'Mining' NVIDIA/AMD graphics card sales both dropped sharply;

1. Xiaolong 1000 exposure for the PC to create the first product or from Asus; 2. MediaTech and Alibaba deepening AI cooperation to open a new era of smart office; 3. Low-end mobile phone processors were exited, the mid-high-end market is facing a fierce battle; 4. Cool Core obtains license and deploys the CEVA-XM4 smart vision platform; 5. 'Mining' arrives NVIDIA/AMD graphics card sales both drop sharply; 6. Big Three gather together: Micron produces GDDR6 memory for the next generation;

1. Xiaolong 1000 exposure is specially designed for PC First product or from Asus;

Micronet reported on June 25th that Qualcomm is accelerating its innovation in mobile computing. According to reports, a processor named SDM1000 (presumably Xiaolong 1000) is under development testing and may be launched next year. After the evolution of the 835, 850 generation mobile computing platform, the Xiaolong 1000 may be a dedicated chip product for Qualcomm in the true sense of the mobile PC field.

The current news about the SDM1000 comes mainly from foreign media. It is expected that the SDM1000 will have an Arm Cortex A76 large core and a Cortex A55 small core as the CPU architecture, and the foundry will be TSMC.

The SDM1000 measures 20mm*15mm, which is much smaller than the Intel 45mm*24mm, but larger than the Qualcomm mobile chip. The 12W power indicator will bring a 35% speed increase, equivalent to Intel's Atom and Core Y series. , Close to the U series (15W). Intel's Y and U series are super-oriented chips, the main low-power performance, Y series speed is lower than U.

In terms of process, it is not clear whether the SDM1000 uses 7 nm or 10 nm.

It is understood that the SDM1000 will be the first ARM architecture with pin pluggable Opteron processors instead of being soldered on the motherboard. Theoretically, it can be replaced on the motherboard.

In terms of storage, SDM1000 supports 16G LPDDDRA4X RAM and two 128G UFS2.1 storage, which will carry 5G baseband.

So far, for the mobile PC industry, Qualcomm has launched the Xiaolong 835 mobile computing platform last year, and recently launched the Xiaolong 850 mobile computing platform. This is seen as an embodiment of its ability to extend the capabilities of the mobile sector to the PC industry. Mobile chip upgrades and "grafting", and SDM1000 will be a chip specifically for the PC field.

The current Windows 10 notebook equipped with the SDM1000 may come from Asus, codenamed 'Primus', which is equipped with at least a 2K resolution display and supports WiGig (60GHz 802.11ad). In the ultrabook market, Asustek has the world's largest market share. Asus is also a close partner of Qualcomm in the mobile PC field.

Qualcomm's mobile PC products with its chips are called 'Long Long', and its main advantage is its high-speed, high-speed performance advantage. It is also considered to be the advantage of Qualcomm's impact on Intel.

In the recently held 'Benjamin's enjoyment of the freedom - on the Dragon Mobile PC platform Windows notebook tasting', Qusong's senior vice president and president of QCT China District Wu Shangjie announced that it includes the Xiaolong 835 mobile computing platform. ASUS, Lenovo notebook products will be released in the near future. Today, the first ASUS Changkyun 370 with Xiaolong 835 mobile platform has been on sale in Jingdong. (Proofreading / Fan Rong)

2. MediaTech and Alibaba deepen AI cooperation to open a new era of smart office;

On June 25th, Microtek announced that MediaTek and Alibaba.com have jointly announced that they will work together to create an intelligent office system based on artificial intelligence. This is in accordance with the intelligence around the voice interaction with Alibaba artificial intelligence laboratory. After the cooperation in the field of home, MediaTek and Alibaba have reached another important cooperation in artificial intelligence. So far, MediaTek and Alibaba have opened two artificial intelligence applications in AI Voice and AI Vision. On the full cooperation.

'MediaTek is actively deploying in AI, launched the NeuroPilot artificial intelligence platform that integrates hardware and software at the beginning of the year to create an AI ecosystem with open AI strategy. Moreover, MediaTek has a wealth of IP and cross-platform capabilities to help customers Several fields quickly deployed AI to a variety of different terminal types. 'The deputy general manager of MediaTek Technologies and general manager of the smart equipment business group Tong Jie said: 'Half a year, MediaTek has cooperated with Alibaba's two major business units in AI. Touching on the two application fields of daily life and corporate office. Alibaba has a strong Internet gene and user base. We are very pleased to have a deep and comprehensive cooperation with Alibaba in the field of artificial intelligence. We look forward to the cooperation between both parties to accelerate artificial intelligence in the terminal. Side of the large-scale landing. '

Alibaba’s nails are the world’s leading smart mobile office platform. Currently, it has hundreds of millions of users. The total number of active users exceeds the sum of second to tenth in the industry, and the number of service companies has exceeded 5 million. In the early days, the cooperation between the two parties was mainly Around the smart frontier supporting face recognition, the future will also expand to more applications, such as smart cloud print cartridges. The smart frontier of this cooperation is based on MediaTek's high-integration, high-performance chip platform to meet AI computing needs. Efficient and stable face recognition brings more efficient, smart office and management methods to the business.

