作者：斯蒂芬·内利斯 (Stephen Nellis) 旧金山9月29日电：从卡内基梅隆大学剥离出来的芯片初创公司Efficient Computer于周二表示，该公司已通过新一轮融资筹集了9700万美元，使其估值达到6.5亿美元。
By Stephen Nellis SAN FRANCISCO, Sept 29 : Efficient Computer, a chip startup spun out of Carnegie Mellon University, on Tuesday said it has raised $97 million in additional funding, bringing its valuation to $650 million.
Efficient正在开发采用所谓数据流架构的芯片，这种芯片的速度可以更快，且能效比英特尔公司或英伟达公司使用的架构高出10到100倍。
Efficient is developing chips with what is known as data flow architectures that can be both faster and 10 to 100 times more energy-efficient than the architectures used by Intel Corp or Nvidia Corp.
数据流芯片在学术文献中已经出现了几十年，但一直未能获得商业上的成功，这主要是因为软件开发人员很难针对英特尔或英伟达芯片所能处理的各种繁杂任务对它们进行编程。然而，激增的计算需求激发了人们对数据流芯片的新一轮兴趣。
Data-flow chips have bounced around academic literature for decades without commercial traction, largely because they are difficult for software developers to program for the wide variety of tasks that can be handled by chips from Intel or Nvidia. But booming demand for computing has spurred a new wave of interest in data-flow chips.
总部位于宾夕法尼亚州匹兹堡的Efficient表示，该公司从学术实验室的零基础开始，开发了硬件和软件工具，使数据流方法适用于通用处理器。虽然Efficient计划最终制造可用于数据中心的大型芯片，但其首批出货的芯片主要面向无人机和小型机器人。
Pittsburgh, Pennsylvania-based Efficient says it started from scratch in academic labs to develop both the hardware and software tools to make the data-flow approach suitable as a general purpose processor. While Efficient plans to eventually make large chips that can be used in data centers, the first chips that is shipping are aimed at drones and small robots.
尽管这些芯片体积极小，但将人工智能功能塞入这些由电池供电的设备中，意味着芯片需要比过去处理更多样化的任务。
While those chips are tiny, cramming AI features into those battery-powered devices means the chips need to handle a greater variety of tasks than in the past.
首席执行官布兰登·露西亚 (Brandon Lucia) 在接受采访时表示：“我们算是找到了一个平衡点，既容易编程、速度快，又高效。当你构建一个人工智能系统时，该系统最终要做的事情远不止两个微型纳米优化的AI算法那么简单。”Efficient没有透露其营收情况，但领投本轮融资的TQ Ventures合伙人安德鲁·马克斯 (Andrew Marks) 表示，这笔新资金将用于增加到明年的出货量。
"We sort of thread the needle where we're easy to program, fast, and efficient," CEO Brandon Lucia said in an interview. "When you build an AI system, the system ends up doing a lot more than two little nano-optimized AI algorithms." Efficient did not disclose its revenues, but Andrew Marks, partner at TQ Ventures, which led the funding round, said the new cash will go toward increasing shipments through next year.
马克斯说：“有很多芯片初创公司拥有伟大的架构。但能将硅片真正送到客户手中的却寥寥无几。”
"Plenty of chip startups have great architecture. Very few get silicon into customers' hands," Marks said.
凭借这笔新资金，Efficient 迄今累计融资达 1.73 亿美元。除 TQ Ventures 外，本轮融资参与方还包括 Eclipse、Union Square Ventures、Giant Ventures、Triatomic Capital、TO Capital、TF Capital、Mana Ventures、Toyota Ventures、Overmatch 和 Borderless。
With the new capital, Efficient has raised $173 million to date. In addition to TQ Ventures, the funding round included participation from Eclipse, Union Square Ventures, Giant Ventures, Triatomic Capital, TO Capital, TF Capital, Mana Ventures, Toyota Ventures, Overmatch and Borderless.