菜单
press center
trade news

NXP under the wave of AIoT: Where does “competitiveness” come from?

date: 2019-08-28
view count: 9
share to:

The emergence of AIoT has triggered another technological revolution in the chip industry. Meanwhile, more and more players are engaged in it, setting off an unprecedented wave of innovation and entrepreneurship.However, Evergrande, the largest chip industry, the strong constant strong law has not changed, with the continuous intensification of competition, now all the head chip manufacturers are accelerating to establish their own ecology, strengthen their 'absolute dominance', which also makes the AIoT dominant power of the battle becomes more and more exciting.


As the only chip company in the world with two major product lines of MCU and application processor, NXP is also a pioneer in this huge wave of AIoT.With this unique advantage, NXP introduced the concept of 'crossover processor' two years ago, which is a big change not only for NXP, but also for the embedded market.


Of course, this is only the beginning and far from the end.In 2019, NXP has become more active than usual, introducing a series of innovative solutions such as eIQ machine learning and artificial intelligence development environment to the Chinese market, applying its own technology to the joint laboratory of tianjin university to train students, and further deepening cooperation with smic.These new trends all reflect NXP's importance to the Chinese market and its determination and attitude to deeply explore AIoT application scenarios in China.


Where does the 'crossover processor' concept go?


When NXP launched its crossover processor two years ago, it heard plenty of skepticism.But true feedback from the market, in the past two years, more and more customers are adopted i.M X RT solution, grace wisdom PuWei controller division director of global product jin-tao zeng China heavy current subnet reporter said: 'many of our competitors are now launch similar products, from the customer and the competitor's feedback, all can be seen that the concept of 'crossover processor, namely the high performance of MCU has been widely adopted.'


Of course, the crossover processor concept of two years ago is not the end, but only the beginning.Relying on in the past on the crossover processor rich experience in product development, now the grace of the wisdom of pu on the basis of the original architecture is further strengthened the concept of 'crossover processor', grace wisdom pu Yu Xiujie, general manager of LPC and low power consumption micro controller product line is introduced: 'today's i.M X 7 ULP is a new concept, its characteristic is has two cores, an architecture (A7 and an architecture (M4 nucleus.Although there are many products with two cores on the market now, the characteristic of i.mem7 ULP is that the A7 and M4 cores are in two completely independent domains, for example, the cortex-a7 can run Linux operating system, while the cortex-m4 core has a very low power consumption.In most cases, the cortex-a7 core is turned off, while cortex-m4 handles some basic tasks to keep the device running at low power.'


Why the 'cortex-a7 + cortex-m4' combination?'To date, cortex-a7 is the most energy efficient architecture ARM has ever built,' explains Geoff Lees, general manager of enzip micro controller business unit. 'there is a new ARM v8 cortex-a35, which is based on the cortex-a7.While cortex-a7 is much better than cortex-a5, the power consumption is not much higher, but the performance is enhanced.We launched cortex-a5 six years ago, but we think it is the best on i.mp7 ULP.At the same time, we also have cortex-a8 and cortex-a9 products, which are characterized by fast running speed but not low power consumption. The best solution is to use cortex-a7 and cortex-a35 for low-power consumption, while the ULP products we launch later are mainly based on cortex-a35.'


This kind of independent solution with low power consumption and high energy efficiency can meet the two-way demand of AIoT fragmentation market while continuing the innovation of 'crossover processor'.According to NXP, the i.mp7 ULP is now in mass production and has sold more than 1 million units in a very short period of time.'We first introduced the crossover processor two years ago. Now it's just the beginning, and NXP has many new products and concepts to come,' yu said.


In addition to low power and high performance, low cost is another key aspect of NXP's search for crossover processors.Jin-tao zeng said: 'the crossover processor product line, we also have a low cost of new products, namely i.M RT1010 X, the MCU can reach 500 MHZ, Intel' pentium 'with the past, for example, the i.M X RT1010 computing power than a pentium, '6 pieces of eight MAO 8 products can achieve such performance, in the past is not in the market.In comparison, the current market for 500MHz is either very expensive, up to $5, 6, or even $8, 9, or very poor performance products, which can only reach the level of 100MHz or less.And the ability to do that at this price point, the i.mx RT1010 is the first one.'


