【晶片故事之三】晶片的行業之困難〈蕭若元:理論蕭析〉2018-04-26 -  免费在线视频最佳电影电视节目 -  Viveos.Net

【晶片故事之三】晶片的行業之困難〈蕭若元:理論蕭析〉2018-04-26

Share
Embed
  • Loading...
  • Published on:  Friday, April 27, 2018
  • 即時聊天室:http://goo.gl/ToDqof
    謎米香港 www.memehk.com
    Facebook:www./memehkdotcom
  • Source: https://youtu.be/YMkN2uv5ZSM
Loading...

Comment

  • Pun yung Law

    Pun yung Law

     10 months ago

    不過亦係盜版成就M S,唔係成萬銀一套点買,M S扮嘢鬧緊老翻唯害,其實心笑緊,唔係点解W10免費。

  • Pun yung Law

    Pun yung Law

     10 months ago

    係大陸叻老翻,唔係量產令價格低,係盗版沒成本

  • Savio Chau

    Savio Chau

     a years ago

    其實大型高密度瞐片嘅關鍵係e-beam lithography 同設計工具呢兩種技術,越大型嘅芯片,晶体管數量越多,晶体管面積越細,e-beam lithography machine 就要越精確。設計經驗固然重要,但最緊要係有好嘅設計工具,現在晶片設計都係用Verilog,OpenCL呢一類電腦語言寫成,同設計軟件差不多,寫完程式再經過 simulation,synthesis,back annotation,再simulate 然後就可以fabricate。除非大量生產嘅晶片先至用人手將面積optimize,但即使如此,都會盡量用已經測試過嘅大型模塊合成,咁先至可以減低出錯機會,要一次過製片成功嘅關𨫡係simulation,模擬越準確,first silicon success 嘅機會越大,而精確嘅模擬,就要有好嘅設計工具,因此中國要趕上晶片技術,必須先要明白晶片設計嘅關鍵技術

  • Darkee Wang

    Darkee Wang

     a years ago

    分析有理,中国工程师有点小聪明

  • Lai Ranky

    Lai Ranky

     a years ago

    Mark仔搞搞個白平衡,其實好簡單但係視覺上好好多

  • 詠春盲 詠春盲

    詠春盲 詠春盲

     a years ago

    又係到扮专家,騙流量

  • enaungong

    enaungong

     a years ago +1

    只要放棄一黨專政,就可以做出領先晶片。

  • fire fire

    fire fire

     a years ago

    請報中央牌手機 、全部諮詢一致、 共產黨台 、合符中國特色

  • Chin Kang

    Chin Kang

     a years ago

    蕭先生間接宣佈中國永遠落後。

  • Raymond Chan

    Raymond Chan

     a years ago

    聯發科

  • Hoi Lui Lee

    Hoi Lui Lee

     a years ago

    燒山你錯喇,Intel 以前有粒叫 Xscale 嘅 CPU 行 Strong ARM 架構。

    https://zh.wikipedia.org/wiki/XScale

    而且早幾年 Intel 亦有粒 Atom Z series SOC用係流動產品,如 ASUS Zenphone 2 及 Tag Heuer Connected Modular 45等。行 x86架構。

  • fang sir

    fang sir

     a years ago

    联发科

  • 李丹

    李丹

     a years ago

    我在惠州都没得吃这种粽

  • kin chuen Chan

    kin chuen Chan

     a years ago

    當年列根用星球大戰計劃,今天就用晶片計劃?

  • KIU WAI LAM

    KIU WAI LAM

     a years ago +4

    蕭生:台灣最大晶片廠係台積電 TSMC ,第二應該係聯華電子 UMC,聯發科技 MediaTek 係大,不過佢係 Fabless ,只做晶片設計。另外台灣另一間晶片設計廠叫「 細 M」Mstar,已經畀「大 M」MediaTek 買了股份了。

  • Alex Ford

    Alex Ford

     a years ago +3

    中国工程师收入吴低,几万人仔一个月闲闲地

  • Ronald Au

    Ronald Au

     a years ago

    老蕭 D 英文發音認真麻撚麻 !

  • Silicon AI

    Silicon AI

     a years ago +1

    For the host: 1) Moore's law is not a law but simply a trajectory prediction that the "number" of cmos xtors can double every two years largely based on the prevailing electro-optics properties of the Silicon and it's optimal doping level (originally 18 months at the beginning becoming 2 years after Pentium time frame). Doubling the number of xtors DOES NOT mean doubling the computing power -- the correlation is not linear. Computing power and its measure is a separate topic 2) The difficulty of getting into semiconductor biz is largely driven by the dynamics of economies of scale, a balance between downward compatibility and upward innovation with multiple technology inflections in the market, deflation (e.g. price drop) is only a reactive biz move by trade-off curve driven by the power/speed curve 3) 7nm from Intel ~7 years ago is different than the 7nm from TSMC just announced, you need to have a PhD degree to understand the differences 4) 7nm even with finfet and below is hitting the electron gate leakage limit so the economies of scale will be wrapped by electro-optic's physics limits. ROI of 7nm and below is being examined. Parallel R&D effort is in material science for Quantum chip fab - we just showcased one in CES last year 4) Chinese super computer is ranked at the top in the world. If you are still curious about semiconductor TD stuff and biz, you can visit the Intel museum and the tech museum of technology if you happen to be in Silicon valley.

  • cookie chan

    cookie chan

     a years ago +4

    中共駛左五仟億都只係做到發聲芯片

  • 趙日天

    趙日天

     a years ago

    我吊你老母,是聯科發不是聯想發