{"id":607,"date":"2025-12-12T11:08:52","date_gmt":"2025-12-12T03:08:52","guid":{"rendered":"https:\/\/aitimes.link\/?p=607"},"modified":"2025-12-12T11:08:52","modified_gmt":"2025-12-12T03:08:52","slug":"fsd-v12","status":"publish","type":"post","link":"https:\/\/aitimes.link\/index.php\/2025\/12\/12\/fsd-v12\/","title":{"rendered":"FSD V12"},"content":{"rendered":"\n

\u7279\u65af\u62c9 FSD V12 \u4e4b\u6240\u4ee5\u88ab\u79f0\u4e3a\u201c\u81ea\u52a8\u9a7e\u9a76\u7684\u5168\u65b0\u8303\u5f0f\u201d\uff0c\u6838\u5fc3\u5728\u4e8e\u5b83\u7b2c\u4e00\u6b21\u628a\u201c\u611f\u77e5-\u51b3\u7b56-\u63a7\u5236\u201d\u6574\u6761\u94fe\u8def\u538b\u8fdb\u4e00\u4e2a\u7aef\u5230\u7aef\u795e\u7ecf\u7f51\u7edc<\/strong>\uff0c\u7528\u6570\u636e\u9a71\u52a8\u5f7b\u5e95\u66ff\u4ee3\u4e86\u4eba\u5de5\u5199\u7684\u89c4\u5219\u4ee3\u7801\uff0c\u4ece\u800c\u628a\u81ea\u52a8\u9a7e\u9a76\u4ece\u201c\u5de5\u7a0b\u5e08\u5199\u903b\u8f91\u201d\u53d8\u6210\u4e86\u201c\u6a21\u578b\u81ea\u5df1\u5b66\u9a7e\u9a76\u201d\u3002\u5177\u4f53\u7279\u6b8a\u6027\u4f53\u73b0\u5728\u4e09\u70b9\uff1a<\/p>\n\n\n\n

    \n
  1. \u4ee3\u7801\u91cf\u4ece 30 \u4e07\u884c C++ \u76f4\u63a5\u780d\u5230 2 000 \u884c\u4ee5\u5185\uff0c\u89c4\u5219\u5f15\u64ce\u57fa\u672c\u6d88\u5931\uff0c\u8f66\u8f86\u52a8\u4f5c\u7531\u795e\u7ecf\u7f51\u7edc\u76f4\u63a5\u8f93\u51fa\uff0c\u5b9e\u73b0\u201c\u5149\u5b50\u8fdb\u3001\u63a7\u5236\u51fa\u201d\u3002<\/li>\n\n\n\n
  2. \u8bad\u7ec3\u65b9\u5f0f\u53d8\u6210\u7eaf\u6a21\u4eff\u5b66\u4e60\uff1a\u7528\u6570\u767e\u4e07\u6bb5\u4eba\u7c7b\u9a7e\u9a76\u89c6\u9891\u505a\u76d1\u7763\uff0c\u8ba9\u7f51\u7edc\u590d\u5236\u4eba\u7c7b\u53f8\u673a\u7684\u201c\u76f4\u89c9\u201d\uff0c\u957f\u5c3e\u573a\u666f\u6cdb\u5316\u80fd\u529b\u5927\u5e45\u63d0\u5347\uff0c\u57ce\u5e02\u590d\u6742\u8def\u51b5\uff08\u65e0\u4fdd\u62a4\u5de6\u8f6c\u3001\u65bd\u5de5\u7ed5\u884c\u7b49\uff09\u5e72\u9884\u6b21\u6570\u4e0b\u964d 65%\u3002<\/li>\n\n\n\n
  3. \u67b6\u6784\u4e0a\u5f7b\u5e95\u7aef\u5230\u7aef\uff0c\u6d88\u9664\u6a21\u5757\u5316\u7cfb\u7edf\u201c\u5c42\u5c42\u4fe1\u606f\u4e22\u5931\u201d\u7684\u7f3a\u9677\uff0c\u9a7e\u9a76\u98ce\u683c\u66f4\u63a5\u8fd1\u4eba\u7c7b\uff0c\u4f53\u611f\u4e1d\u6ed1\uff0c\u4e0d\u518d\u51fa\u73b0\u65e9\u671f\u7248\u672c\u201c\u673a\u68b0\u5f0f\u52a0\u51cf\u901f\u201d\u3002<\/li>\n<\/ol>\n\n\n\n