'Meaningful' platforms EdgeVerse and eIQ


In addition to the new layout on the hardware side, NXP also brings with it the EdgeVerse platform, which will be launched at the future tech summit on June 10.By gathering one of the world's most comprehensive portfolio of edge computing products on a common brand platform, NXP integrates NXP's strong ecological environment support and Internet of things support, and promotes the edge computing industry to accelerate forward with the help of NXP's more than 30 years of successful experience in embedded processing.


According to the reporter, the EdgeVerse platform includes NXP's eIQ machine learning software development environment, Immersiv3D audio experience package, and EdgeScale device management platform. NXP customers can use the Immersiv3D audio experience package with existing models, or quickly design, train, learn, and optimize new models for deployment across NXP's entire product portfolio.By providing tools and engines that help drive machine learning, reasoning, and easy cloud connectivity, edge ai is stimulated and simplified.From small hardware to high-end NPU with strong processing capabilities, from the lowest end of IoT nodes to the cloud, EdgeVerse has a matching solution, even providing services that support IoT authentication.


EIQ machine learning as the blockbuster EdgeVerse platform, software development environment has become the focus, grace wisdom PuWei controller division director of global product jin-tao zeng said: 'the eIQ is a software platform, the current market there are a lot of this kind of machine learning algorithm, such as simple voice, but the speech in a big run on the coprocessor is unnecessary, the power consumption and high cost.For many small applications, just take a 500MHz cortex-a7 core and you can do good speech processing and even voice noise reduction.Therefore, grace of pu eIQ hardware platform is not for other company, but have to be combined with the software, such as some customers have the Google algorithm, can choose according to the algorithm of Google we a hardware platform, such as choosing i.M X 8, or can recommend eIQ platform based on different scheme and its application to choose different combination chip.'


This means that customers can choose an optimized hardware package based on different applications, power consumption, price and other requirements. NXP platform can provide very comprehensive evaluation and test support for this.Geoff Lees said: 'the real point of the eIQ software development platform is to enable customers to choose a hardware from our product and match it with the machine learning model they want to meet their final application requirements based on their needs.This development environment is currently compatible with almost every open source training tool, modeling, and editor on the market, and almost every open standard or custom standard on the market can be developed in this development environment.Because of this software platform, we can apply machine learning to all product lines, from low-end microcontrollers to high-end microprocessors.Now some of the latest embedded machine learning model, there are many different engines run together on the platform, there are all kinds of models on the market, depending on the model with the hardware is different also, you can use the CPU running, can use DSP operation, machine learning coprocessor can also be used to run, our software can be optimized according to the different hardware machine learning model.'


To the top of the cloud, from the smallest MCU eIQ has all kinds of machine learning model, Geoff Lees told reporters: 'we the eIQ platform can help customers to optimize their algorithms, customers can choose according to their own demand for power or performance in a suitable for the application of the hardware, based on grace wisdom pu MCU or high-end processors development and their applications, not only the high performance of the processor is the most suitable for machine learning, even in very low on the MCU can run up to 10 times, 20 times or higher performance.'


summary


The innovation potential of crossover processors is still huge in the future. After all, the AIoT boom has just begun, and with the implementation of more and more applications, the long-lost market boom is bound to ignite. The solutions with low cost, low power consumption and high performance will be welcomed by the market.


Grace for the 'crossover processor' of the creator of the wisdom of pu, now more and more competitors are power and layout under the situation of 'crossover processor technology, strengthening its ecological control ability is crucial, and launch EdgeVerse and eIQ platform is also let us see the grace wisdom pu's efforts in this aspect, next, as long as you continue to technology innovation, grace of pu will surely in the era of AIoT chip market more' like a duck to water.'


WeChat Official Platform

share to:
Copyright 02019 - 2022 Shenzhen likun technology co., LTD
Copyright @ shenzhen likun technology co., LTD. All rights reserved
犀牛云提供云计算服务