IF / News

Samsung Electronics has allegedly received an order for Google’s upcoming Tensor G4 SoC, which will make its way to the Pixel 9 series. The Tensor G4 will feature an improved CPU from the G3. It bears an internal project codename “Zuma Pro”, indicating an evolution from the G3’s “Zuma”. The SoC is said to be produced using Samsung’s 4nm third-generation SF4P process, as opposed to the G3’s last-gen SF4 process. Google was considering a shift to TSMC with the Tensor G4. However, the plans fell through due to a lack of coordination on production timing and quantity. (Android Authority, Maeil Business)

Park Yong-in, President of Samsung System LSI, has revealed that the flagship Exynos chip, presumably Exynos 2500, which will launch in 2024, will use a 3nm fabrication process. It will use Samsung Foundry’s new 3nm GAA fabrication process, which is said to offer higher power efficiency and performance even against TSMC’s 3nm process. It was also revealed that in the future, NPUs will be used for AI processing rather than GPUs. Currently, GPUs are being used in cloud servers for AI and Generative AI applications, but they are too costly and consume a lot of power. So, Park Yong-in claims that NPUs are better suited for AI use cases. (Android Headlines, SamMobile, WCCFtech, Twitter)

Apple is unveiling its new M3 processors, and all three include big improvements to the GPU side of the chips. Built on TSMC’s N3B process, Apple is looking to once again raise the bar on both CPU and GPU performance, as well as setting a new record for the number of transistors used in a single laptop SoC. All three M3 chips, M3, M3 Pro and M3 Max will debut in new MacBook Pro models, with the M3 also part of a new 24” iMac.(The Verge, AnandTech, Apple)

Nvidia has trained its NeMo large language model (LLM) on internal data to help chip designers with tasks related to chip design, including answering general questions about chip design, summarizing bug documentation, and writing scripts for EDA tools. Nvidia’s chief scientist, Bill Dally, has presented the LLM, dubbed ChipNeMo. Foundation models like NeMo are generally trained on many trillions of words scraped from the internet to allow them to gain a good general understanding of language. LLMs can then be further pre-trained on domain-specific data to gain additional context in a particular field, and they can be fine-tuned with example question-answer pairs. Nvidia pre-trained ChipNeMo using a single data set scraped from internal repositories of code and text, including architecture documents, design documents and Nvidia’s code base, Dally said, and the company then fine-tuned on a subset of that data.(Android Headlines, EE Times, Nvidia, HPC Wire)

Taiwan Semiconductor Manufacturing Company (TSMC) founder Morris Chang has voiced a deep concern that is probably shared by many in the tech industry: the possible decoupling of the United States and China, the world’s two largest economies. He warned that such a separation would “slow down everybody”, reflecting on the complex web of global supply chains that both countries are part of. He believes “this will be harmful for everybody”. (Gizmo China, SCMP, 163.com, Takungpao, Yahoo)

Qualcomm CEO Cristiano Amon has not only confirmed that Samsung’s upcoming flagship (Galaxy S24 series) will once again adopt a two-chip strategy but that Qualcomm should have a majority share. The Galaxy S24 Ultra will allegedly be featured exclusively by the Snapdragon 8 Gen 3 “for Galaxy” SoC. However, the base Galaxy S24 model and the Galaxy S24+ will use both Exynos and Qualcomm chipsets. (GSM Arena, The Elec, SamMobile)

Qualcomm Snapdragon 8 Gen 4 will be the first from Qualcomm for this product category to feature Oryon cores. Qualcomm’s Senior Vice President Chris Patrick has said that custom CPU cores do not necessarily have to be expensive but also mentioned that Qualcomm needs to bring a balance to cost, power consumption, and performance. To achieve this balance would result in increasing the price of the Snapdragon 8 Gen 4. (CN Beta, WCCFtech, Android Authority)

Arm has announced that it has made a strategic investment, a minority stake in Raspberry Pi, the company of Raspberry Pi responsible for the new Raspberry Pi 5 and past Raspberry Pi products. Arm’s minority stake extends the long-term partnership between Arm and Raspberry Pi, which has seen Arm CPUs feature in all of the Raspberry Pi and Raspberry Pi Pico SoC.(CN Beta, The Register, Tom’s Hardware, Arm)

Samsung has reportedly worked with AMD and Qualcomm to jointly develop FSR (FidelityFX Super Resolution). The feature is expected to come to the company’s next-generation flagship smartphone, the Galaxy S24. This technology renders a game’s graphics at a lower resolution to improve frame rates and then uses AI to upscale the resolution. The higher frame rates could be acheived at the cost of a slight downgrade in graphics details. FSR aims to compete with Nvidia’s DLSS (Deep Learning Super Sampling), which is currently the industry’s leading spatial upscaling technology for PC games. (Android Authority, Business Standard, SamMobile, WCCFtech, Ajunews, Twitter)

