UNISOC enters the 5G era – An interview with UNISOC

The 5G era has opened new opportunities for the semiconductor industry. In its latest market report 5G’s Impact on RF Front-End and Connectivity for Cellphones 2020, Yole Développement forecasts that the radio-frequency (RF) front-end and connectivity market will be worth $25B by 2025, a 11% compound annual growth rate. This would not have been possible without the support of the so-called “platform providers”. Multiple such System on Chip (SoC) providers are offering advanced chipsets that support the most advanced cellular communication standard ever. Additionally, integrated artificial intelligence (AI) inference processing has moved beyond the realm of just fancy camera filters. It is becoming a critical piece of security and human-machine interaction, as mentioned in the Processor Quarterly Market Monitor.

Cédric Malaquin, Technology & Market Analyst, RF Devices & Technology and John Lorenz, Technology & Market Analyst, Computing & Software, both at Yole Dévelopemment, had the chance to interview Yuning (Alan) Huang, Vice President of Marketing at one of the largest players in China, UNISOC. He kindly answers Yole’s questions and shares his insights below.

Yole Développement (YD): Could you please introduce UNISOC, its history and current activities?

Alan Huang (AH): UNISOC, formerly Spreadtrum, is a leading fabless semiconductor company committed to R&D of core chipsets in mobile communications and artificial intelligence/internet of things (AIoT). UNISOC’s products cover mobile chipset platforms supporting 2G/3G/4G/5G communication standards and various chipset solutions in the field of IoT, RF Front Ends (RFFE), wireless connections and TV. With more than 4,500 staff, 17 R&D centers and seven customer support centers around the world, UNISOC has become one of the top three mobile chipset suppliers in terms of global market share, the leading 5G company in the world and one of the largest chipset providers for IoT and connectivity devices in China.

YD: UNISOC’s V510 5G modem is built using a 12nm process and is reportedly powering internet customer premises equipment (CPE) devices in China. Do you plan further scaling of this product to support mobile handsets?

AH: V510 is the first 5G Modem from UNISOC. As a highly integrated, high performance, power-efficiency 5G modem with a flexible architecture, V510 supports both standalone (SA) and non-standalone (NSA) network configurations across a range of scenarios.
V510 has already been applied to a number of 5G devices developed by OEMs and solution providers, including mobile phones, modules, CPE, and so on.

CPE with UNISOC ivy 510 - Courtesy of UNISOC
CPE with UNISOC ivy 510 – Courtesy of UNISOC

YD: UNISOC announced that HiSense’s F50 5G smartphone was powered with the T7510 5G chipset supporting SA/NSA, sub-6 GHz frequency bands and Wi-Fi 5/BT 5.0. Are there any additional features in the T5720 5G chipset built using 6nm EUV lithography from TSMC?

AH: The T7520 was developed on UNISOC’s Makalu 5G platform. It integrates the world’s first 5G modem to support coverage enhancement for all application scenarios.
By using advanced 6nm EUV process technology from TSMC, compared to the previous 7nm process, there is an 18% improvement in density of transistors. This allows more transistors to be integrated per unit area and reduces power consumption by 8%, allowing the battery to last longer.
For more information on the T7520, please visit this link: http://www.unisoc.com/breaking-unisoc-launches-new-gen-5g-soc-t7520

T7520 5G chipset - Courtesy of UNISOC
T7520 5G chipset – Courtesy of UNISOC

YD: Does UNISOC plan to introduce an RF transceiver supporting 5G? If yes, when will it come to the market?

AH: Yes. It has already been used in the T7510, and it will be used in T7520 with updates as our new generation SoC.

YD: When can we expect UNISOC’s chipset to support 5G millimeter wavelength (mmWave) bands?

AH: Although the demand for 5G mmWave bands in almost all markets is not strong except in the United States, UNISOC is still investing in research and development and promises to disclose the progress to the public gradually. Being one of the only three chipset vendors that is capable of providing 5G mobile chip solutions to various customers, we are always committed to creating value for our customers whether it is sub-6 GHz or mmWave bands.

YD: UNISOC is developing 5G mmWave Antennae-in-Packages. Could you give insights into the technology used and associated performance? When do you expect this device to be mass produced and for which market?

AH: We have many products and technologies that are being developed internally to ultimately fulfill our customers’ needs. Stay tuned!

YD: Will UNISOC develop its RF Front-End product portfolio to support 5G signal amplification, rooting and filtering? Or will you develop partnerships with other RF Front-End companies?

AH: 5G RF Front-End mass production is planned to start in 1Q 2021. However, we do not rule out the possibility of cooperating with other companies.

YD: You announced a partnership with Imagination Technologies to license their neural network accelerator (NNA) tech in 2019. Do the newest Tiger chipsets deploy this AI acceleration technology? If not, when should we expect to see that integrated?

