最新公告:
新闻动态
联系方式

地址:
电话:
联系:

新闻动态

当前位置:主页 > 新闻动态 >      
自家AI芯片让特斯拉对自己的未来拥有了更多的控制权
作者:admin 发布于:2018-11-08 13:26

据外媒techcrunch8月2日报道,埃隆•马斯克在今天的财报电话会议上表示,“在过去2-3年里,我们在AI芯片方面基本上一直处于半保密模式,我想是时候把秘密泄露出去了……”
 
According to foreign media techcrunch on August 2, Elon Musk said at today's earnings conference that "in the past two to three years, we have been basically in a semi-confidential mode in the field of AI chips, and I think it's time to let the secret out..."
 
 
 
AI芯片制造的问题的关键在于:特斯拉自身的硬件,也就是“Hardware 3”。这是一款由特斯拉制造的硬件,被用于替换Model S、X和3,来完成所有需要的数字运算,以提高汽车自动驾驶能力。
 
The key to the problem of AI chip manufacturing is Tesla's own hardware, that is, "Hardware 3". This is a Tesla-made hardware, which is used to replace Model S, X and 3 to complete all the required digital operations to improve the automotive driving ability.
 
 
 
截至目前,特斯拉一直依赖于英伟达的驱动平台。那其为何选择现在开始改变?特斯拉表示,自己进行研发生产时可集中精力于自己的需要,以提高效率。
 
Up to now, Tesla has always relied on the Vivian driving platform. So why does it choose to start changing now? Tesla says that when it comes to R&D and production, it can concentrate on its own needs to improve efficiency.
 
 
 
 
 
特斯拉Hardware 3项目的负责人皮特·班农说:“我们有一个优势,那就是知道我们的神经网络是什么样子,以及它们在未来是什么样子。”同时,硬件升级将于明年开始。
 
"We have an advantage in knowing what our neural networks look like and what they will look like in the future," said Peter Bannon, head of the Tesla Hardware 3 project. Meanwhile, the hardware upgrade will start next year.
 
 
 
“AI芯片的关键在于能否运行基本的神经网络。你必须在电路中做这些计算,而不是在某种模拟模式下,这就是GPU或CPU的工作方式。你要用这里的内存做大量的计算。”马斯克补充说道。
 
The key of AI chip is whether we can run the basic neural network. You have to do these calculations in the circuit, not in some analog mode, which is how the GPU or CPU works. You need to do a lot of computation with the memory here. " Musk added.
 
 
 
据马斯克透露,最终测试结果相当惊人,特斯拉的计算机视觉软件在Nvidia的硬件上运行,每秒可处理200帧,但在专用芯片下每秒以2000帧的速度运行。
 
According to Musk, the final test results are quite amazing. Tesla's computer vision software runs on Nvidia's hardware and can process 200 frames per second, but runs at 2000 frames per second on dedicated chips.
 
 
 
正如人工智能分析师James Wang所说,这让特斯拉对自己的未来拥有了更多的控制权。至于未来如何将这些芯片植入现有的特斯拉车型,不妨期待一下。
 
As James Wang, an AI analyst, puts it, Tesla has more control over her future. As for how to put these chips into the existing Tesla models in the future, we may as well look forward to it.

 


下一篇:岳阳全域旅游推介会拓展市场 当场签约8个优质旅游项目

上一篇:鸡蛋比汽油贵20万倍!委内瑞拉能靠“石油币”救国吗?

友情链接: 织梦CMS官方 DedeCMS维基手册 织梦技术论坛