Yangtze Memory Technologies Company (YMTC) has been reported as successfully developing and mass-producing 232-layer 3D NAND flash. The new 3D NAND memory devices boast the industry’s highest recording density of as well as extreme performance due to Xtacking 3.0 architecture. TechInsights has discovered the 232-layer Quadruple Level Cell (QLC) 3D NAND manufactured by YMTC in the 1 terabyte (TB) consumer solid-state drive (SSD) ZHITAI 600 that was quietly launched Jul 2022. In comparison, SK hynix’s NAND has a maximum layer count of 238 layers. Samsung Electronics has not disclosed its layer count specifically but is rumored to be around 236 layers. YMTC’s 1Tb 3D QLC NAND device has a recording density of 19.8 Gbit/mm2, which is the world’s highest density for a commercial IC. In fact, even a 232-layer 3D TLC NAND chip from YMTC boasts a recording density of 15.47 Gbit/mm2, which is higher than competing offerings that are in mass production. (Gizmo China, SCMP, Tom’s Hardware, Business Korea, TechInsights)

A long-expected merger of two of the world’s biggest memory chip companies — Western Digital and Japan’s Kioxia — may be hampered by one of Kioxa’s shareholders, SK Hynix. The company’s CFO Woohyun Kim has indicated that SK Hynix is not agreeing on the deal given the overall impact on the value of the company’s investment in Kioxia. Western and Kioxia have been working together in a joint venture since 2022. A deeper tie-up has allegedly been in the works for some time, with reports dating back to 2021. Western Digital and Kioxia allegedly have been talking to top Japanese banks to secure JPY1.9T (nearly USD13B) to finance the deal. Sumitomo Mitsui Financial, Mizuho Financial, Mitsubishi UFJ Financial and the Development of Bank of Japan are among the names floated as lenders. (TechCrunch, SK hynix)

Google has detailed its research into audioplethysmography (APG) that adds heart rate sensing capabilities to active noise canceling (ANC) headphones and earbuds “with a simple software upgrade”. Google says the “ear canal is an ideal location for health sensing” given that the deep ear artery “forms an intricate network of smaller vessels that extensively permeate the auditory canal”. This audioplethysmography approach works by “sending a low intensity ultrasound probing signal through an ANC headphone’s speakers”.(Android Authority, Google, 9to5Google)

Apple was allegedly so close to selling an Apple Watch for Android and how the Apple Watch Series 10 could come with blood pressure (BP) and Sleep Apnea monitoring. Apple is reportedly working on a blood glucose monitoring system that would warn Apple Watch users if they are prediabetic. Apple’s noninvasive blood sugar monitoring technique involves shining a light through the skin onto the fluid between the blood vessels and the cells. The light that reflects back can be used to access glucose levels. Apple’s solution also requires AI to analyze raw data and determine if a person is prediabetic. (Android Authority, Bloomberg)

Samsung is allegedly contemplating a significant enhancement in fast charging speeds. Samsung’s transformer supplier Atum has quietly introduced chargers boasting 65W and a 140W in their product lineup. All signs point to these new products being destined for Samsung’s future smartphones. This suggests that Samsung is actively working on a 65W charger. (GizChina, 4gnews, Twitter)

Samsung is starting to roll out a “Temporary Cloud Backup” feature to Samsung Galaxy smartphones and tablets running on One UI 6. It will allow users to back up and transfer an unlimited number of photos, videos, and other data for up to 30 days for free. The service will allow users to upload data from their smartphones or computers to Samsung Cloud storage for later use, such as when users are waiting for the delivery of their new device or when getting a smartphone repaired.(The Verge, Samsung, XDA-Developers)

Samsung Electronics has announced plans to unveil a smartphone with integrated generative artificial intelligence (AI) technology in 2024. The technology Samsung referred to is an on-device generative AI without connecting to an external cloud. This technology is believed to be featured in the Galaxy S24, expected to be revealed early 2024, or the Galaxy Z6 series to be launched in 2H24.(Android Authority, Business Korea, SamMobile)

Huawei has announced it reached 100M downloads of the new OS, HarmonyOS 4. It includes both newly bought and updated devices and happened in exactly 87 days – averaging over 1.1M installs per day. HarmonyOS 4 is equipped with the new Huawei “Ark” engine. Compared with HarmonyOS 3, the system fluency is improved by 20%, the battery life is extended by 30 minutes, and thousands of pictures in the gallery can be loaded instantly.(GSM Arena, Weibo, QQ, Yicai)

vivo has introduced Origin OS 4, based on Android 14. It brings a bunch of new features, a UI redesign and improvements in performance. Improvements in performance include faster app response, enabled by a “virtual graphics card”. It reduces opening and closing times of apps by 36% and boosts frame rate stability by 70%. Visual improvements include over 2,000 redesigned icons to improve visual comfort. The company brought a custom vivo Sans font across the whole system, and there are also UI Interactions with real-time blurring effects, contextual wallpapers, and lock screen customizations. The new version of Origin OS also comes with vivo smart car 4.0. It is the brand’s alternative to Android Auto, which is unavailable in China. (GSM Arena, Weibo, Fonearena, IT Home)

