积电。
“如果你对其他竞争对手有所了解，也知道他们说10纳米对性能改进的余地很小，他们会选择在7纳米全力改进性能。而我们的计划是在这些工艺节点上，每一代都能有10%至20%较为一致的性能改进。”科兹安尼说。
40年来以来，英特尔一直遵循摩尔定律作为晶体管发展的指导，将其作为技术创新的核心思想。按照摩尔定律，集成电量上的晶体管数量每隔两年翻一倍，并且成本更低。不过，现在来看越来越难维持，该战略出现了停摆，反而更多地设计芯片制造的经济型，希望从中融入更多高级的PC技术，同时确保成本更加低廉。
准确的说，英特尔似乎已经开始考虑三年推进一次工艺制程的新策略了，而不再是两年一更。但需要注意一个事实，工艺进步到10nm之后越接近硅基半导体的物理极限，制造难度更大，成本和风险也更高，英特尔现在考虑以数十亿美元投资到新工厂上，所以花更多时间也在情理之中。
当然了，英特尔已经确认第一款10纳米代号为Cannonlake 的芯片，肯定会在2017年下半年发布。
工艺制程不重要了？
Tirias Research首席分析师吉姆·麦格雷戈（Jim McGregor）,对英特尔大佬在芯片性能改进上持有不同的意见。他表示，英特尔所说的“竞争对手工艺进步性能改进也不会很大”论没错，但是随着工艺制程的成熟性能同样得到明显提升，只有以先进的工艺支持为基础才有可能刺激性能的增长。
若以事实证明，直接拿英特尔就能做例子。最初英特尔刚刚踏入14纳米的工艺，性能上其实仍与22纳米差不多，但是随着工艺的成熟，如今14纳米芯片的性能得到了更高的提升。所以关于工艺制程的发展，英特尔的竞争对手都在奋力超车，不出意外的话三星有可能是第一家可投产10纳米的制造商，不过台积电表示，今年年底10纳米芯片就可以量产。
但是，芯片工艺制程的每一次提升，性能就一定会稳定循环增长？谁也说不定，有可能这一次ZEN架构40%的提升之后就无法做到了呢？毕竟这一次AMD的ZEN架构不只是升级14纳米新工艺那么简单，还伴随着复杂3D晶体管结构的改进，这些都是此前长期AMD没有从根本上改进的重点工程。
另外，为了在7纳米工艺结点上取得性能优势，现在大多数芯片厂商都在押注更先进的制造技术，比如名为EUV（极紫外光刻）的新技术，该技术有助于消除芯片制造过程的复杂问题，制造出的芯片更为精细。英特尔之所以时候7纳米是一个重要的升级，说明该工艺制程目前还是各大制造商的一大挑战。
在芯片工艺制程方面，英特尔曾多年领先，让人难以忘怀的一次又一次重大突破实在太多，比如2007年的采用high-k材料制作晶体管闸极电介质，这是自1960年代以来晶体管技术的最大变革，号称可确保摩尔定律再延续10年，还有2011年3D晶体管让英特尔辉煌的走进22纳米时代。
总之，英特尔肯定还在关注整个半导体行业的发展，相信英特尔有自己的技术以确保再一次在工艺制程上领先。布莱恩·科兹安尼在上周的会议上也提到，“至少在未来两个工艺节点上，我们知道如何保持领先地位”。