谷歌自研芯片Tensor正式亮相：5nm，8核CPU，支持手機運行AI模型

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"北京時間 10 月 20 日凌晨，Google 2021年秋季硬件發佈會如期而至。此次的發佈會上並沒有嚮往年一樣發佈多款硬件產品，而是主推搭載了其首款自研芯片Tensor的旗艦手機Pixel 6系列，此外，谷歌歷時四年研製的芯片Tensor的一些細節也正式對外公佈。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"谷歌官宣首款自研Tensor芯片"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"發佈會上的兩款硬件Pixel 6和6 Pro內部都配備了Google Tensor處理器，就像蘋果的 A 系列芯片一樣，它試圖將谷歌的軟件與一些定製的硬件結合起來。該芯片由谷歌研究院設計，被谷歌譽爲“機器學習里程碑”。早在今年夏天，谷歌首次談到了Tensor，它爲谷歌的定製硬件提供人工智能支持。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"據悉，Tensor芯片採用5nm，CPU爲八核設計，分爲大中小三核——兩個以2.8GHz運行的ARM Cortex-X1內核、兩個以2.25GHz運行的Cortex A76內核，以及四個1.8GHz A55內核。GPU爲20核設計。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"不過，我們從這些早期泄露的信息中知道的是，Tensor由兩個Arm Cortex-X1芯片驅動，這是"},{"type":"link","attrs":{"href":"https:\/\/www.infoq.cn\/article\/EVU8dfiSkTZH3p0oIVdR","title":"xxx","type":null},"content":[{"type":"text","text":"Arm"}]},{"type":"text","text":"面向性能的旗艦移動設計。相比之下，爲三星 Galaxy S21 和許多其他高端手機提供動力的高通旗艦驍龍 888芯片只有一個 X1 內核。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"與現代SoC的慣例一樣，Tensor處理器也設計性能較低、節省電池的內核。新的 SoC 允許 Pixel 6 通過實時翻譯功能快速翻譯視頻和消息，並且在識別聲音方面也會更加智能。在使用語音輸入、編輯和發送消息時，這將會特別有用。最重要的是，它將能夠在不消耗太多電池壽命的情況下完成所有這些工作。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"此外，谷歌表示 Tensor 還爲 "},{"type":"link","attrs":{"href":"https:\/\/www.infoq.cn\/article\/wjmYeU0bfwIxTSJTav0j","title":"xxx","type":null},"content":[{"type":"text","text":"Pixel 6"}]},{"type":"text","text":" 提供了額外的安全層。它將與手機中的 Titan M2 芯片配合使用，以防止惡意軟件和其他潛在攻擊。這對谷歌來說是一個很好的進步，我們希望在其他 Android 手機中看到更多的安全硬件。據谷歌稱，總體而言，Tensor 芯片的性能將比 Pixel 5 快 80% 左右。"}]},{"type":"heading","attrs":{"align":null,"level":2},"content":[{"type":"text","text":"蘋果在前打樣，谷歌加緊追趕"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在谷歌攻堅手機市場的那些年裏，它分別收購了 HTC 和 Fitbit，雖然它將計算理念推向了新的高度，但卻幾乎完全沒有獲得有意義的智能手機市場份額。此外，想起命運多舛（但極具創造力）的Moto X，也不禁讓人唏噓。儘管 Pixel 具有影響力，但頂級 Pixel 並沒有大賣——而較便宜的低端機的表現也只是好一點。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"此前的多番嘗試都沒有掀起大的水花，那麼谷歌爲何下決心要推出自研芯片呢？因爲對手已經跑很遠了，再不行動就真的追趕不上了。作爲谷歌的競爭對手之一，蘋果早已在其iPhone上使用A系列自研芯片，目前已迭代到了A14。而且，"},{"type":"link","attrs":{"href":"https:\/\/mp.weixin.qq.com\/s?__biz=MzU1NDA4NjU2MA==&mid=2247488150&idx=1&sn=7dae5483d82b63c1eebe0899971a6924&chksm=fbe9ab59cc9e224fcab84a6688047ceb21b168d3b06f60ab2e6140c94e176a7736f16ad9be15&scene=27#wechat_redirect","title":"xxx","type":null},"content":[{"type":"text","text":"蘋果"}]},{"type":"text","text":"從去年開始在Mac上使用M1自研電腦芯片，並計劃在明年底完成過渡。谷歌感受到了危機。今年夏天，谷歌硬件業務負責人Rick Osterloh在接受採訪時稱，谷歌要利用這顆自研的芯片徹底改變智能手機，爲未來AR和AI等新技術的應用打下重要基礎。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Osterloh確信，AR將在未來的計算中發揮重要作用，儘管他並不十分清楚這會在何時、以怎樣的方式發生。谷歌之所以推出自研芯片，一定程度上是因爲AR和AI等前瞻性技術依舊對手持設備構成重大計算挑戰，因此谷歌進行了大膽嘗試，利用Tensor芯片來進行補救。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Tensor 似乎在新的移動芯片中提供了我們想要的一切：快速的速度和大量具有前瞻性的 AI 功能。如果谷歌真的成功了，也許 Tensor 可以應用到其他公司的設備上。Osterloh稱，這款芯片十分強大，它甚至能夠在智能機上運行Google AI模型。放在以前，手機要運行Google AI模型還得連接到數據中心纔行。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":"現在，有了 Pixel 6 和 6 Pro，Osterloh希望通過Tensor，谷歌能在智能手市場上與蘋果和三星並駕齊驅。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}},{"type":"strong"}],"text":"參考鏈接："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}],"text":" "}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"link","attrs":{"href":"https:\/\/www.engadget.com\/google-tensor-soc-pixel-6-chip-170059466.html?guccounter=1&guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS8&guce_referrer_sig=AQAAAApaAzPTpeQ6AeyM8mlB9m5K3N2yrDVwBHxvUu34n97B7462CqrL9wGqepMmdAlHPDycTUd2RBl0L1eVDhswANce2PLE6iXHRWYf4JONznngxTcNSupvkOuFQrqEXnlsRjjM6ZRgpL403kwp-mUDonugCnZc1lhVSglg9uG4mijB","title":null,"type":null},"content":[{"type":"text","text":"https:\/\/www.engadget.com\/google-tensor-soc-pixel-6-chip-170059466.html?guccounter=1&guce_referrer=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS8&guce_referrer_sig=AQAAAApaAzPTpeQ6AeyM8mlB9m5K3N2yrDVwBHxvUu34n97B7462CqrL9wGqepMmdAlHPDycTUd2RBl0L1eVDhswANce2PLE6iXHRWYf4JONznngxTcNSupvkOuFQrqEXnlsRjjM6ZRgpL403kwp-mUDonugCnZc1lhVSglg9uG4mijB"}],"marks":[{"type":"color","attrs":{"color":"#494949","name":"user"}}]}]}]}