哪位朋友谈谈“中国芯”?

  • 主题发起人 主题发起人 cobra
  • 开始时间 开始时间
最初由 Kent以东首帅哥 发布
再过15年, 做到0.02 micron的时候就差不多是极限了.

难道摩尔定律15年后就不会成立了吗,
 
最初由 spacespace 发布


0.13u with 38M gates, that is too much. I am not saying that is impossible, I am just saying it is too hard. How big the chip going to be, I am not sure if there is any package can fit into it. And how much power it going to consume, must like a heater. 0.09u, there is too many unknow factors, when desing move toward deep sub-micron, a lot of things people do not know. Everyone is learning when they move forward.

Yeah, most Cisco designs are under 0.13u, but does not mean that is cheaper to them. The reason they have to go with 0.13u is that they require high speed IO, DDR RAM and other high speed macro support which is only available in 0.13 technology. The price to Cisco is definite not cheaper. Actually IBM or TI take the price advantage of the technology, they can make more money, not Cisco. But all those based on one assumption, high volume. The fact is that in the high-end ASIC cummunication area, the volume is really low, low volume means low profit, or even losing money.

At the begining of the project, everyone will claim high volume to bargain for the price. But after the project is done, you will find they only want 1000 parts per year. How many wafers for 1000 parts? probably just 3 wafers. That is why not many company would like to support high-end customers, like Cisco. Do u work for Cisco?

看来老兄在理论上很有研究呀。38Mgates的确很难,但是实际上现在IBM0.13就是可以做到了。
另外,锤泥春说得对,象cisco这样的客户是最受欢迎的,事实上cisco就是IBM和TI最受欢迎的客户,对这样的客户真是一切优先,有求必应呀。
 
最初由 spacespace 发布


难道摩尔定律15年后就不会成立了吗,

之后城里人会把片子做大,做成三纬的,还能撑十年八年,半导体的极限才来临.
 
最初由 Kent以东首帅哥 发布


之后城里人会把片子做大,做成三纬的,还能撑十年八年,半导体的极限才来临.

到那是纳米技术和超导技术开始成熟,所有的芯片和现在的概念几乎不同。也就是是说,现有的IT技术通通过时。而机械技术等传统技术仍然存在。
 
后退
顶部