Yi Tong (Ren Qing), Vice President of Alibaba Spikes, stated: 'Stuck has always been committed to enabling companies to achieve zero barriers, zero-cost access to the convenience of the mobile Internet era, making corporate offices more efficient, safe, simple, and happier employees. Nail will export the unified standards of DingTalk Inside, provide companies with integrated office solutions for hardware and software, and work together with eco-partners such as MediaTek to create an open and intelligent office hardware ecosystem and use intelligent hardware products based on AI technology. China's 43 million companies provide state-of-the-art hardware and software services. ' MediaTek is committed to becoming a driver for terminal AI, bringing terminal artificial intelligence (Edge AI) to various cross-platform terminal devices - from smart phones, smart homes to self-driving cars Etc. Each year, about 1.5 billion Taiwan-based consumer electronic products using MediaTek chips have AI capabilities.

The first Alibaba intelligent front-end product equipped with MediaTek's chip platform will be available in July this year.

3. The low-end mobile phone processor was withdrawn, and the mid-to-high-end market was facing a fierce battle;

Gathering micro-messages (text/Xiaobei) Recently, MediaTek officials stated that shipments of smart phone processors increased significantly in the second quarter, shipments were stronger than expected, and handset processor shipments are expected to exceed 100 million. The gross profit margin is expected to rise back to above 37%. In the second half of the year, low-end and mid-range mobile phone processors will be introduced. The annual gross profit margin is expected to exceed 38%.

As predicted by Strategy Analytics analyst Sravan Kundojjala, 2018 is a crucial year for MediaTek and ZF.

This year, MediaTek switched from the Helio X high-end line to the Helio P mid-to-high-end processor line with a cost-effective advantage, and released the mid-range processor Helio P60 in March.

As mobile phone processor manufacturers increasingly give high-end performance to mid-range mobile phone chips, the boundaries of mid- to high-end mobile phone processors will become increasingly blurred, and low-end mobile phone processors may speed up.

At one time, Qualcomm had a clear lineup of mobile phone processors, the low-end 400 series, the mid-range 600 series, and the high-end 800 series. Last year, Qualcomm gave the Xiaolong 600 Series the 'characteristic' of the 800 series, which made people feel that Xiaolong 660 did not It is a mid-range processor. This year, Qualcomm launched the 700 series between the 800 series and the 600 series, targeting the high-end smartphone market.

The introduction of Qualcomm's 700 series confirms the blurring of the boundaries of Xiaolong's mid- to high-end mobile phone chips, and shows that Qualcomm is aware of the competitiveness of MediaTek's mid-range mobile phone processors. This processor makes Qualcomm feel a sense of "crisis." It is Helio P60.

The Helio P60 is preemptively equipped with an AI function and is manufactured on TSMC's 12nm process.

In a bid to win, MediaTek launched the Helio P22 processor in May, which will inject a stimulant into the midrange mobile phone market. It will be built on TSMC's 12nm FinFET process, with an 8-core A53 CPU and a maximum frequency of 2.0GHz. It will be equipped with MediaTek's CorePilot 4.0 and NeuroPilot. Smart technology, support for face recognition, smart album, single/double camera depth of field AI camera.

In addition, Ziguang Zengru also made it clear that its 5G chip will be cut into the high-end market in 2021.

It is not difficult to find that the mid-range mobile phone processor market will become the most competitive market in the next few years, and the low-end mobile phone processor will soon be 'out of business'.

According to industry insiders, as mobile phone processor manufacturers compete for the mid-to-high end chip market, they will also appropriately weaken the low-end processor strategy.

This year, high-end mobile phone processors are generally equipped with AI functionality, and the selling point of 'AI' has become not new. Mobile phone processor makers and smart phone manufacturers need to find the next selling point. Insiders said, no doubt, the next selling point is set '5G'.

The period before the arrival of the 5G era has become a great time for major mobile phone manufacturers and mobile phone processor companies to seize market and resources. Industry experts believe that at this stage of slowing demand, mobile phone and chip manufacturers will pass a The cost-effective' battle will determine the future market pattern. Those mobile phone supply chain manufacturers that are not competitive will face the crisis that has been eliminated. (Proofreading/Spring/Summer)

4. Cool Core obtains license and deploys CEVA-XM4 intelligent vision platform;

CEVA announces that Shanghai Coolcore Microelectronics, China's leading provider of drone and robot system-on-chip (SoC) products, has licensed the CEVA-XM4 smart vision platform and deployed on the soon-to-be-released AR9X01 artificial intelligence (AI) SoC device. Provides support for computer vision and deep learning workloads.