    \u4e1a\u754c\u628a V12 \u6bd4\u4f5c\u201c\u81ea\u52a8\u9a7e\u9a76\u7684\u5965\u672c\u6d77\u9ed8\u65f6\u523b\u201d\u2014\u2014\u5b83\u9996\u6b21\u8bc1\u660e\u7eaf\u6570\u636e\u9a71\u52a8\u8303\u5f0f\u5728\u57ce\u5e02\u573a\u666f\u53ef\u5de5\u7a0b\u5316\u843d\u5730\uff0c\u6807\u5fd7\u7740\u884c\u4e1a\u4ece\u201c\u89c4\u5219\u9a71\u52a8\u201d\u8fc8\u5165\u201c\u6570\u636e\u9a71\u52a8\u201d\u4e34\u754c\u70b9\u3002<\/p>\n\n\n\n

    <\/p>\n\n\n\n

    \u4e0b\u9762\u4ece\u201c\u56e2\u961f\u7ed3\u6784\u201d\u548c\u201c\u5173\u952e\u6280\u672f\u201d\u4e24\u6761\u7ebf\uff0c\u628a\u7279\u65af\u62c9 FSD \u7684\u201c\u4eba\u201d\u4e0e\u201c\u4e8b\u201d\u4e32\u8d77\u6765\uff0c\u65b9\u4fbf\u4f60\u4e00\u773c\u770b\u6e05\u8fd9\u5957\u7cfb\u7edf\u662f\u600e\u4e48\u88ab\u505a\u51fa\u6765\u7684\uff0c\u53c8\u5230\u5e95\u65b0\u5728\u54ea\u3002<\/p>\n\n\n\n

    \u4e00\u3001\u56e2\u961f\uff1a\u5c0f\u800c\u7cbe\u3001\u5782\u76f4\u6c47\u62a5\u3001\u8de8\u5b66\u79d1\u6df7\u7f16<\/p>\n\n\n\n

      \n
    1. \u9876\u5c42\uff1a\u76f4\u63a5\u5411\u9a6c\u65af\u514b\u6c47\u62a5\u7684\u662f\u201cAI & Autopilot\u201d \u526f\u603b\u88c1 Ashok Elluswamy\uff0c\u4ed6\u7edf\u7ba1\u611f\u77e5\u3001\u89c4\u63a7\u3001\u82af\u7247\u3001\u6570\u636e\u3001\u4eff\u771f\u4e94\u5927\u65b9\u5411\u3002<\/li>\n\n\n\n
    2. \u6838\u5fc3\u5c0f\u7ec4\uff1a
      \u2013 \u89c6\u89c9\u4e0e\u7f51\u7edc\u7ec4\uff1a\u8d1f\u8d23\u7aef\u5230\u7aef\u5927\u6a21\u578b\u3001\u591a\u6a21\u6001\u878d\u5408\uff08\u89c6\u89c9+\u8bed\u8a00+\u97f3\u9891\uff09\uff0c<20 \u4eba\uff0c\u4e3b\u529b\u662f\u524d OpenAI \u7814\u7a76\u5458\u3001ICCV \u6700\u4f73\u8bba\u6587\u5f97\u4e3b\u7b49\u3002
      \u2013 \u82af\u7247\u4e0e\u7f16\u8bd1\u5668\u7ec4\uff1a2016 \u5e74\u6210\u7acb\uff0cJim Keller \u4e0e Pete Bannon \u5e26\u961f\uff0c18 \u4e2a\u6708\u6d41\u7247\uff0c\u73b0\u5728\u8fed\u4ee3\u5230 HW4.0 \u53cc NPU\uff08\u5355\u9897 72 TOPS\uff09\uff0c\u529f\u8017 <7 W\u3002
      \u2013 \u6570\u636e\u5f15\u64ce\u7ec4\uff1a\u4e2d\u7f8e\u4e24\u5730\u8d85 100 \u4eba\u505a\u81ea\u52a8 4D \u6807\u6ce8\uff0c12 \u5c0f\u65f6\u53ef\u6807 1 \u4e07\u6bb5\u884c\u7a0b\uff0c\u76f8\u5f53\u4e8e 500 \u4e07\u5c0f\u65f6\u4eba\u5de5\u91cf\u3002
      \u2013 \u4eff\u771f\u4e0e\u9a8c\u8bc1\u7ec4\uff1a\u7528 3D Gaussian \u91cd\u5efa+\u5bf9\u6297\u751f\u6210\uff0c\u628a\u771f\u5b9e\u8857\u9053\u79d2\u53d8\u53ef\u7f16\u8f91\u865a\u62df\u573a\u666f\uff0c\u6bcf\u5929\u8dd1\u6570\u767e\u4e07\u516c\u91cc\u95ed\u73af\u6d4b\u8bd5\u3002<\/li>\n\n\n\n
    3. \u7ec4\u7ec7\u7279\u8272\uff1a\u529f\u80fd\u578b\u7ed3\u6784\u3001\u5c42\u7ea7\u6781\u6241\uff1b\u7b97\u6cd5\u3001\u82af\u7247\u3001\u8f66\u8f86\u3001\u6cd5\u89c4\u540c\u4e00\u5c42\u697c\u529e\u516c\uff0c\u4fdd\u8bc1\u201c\u4eca\u5929\u53d1\u73b0 bug\u3001\u660e\u5929\u5c31 OTA\u201d\u3002<\/li>\n<\/ol>\n\n\n\n

