Nordic首席执行官：RISC-V是Arm的补充，不是威胁

Nordic Semiconductor 首席技术官 Svein-Egil Nielsen 撰写了一篇文章，详细介绍了该公司对加入 RISC-V 潮流的想法——它认为这补充了该公司与专有芯片 IP 巨头 Arm 的现有关系，而不是与之竞争。

Nielsen在文章中承认：“乍一看，Nordic 宣布将全力支持开源芯片架构的开发和采用，这可能会让人质疑该公司与 Arm 的持续关系有多牢固。” 。“毕竟，Arm 提供的商业芯片架构绝不是开源的，并且自 2012 年推出 Nordic 的 nRF51 系列片上系统 (SoC) 以来，一直用于 Nordic 的半导体无线连接产品。”

Nordic早在 8 月份就宣布了对免费开源 RISC-V 指令集架构 (ISA) 的兴趣，并宣布与博世、英飞凌、恩智浦半导体和高通技术公司成立合资企业。五位行业重量级公司正在组建一家德国公司，旨在加速该平台及其软件生态系统的商业化——首先是在汽车市场中使用，然后扩展到物联网 (IoT) 和移动领域。

从低功耗微控制器一直扩展到多核超级计算机，RISC-V 指令集架构是免费且开源的——没有人需要签署任何保密协议、支付大笔许可或版税，甚至不需要获得许可开始围绕它进行构建。尽管所有这些似乎都给行业巨头 Arm 带来了麻烦，该公司的收入仅由其处理器核心和其他知识产权 (IP) 的许可和特许权使用费驱动，但Nielsen表示，两者都有发展空间。

“我对 RISC-V 的看法与我对无线物联网连接标准激增的看法类似：没有一种技术能够万能（解决）所有应用问题。那么，蓝牙 LE 真的能与 Thread 或蜂窝物联网竞争吗？当然不是。每种产品都旨在做好不同的事情，但不是所有事情，”Nielsen认为。“因此，RISC-V 实际上是 Arm 的补充替代品，而不是威胁。在传统上 Arm 占据主导地位的功耗关键型移动和物联网应用中尤其如此。”

Nielsen 声称 Nordic 对 RISC-V 的兴趣主要在于为客户提供“自由和灵活性”，以减少模块化 ISA 的功能，从而尽可能降低功耗，但它将仅限于“某些特定的情况”。以及高度专业化的应用。” 正如首席执行官兼总裁 Teo Swee Ann 当时所说， “我们有一些特殊的需求。”

“那么，开发超低功耗指令集的能力在哪里才能真正发挥作用呢？对我来说，它处于边缘，例如，用于传感器的更简单的嵌入式芯片，需要少量的处理能力才能提供本地化的机器学习，”Nielsen说。“在此类应用中，拥有 Arm 核心完全是多余的。但该传感器可能仍然需要与基于 Arm 核心的 Nordic 设备通信并协同工作。

“RISC-V 不会与 Nordic 长期以来在其无线物联网连接设备中使用 Arm 内核发生冲突。”

Nordic recently announced that it is joining an industry consortium of semiconductor companies to drive the adoption of an open source chip architecture called RISC-V. Why?

RISC-V versus Arm

At first glance, Nordic announcing that it is putting its full technological and commercial weight behind the development and adoption of an open-source chip architecture might raise questions about how strong the company’s on-going relationship with Arm is.

After all, Arm provides a commercial chip architecture that is anything but open source, and has been used on Nordic's semiconductor wireless connectivity products since the 2012 launch of Nordic's nRF51 Series Systems-on-Chip (SoCs).

The nRF51 Series redefined the leading-edge of the Bluetooth LE market once and for all. And one of the biggest innovations was the inclusion of a powerful on-board Arm processor for the very first time on a Bluetooth chip. Nothing like it had ever been seen before in the Bluetooth LE market. In fact, the nRF51 Series literally sent Nordic's competitors scrambling back to the drawing board when it came to how to design a Bluetooth LE chip.

RISC-V will complement Arm

RISC-V was developed about 10 years ago at UC Berkeley in the U.S. and at first glance would seem to be competitive with Arm.

But I think this is way too narrow a view. I see RISC-V in a similar way to how I see the proliferation of wireless IoT connectivity standards: no one technology can be all things to (solve) all application problems. So does Bluetooth LE really compete with Thread or cellular IoT ? Of course not. Each is designed to do different things well, but not everything.

What RISC-V is in reality, therefore, is a complimentary alternative to Arm, and not a threat. And this is particularly true in power consumption-critical mobile and IoT applications where Arm has traditionally been dominant.

Optimizing for lowest power

At its heart, RISC-V is all about promoting innovation by giving users the ability to develop leading-edge, customized hardware based on an open-source chip architecture.

What RISC-V will do for Nordic's customers is give them the freedom and flexibility, in certain specific and highly specialized applications, to strip down the instruction set to ensure extreme levels of low power consumption.

To me, this is no different to how you can strip down a 2.4GHz proprietary wireless protocol to optimize for the ultra lowest power consumption in a way you simply could not match at such extremes with a standards-based protocol such as Bluetooth low energy.

Now RISC-V isn't of course proprietary. But just like a Nordic proprietary 2.4GHz protocol, if you are the developer behind it, you get far more control over every line of code to customize for the specific needs of your specific application.

Lowest power counts at the edge

So where might the ability to develop an ultra lowest power instruction set really come into its own?

To me it's at the edge in, for example, simpler embedded chips for sensors that require a small bit of processing power in order to deliver localized machine learning.

In such applications having an Arm core would be complete overkill. But that sensor may still need to communicate with and work alongside an Arm core-based Nordic device.

More options

What RISC-V will do is give our customers an extra option when seeking to minimize power consumption in certain applications where the (many) tradeoffs of not using an Arm-based core are acceptable.

RISC-V will also lower the barriers-to-entry and level the playing field when it comes to developing IoT applications. This will encourage even greater innovation in the IoT market.

What RISC-V will not do is conflict with Nordic's long-established use of Arm cores in its wireless IoT connectivity devices.

RISC-V will simply further enhance the design options for Nordic customers and particularly in simpler, ultra lowest power applications where every Joule and Watt counts.

*免责声明：本文由作者原创。文章内容系作者个人观点，半导体行业观察转载仅为了传达一种不同的观点，不代表半导体行业观察对该观点赞同或支持，如果有任何异议，欢迎联系半导体行业观察。

今天是《半导体行业观察》为您分享的第3584期内容，欢迎关注。