Cool core chief technology officer Shen Bo said: 'The cool core continues to promote the innovation of drones, adding new features and new technologies that can improve performance, flight time and autonomy. The CEVA-XM4 intelligent vision platform provides us with processing capabilities and Development Kit, which truly harnesses the power of AI in our drone SoC design. '

Since the company was founded in 2011, Coolcore has been the leading chip supplier in the UAV market. To date, its controller chips have helped millions of UAV manufacturers worldwide. AR9X01 It is the most complex chip designed by Coolcore so far. It uses multiple CEVA-XM4 cores to analyze real-time flight environments and uses artificial intelligence to accomplish various tasks including object detection, classification, and tracking. Compared to CPU or GPU-based alternatives Due to the inherent low power consumption of the CEVA-XM4 platform running computer vision algorithms and deep learning reasoning, the AR9X01 allows drone manufacturers to maximize flight time and improve overall drone performance.

Ilan Yona, CEVA vice president and general manager of the vision business unit, said: 'Our CEVA-XM series of intelligent vision processors and artificial intelligence coprocessors can better realize vision-based machine learning in embedded devices and continue to lead the industry. We are pleased to announce that the recognized leader in the field of drones has become one of the many customers on the CEVA-XM4 platform, enabling its AR9X01 SoC devices to fully utilize AI functionality.

CEVA's latest generation of image and vision DSP platforms can not only meet extreme processing power requirements, but also reduce power consumption constraints for the most sophisticated machine learning and machine vision applications such as smartphones, surveillance, AR and VR, drones and autonomous cars. These DSP platforms include a hybrid architecture consisting of scalar and vector DSP processors and hardware accelerators, as well as a comprehensive Application Development Kit (ADK) that simplifies software development. The CEVA ADK includes: CEVA-Link for software level assistance with the host processor Seamless integration; A wide range of widely used and optimized software algorithms; The CEVA Deep Neural Network (CDNN) real-time software framework that simplifies machine learning deployments and consumes far less power than GPU-based systems; and advanced development and debugging tools.

5. 'Mining' Arrival The sales of NVIDIA/AMD graphics cards have fallen sharply;

Crazy mining made NVIDIA, AMD, the two major graphics card manufacturers gain a lot in the past, but they suffered from a large number of gamers. Sometimes even if they could not afford to buy more money, they could not buy their favorite products. The continued decline in the prices of virtual electronic money, such as bitcoin has fallen below 6000 US dollars, a new round of 'mine accident' has been staged, NVIDIA, AMD sold hot miner cards have suffered heavy losses.

The first quarter of 2018 can be said to be the peak of the mine card. Statistics show that the world sold 1.7 million mineral cards in the quarter. AMD's 10% of this quarter's revenue comes from this, NVIDIA also has 9%.

However, the higher the climb, the easier it would be to fall. After entering the second quarter, it was not just the plummeting Chinese mining card market. North America has not been spared.

According to the latest figures from BlueFin Research Partners, sales of NVIDIA graphics cards in the North American market will be at least 50% lower in the second quarter than in the first quarter. AMD is more miserable because of more mine cards, and only 35 sales in the past quarter. -40%.

Of course, the hardship of the game players has passed, and the game card market is slowly returning to normal.

With the return of major manufacturers' attention, many of the video cards that are out of stock prices are gradually recovering. The former popular mine cards such as RX 580 and GTX 1060 have returned to the price of around 2,000 yuan.

6. The Big Three gather together: Micron produces GDDR6 memory for the next generation;

Micron announced that it has begun mass production of next-generation GDDR6 memory chips. It is also the last of the DRAM giants. Samsung and SK Hynix have already mass-produced and shipped GDDR6. Not only is the progress slow, but the Micron GDDR6 also lags behind the other two in terms of specifications. Home, the current single-capacity is 8Gb (1GB), will be later to achieve 16Gb (2GB) - GDDR6 maximum allow a single 32Gb (4GB).

In terms of speed, Micron is the first to offer 10Gbps, 12Gbps, 14Gbps three options, with voltages of 1.25V and 1.35V. Compared with GDDR5 1.5-1.6V, it is naturally advantageous to save power.

16Gbps particles are also under development. Micron has even demonstrated 20Gbps ultra-high speed internally. Of course, other people started to mass-produce 18Gbps at the beginning of this year.

However, even with low speeds of 12Gbps, GDDR6 has a high bandwidth of 48GB/s, which is half that of the fastest GDDR5. 14Gbps GDDR6 is up to 17% faster than the fastest GDDR5X.

If it is 20Gbps, that bandwidth is a terrible 80GB/s. The total bandwidth under a 256-bit bit-width can reach 640GB/s, twice the GTX 1080, and even approaching the Titan V of HBM2 memory.

Micron said that in addition to traditional GPU graphics, their GDDR6 memory can also be used in AI artificial intelligence, networking, automotive and other fields.

2016 GoodChinaBrand | ICP: 12011751 | China Exports