AH: UNISOC T7510 and T7520 all integrated the latest NNA tech from Imagination Technologies, which proves that Imagination has the leading AI technology. We have launched an AI strategic cooperation with Imagination to jointly promote the commercial use of AI in the fields of industry, commerce, medical treatment, home furnishing and education, and accelerate the intelligent upgrading of traditional industries.

T710 AI chipset - Courtesy of UNISOC
T710 AI chipset – Courtesy of UNISOC

YD: How does NNA technology differ from previous approaches you have taken with AI inference?

AH: NNA tech is the fastest and most energy-efficient neural network hardware-accelerated embedded solution on the market. Based on the award-winning predecessor products, the new version of NNA technology provides unparalleled scalability, enabling SoC manufacturers to optimize computing power and capabilities for a range of embedded markets such as automobiles, mobile devices, smart video surveillance, and IoT edge devices. Thanks to architectural enhancements including lossless weight compression, the performance of the architecture can be increased by 40% over the previous generations on the same chip area, allowing SoC manufacturers to increase performance efficiency by nearly 60% and reduce bandwidth requirements 35%.

YD: Where will we find embedded AI processing using the NNA from your strategic partnership?

AH: Our consumer electronics and industrial electronics businesses will add AI computing to enable all kinds of products, including mobile phones, smartwatches and industrial related products.

YD: GPU architecture has performed AI acceleration tasks in other computational situations and in some of your competitors’ processors.  How do you see the processor industry’s evolution in AI acceleration from GPUs towards the newer neural network architectures that are becoming more common?

AH: GPU architecture is very common for AI acceleration. Both Qualcomm and Apple all use GPU architecture for their AI acceleration. We believe that GPUs will support some AI calculations, and UNISOC is open to working on this architecture in the future.

YD: There is some interest in using spiking neural networks, specifically on edge/IoT applications. Do you see that as a viable path for your product roadmap?  Why/why not?

AH: We have a strong interest in spiking neural networks. We believe that spiking neural networks is a good direction. We are doing related research and at the same time we have a lot of communication with and are learning from many parties.

YD: What can we expect from UNISOC in the next two to three years?

AH: UNISOC is set to become the technology enabler behind innovation for the public. We believe 5G devices should not be deemed as pricey and high-end, they are for everyone.

Interviewee

Yuning (Alan) Huang is Vice President of UNISOC. He has worked in the mobile communication semiconductor industry for nearly 20 years. He has experienced many key positions in chip research and development, AI technology, architecture design, product planning, strategic cooperation, and overseas market expansion. He has outstanding achievements and successful experience in almost all fields. After obtaining a master’s degree from the University of Bristol in the United Kingdom, he became the backbone of the chip R&D team when HiSilicon was founded. Alan served as chief architect of HiSilicon Kirin AI and made important contributions to the AI competitiveness of Kirin chips. He is responsible for the construction of basic software and hardware open source ecosystem.

Interviewers

As a Technology & Market Analyst, specialized in RF devices & technologies within the Power & Wireless division at Yole Développement (Yole), Cédric Malaquin is involved in the development of technology & market reports as well as the production of custom consulting projects.
Prior his mission at Yole, Cédric first served Soitec as a process integration engineer during 9 years, then as an electrical characterization engineer during 6 years. He deeply contributed to FDSOI and RFSOI products characterization. He has also authored or co-authored three patents and five international publications in the semiconductor field. Cédric graduated from Polytech Lille in France with an engineering degree in microelectronics and material sciences.

John Lorenz is a Technology and Market Analyst within the Computing & Software division at Yole Développement (Yole), part of Yole Group of Companies. John is engaged in the development of market and technology monitors for the logic segment of advanced semiconductors, with an initial focus on processors. Prior to joining Yole, John held various technical and strategic roles at Micron Technology.
On the engineering side, his roles included thin film process development and manufacturing integration on DRAM, NAND, and emerging memory technologies and industrial engineering / factory physics for the R&D fab.
On the strategic side, John ran the memory industry supply & capex model for corporate strategy / market intelligence, and established the industry front-end costing model within strategic finance.
John has a Bachelor of Science degree in Mechanical Engineering from the University of Illinois Urbana-Champaign (USA), with a focus on MEMS devices.

Related reports

Application Processor – Quarterly Market Monitor BD

Processor Quarterly Market Monitor
With SMIC, China is positionning as a competitor in foundry business and thereby catching the dynamic of the APU market.



5G’s Impact on RF Front-End and Connectivity for Cellphones 2020
An intensifying US-China competition for RF technology supremacy.


Source: https://www.unisoc.com/#/home/main

Related presentations

Cet article vous a plu ?

Partagez-le sur vos réseaux sociaux