      \u4e8c\u3001\u6280\u672f\u521b\u65b0\uff1a\u4ece\u201c\u5199\u89c4\u5219\u201d\u5230\u201c\u5b66\u9a7e\u9a76\u201d\u7684\u4e09\u7ea7\u8df3<\/p>\n\n\n\n

        \n
      1. \u7aef\u5230\u7aef\u5355\u7f51\u7edc\uff08V12 \u8d77\uff09\uff1a
        \u8f93\u5165 7 \u8def 8-bit \u89c6\u9891\u6d41 \u2192 \u76f4\u63a5\u8f93\u51fa\u8f6c\u5411\/\u5236\u52a8\/\u52a0\u901f\uff0c\u4eba\u5de5 C++ \u4ece 30 \u4e07\u884c\u780d\u5230 2 000 \u884c\uff0c\u6d88\u9664\u6a21\u5757\u5316\u63a5\u53e3\u74f6\u9888\uff0c\u5b9e\u73b0\u201c\u5149\u5b50\u8fdb\u3001\u63a7\u5236\u51fa\u201d\u3002<\/li>\n\n\n\n
      2. \u591a\u6a21\u6001\u5927\u6a21\u578b\uff08V14\uff09\uff1a
        \u5728\u7aef\u5230\u7aef\u57fa\u7840\u4e0a\u5f15\u5165 VLA\uff08Vision-Language-Action\uff09\u6846\u67b6\uff0c\u7f51\u7edc\u5185\u90e8\u540c\u65f6\u751f\u6210 3D \u5360\u7528\u6805\u683c\u3001\u8bed\u8a00\u51b3\u7b56\u94fe\uff08Chain-of-Thought\uff09\u548c\u63a7\u5236\u6307\u4ee4\uff0c\u53ef\u201c\u770b\u61c2\u201d\u65bd\u5de5\u724c\u5e76\u201c\u81ea\u8a00\u81ea\u8bed\u201d\u63a8\u7406\u51fa\u201c\u9700\u501f\u9053\u5de6\u4fa7\u201d\u3002<\/li>\n\n\n\n
      3. \u81ea\u7814\u82af\u7247 + Dojo \u8d85\u7b97\uff1a
        \u2013 FSD Chip \u91c7\u7528\u53ef\u914d\u7f6e DMA \u4e0e\u7247\u4e0a SRAM \u5207\u7247\uff0c\u628a\u6743\u91cd\u9884\u53d6\u548c\u8ba1\u7b97\u5e76\u884c\u5316\uff0c\u5ef6\u8fdf <2 ms\uff0c\u529f\u8017\u53ea\u6709 GPU \u65b9\u6848\u7684 1\/5\u3002
        \u2013 Dojo \u7528 7 nm \u8bad\u7ec3 tiles \u62fc\u6210\u65e0\u4ea4\u6362\u673a\u62d3\u6251\uff0c\u5355\u67dc 1.1 EFLOPS\uff0c\u4e13\u95e8\u8dd1\u89c6\u9891\u7ea7\u65e0\u76d1\u7763\u9884\u8bad\u7ec3\uff0c\u628a OTA \u8fed\u4ee3\u5468\u671f\u4ece\u6708\u7f29\u5230\u5468\u3002<\/li>\n\n\n\n
      4. \u6570\u636e\u95ed\u73af\uff1a
        \u8f66\u7aef\u5f71\u5b50\u6a21\u5f0f\u91c7\u96c6 \u2192 \u4e91\u7aef\u81ea\u52a8\u6807\u6ce8 \u2192 Dojo \u8bad\u7ec3 \u2192 \u4eff\u771f\u9a8c\u8bc1 \u2192 OTA \u63a8\u9001\uff0c\u5168\u6d41\u7a0b\u65e0\u4eba\u503c\u5b88\uff1b\u957f\u5c3e\u6837\u672c\uff08\u62d6\u8f66\u3001\u5f02\u5f62\u969c\u788d\u7269\uff09\u901a\u8fc7 Occupancy Network \u5728\u7ebf\u6316\u6398\uff0c\u8bc6\u522b\u7387\u63d0\u5347 3 \u500d\u3002<\/li>\n\n\n\n
      5. \u5b89\u5168\u4e0e\u53ef\u89e3\u91ca\uff1a
        \u7aef\u5230\u7aef\u7f51\u7edc\u5728\u63a8\u7406\u65f6\u4f1a\u540c\u6b65\u8f93\u51fa\u8bed\u8a00\u89e3\u91ca\u3001\u5360\u7528\u6805\u683c\u3001\u8bed\u4e49\u5206\u5272\u4e09\u5f20\u201c\u4e2d\u95f4\u56fe\u201d\uff0c\u4f9b\u76d1\u7ba1\u4e0e\u5de5\u7a0b\u5e08\u56de\u6eaf\u51b3\u7b56\u8def\u5f84\uff0c\u5b9e\u73b0\u201c\u9ed1\u76d2\u91cc\u7684\u7070\u76d2\u201d\u3002<\/li>\n<\/ol>\n\n\n\n