vivo has officially released its self-developed operating system called BlueOS. BlueOS has three main features: smart interactions, smoother user experience, and user data safety. The OS is designed to support large models and multi-modal interactions, which means that it can understand and respond to a variety of input methods, including voice, text, gestures, and even brain waves. BlueOS also supports the BlueXlink connection protocol, which adopts a distributed design concept and is compatible with industry-standard protocols. This allows data to be freely and securely transferred and accessed between multiple devices. vivo says that BlueOS has reduced memory usage by 67% and increased rendering efficiency by 48%. It also features a super coroutine mechanism that can increase response speed by 18%.(Gizmo China, IT Home)

According to Canalys, the global smartphone market underwent a slight drop of 1% in 3Q23, signaling a slowdown in its decline. Bolstered by regional recoveries and new product upgrade demand, the smartphone market recorded a double-digit sequential growth in 3Q23, ahead of the sales seasons. Samsung held onto the leading position with a 20% market share, despite a small dip from 3Q22, while Apple claimed second place with a 17% market share. Xiaomi took the third position with a 14% market share, with unit shipments recovering both annually and sequentially. OPPO (including OnePlus) captured the fourth spot with a 9% market share, owing to its strong position in Asia Pacific. Transsion rounded out the top five with a 9% share and remarkable year-over-year expansion. Outside the top five, Huawei made a major comeback in its home market, driven by the new Mate series.  (Android Headlines, Canalys)

Apple has 3 contract manufacturers namely Foxconn, Wistron, and Pegatron. Tata Group is set to manufacture iPhone in the country for both domestic and international markets. Electronics and Technology Minister Rajeev Chandrasekhar has said that the project will be completed within 2.5 years, and that the government will support the Tata Group and other Indian electronics companies in their efforts to partner with global brands. The deal was finalised after Wistron, granted approval for the sale of its India unit to Tata Group for USD125M. Tata Group will own a 100% stake in Wistron InfoComm Manufacturing India (WMMI) following board approval.  (Gizmo China, India Times, Patently Apple, Reuters, Indian Express, Twitter)

Apple, Dell, HP, Samsung and Lenovo are among 110 firms authorised by India to import laptops, tablets and personal computers under a new system aimed at monitoring shipments. Acer, Xiaomi, IBM and Asus have also been issued import authorisations, under India’s new “import management system” effective from 1 Nov 2023. India announced the new system for laptops, tablets and personal computers in Oct 2023 after it rolled back an earlier plan to impose a licensing regime, following criticism from the industry and Washington. Companies must register the quantity and value of imports on a portal, with an authorisation valid until Sept 2024.(Apple Insider, Reuters)

Autonomous vehicle operator Cruise, which is backed by General Motors (GM), will launch a robotaxi service in Japan by 2026. Japan would be Cruise’s second international market, after Dubai. Cruise currently operates robotaxis in several US cities and has said it plans to launch in 12 more in the years to come. GM made the announcement alongside Honda, which has a three-year-old partnership to co-develop a series of affordable electric vehicles. The three companies aim to establish the joint venture in 1H24pending regulatory approval. (The Verge, TechCrunch, Reuters)

The Ford subsidiary that operates the train station, Michigan Central, has announced that it is teaming up with the state’s Department of Transportation (MDOT) to test out the use of unmanned aircraft systems (UAM), or drones, to deliver medicine, food, and other small items to nearby residents. The drone delivery testing will take place in what’s being called Detroit’s Advanced Aerial Innovation Region, an area within a three-mile radius of Michigan Central Station in which drones can conduct deliveries to nearby homes and apartment buildings or perform other tasks like building inspections. (The Verge, Benzinga, MSN)

MIT has developed a deep learning system, Air-Guardian, designed to work in tandem with airplane pilots to enhance flight safety. This artificial intelligence (AI) copilot can detect when a human pilot overlooks a critical situation and intervene to prevent potential incidents. The backbone of Air-Guardian is a novel deep learning system known as Liquid Neural Networks (LNN), developed by the MIT Computer Science and Artificial Intelligence Lab (CSAIL). LNNs have already demonstrated their effectiveness in various fields. Their potential impact is significant, particularly in areas that require compute-efficient and explainable AI systems, where they might be a viable alternative to current popular deep learning models. (VentureBeat, MIT)

Researchers from the Australian National University, the University of Oxford, and the Beijing Academy of Artificial Intelligence have developed a new AI system called “3D-GPT” that can generate 3D models simply from text-based descriptions provided by a user. The system offers a more efficient and intuitive way to create 3D assets compared to traditional 3D modeling workflows. 3D-GPT is able to “dissect procedural 3D modeling tasks into accessible segments and appoint the apt agent for each task”. It utilizes multiple AI agents that each focus on a different part of understanding the text prompt and executing modeling functions. (VentureBeat, Github)