        \u4e00\u53e5\u8bdd\u603b\u7ed3\uff1a\u7279\u65af\u62c9\u628a\u201c\u82af\u7247-\u6570\u636e-\u6a21\u578b-\u8f66\u201d\u56db\u6761\u94fe\u8def\u5168\u90e8\u81ea\u5df1\u505a\uff0c\u5e76\u7528\u7aef\u5230\u7aef+\u591a\u6a21\u6001+\u8bed\u8a00\u63a8\u7406\u628a\u81ea\u52a8\u9a7e\u9a76\u4ece\u201c\u5de5\u7a0b\u5e08\u5199 if-then\u201d \u5347\u7ea7\u4e3a\u201c\u6a21\u578b\u81ea\u5df1\u8bfb\u89c6\u9891\u3001\u81ea\u5df1\u63a8\u7406\u3001\u81ea\u5df1\u5f00\u8f66\u201d\uff0c\u8fd9\u624d\u662f FSD \u88ab\u4e1a\u5185\u89c6\u4e3a\u201c\u65b0\u8303\u5f0f\u201d\u7684\u6839\u672c\u539f\u56e0\u3002<\/p>\n","protected":false},"excerpt":{"rendered":"

        \u7279\u65af\u62c9 FSD V12 \u4e4b\u6240\u4ee5\u88ab\u79f0\u4e3a\u201c\u81ea\u52a8\u9a7e\u9a76\u7684\u5168\u65b0\u8303\u5f0f\u201d\uff0c\u6838\u5fc3\u5728\u4e8e\u5b83\u7b2c\u4e00\u6b21\u628a\u201c\u611f\u77e5-\u51b3\u7b56-\u63a7\u5236\u201d\u6574\u6761\u94fe\u8def\u538b\u8fdb […]<\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[6],"tags":[],"class_list":["post-607","post","type-post","status-publish","format-standard","hentry","category-ai"],"_links":{"self":[{"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/posts\/607","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/comments?post=607"}],"version-history":[{"count":1,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/posts\/607\/revisions"}],"predecessor-version":[{"id":608,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/posts\/607\/revisions\/608"}],"wp:attachment":[{"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/media?parent=607"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/categories?post=607"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/aitimes.link\/index.php\/wp-json\/wp\/v2\/tags?post=607"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}