1. Shih Jung-Jung: Zhang Chung-mou made the most contribution to Taiwan's industry and "let the world see Taiwan."
Set a micro-channel news, Zhang Zhong Mou led the TSMC record miracles over the past 30 years, in addition to Taiwan's stock market leading the value of the stock market last March, TSMC market value surpass Intel, grabbed the semiconductor leader throne Acer founder Shi Zhenrong that Zhang Zhongmou contribution to Taiwan's industry The biggest, because TSMC "let the world see Taiwan."
"Zhang Zhongmou founded TSMC at the age of 56 and used an innovative business model to subvert the semiconductor industry." Mirror Weekly reported that TSMC's original wafer foundry model has allowed IC design companies to save money on large-scale factory closures by simply designing them TSMC to get out of the chip to get.
Zhang Zhongmou proud of 30 years, he devastated innovation to make TSMC an indispensable force in the semiconductor industry, while Shi Zhenrong praised Zhang Zhongmou's contribution to the industry in Taiwan no one can, "because TSMC let the world see Taiwan Without TSMC, there is no such thing as IC design companies such as Qualcomm, Broadcom, Huida and MediaTek booming.
2. The first meeting of Broadcom Qualcomm ended Qualcomm said it will assess the next move;
SAN FRANCISCO, Feb. 15 morning news, on Broadcom's $ 121 billion acquisition of Qualcomm's trading proposal, the first talks between the two companies concluded this week.Goltone said the board will hold a meeting to discuss the next move, But did not say whether the two companies have reached an agreement.
After the talks in New York, Qualcomm said in a statement: 'Earlier today, we had two hours of talks with Broadcom representatives and listened carefully to their opinions. Qualcomm's board of directors will meet promptly to discuss the talks Results, and determine the next step. '
Sources close to Broadcom said that although Qualcomm listened to the opinions, there was no interaction with Broadcom.
Last week, Qualcomm's board of directors rejected Broadcom's 'best-of-breed' deal, $ 82 a share, which Qualcomm believes is still too low. The company agreed to meet Broadcom CEO Chen Fuyang but said Broadcom needs to be raised Bidding, and better ensure that the transaction can be approved within a reasonable time regulatory authorities.Chen Fuyang responded that the latest bid is he can offer the highest bid.The two sides will be March 6 Qualcomm shareholders conference showdown, Broadcom appeal Qualcomm investors elected the board of directors of the company nominated to facilitate the completion of the transaction.
If you fail to win the support of Qualcomm's board of directors or shareholders, then Chen Fuyang will be an uncommon setback, through a series of transactions, he has made Broadcom into the $ 380 billion semiconductor industry, one of the largest companies by acquiring the market's largest Qualcomm, a maker of smart phone chips, Broadcom will achieve unprecedented impact in the area of contemporary communications system components.
Including chairman and CEO Paul Jacobs and CEO Steve Mollenkopf, Qualcomm's board of directors hopes to market market-leading mobile chip technologies to new areas such as servers, PCs and cars. Chen Fuyang believes that such an expansion plan is unrealistic, the industry is changing, the growth rate is slowing down.
At this meeting, Chen Fuyang did not see all the members of Qualcomm's board of directors, representatives of Qualcomm include Jacopo, Molenkov, Tom Horton, and the other three members of the Qualcomm management team only Huo While Dayton is an independent director, others are Qualcomm's longtime executives who uphold Qualcomm's current strategy and plan to operate as independent companies.
Qualcomm believes the company's future prospects are bright, but it's hard to believe in the past two years, and Qualcomm's technology licensing business is facing a challenge that has become a core issue for Qualcomm.
3.MIT new chip enhance AI neural network instruction cycle seven times, less energy consumption 95%;
AI development potential is optimistic about all parties, but AI operations need to consume a lot of energy, but it is behind the development of the hidden danger.MIT recently announced a new chip, seven times faster than the traditional AI chip neural network instruction cycle, reducing energy consumption by 95% In the future, if the popularization of AI can withstand the general consumer devices such as mobile phones installed above.
Face recognition, speech recognition, which we used to use, are behind the results of the neural network operations.Generally speaking, the neural network is very large, so to put the phone often only has a simplified architecture, can not play the nerve network's strong ability, or The new MIT chips reduce the amount of data passed to the memory and computing modules of the chip, increasing the computational efficiency by three to seven times, accelerating the speed and reducing the power consumption by 95%.
If MIT's new chips popularize, these new chips can drive local neural networks to home-based smart devices, and future human voice commands can be handled locally without having to be connected to the Web as now or Amazon Alexa Spread to the data center, and then back.
When Apple released the A11 bionic chip in 2017, it had the AI capability to identify if it was the owner of a handset using Face ID. A similar application in the future will not be only on the iPhone, or Face ID, but more devices and applications technews
4.16 core 100W power consumption! AMD Zen fought embedded;
Since the advent of the AMD Zen architecture, it has covered different market segments, including consumer desktops / laptops, business-class desktops / notebooks, and server data centers, and has entered the new APU, with the next battlefield to be embedded.
Taiwan's industrial embedded manufacturer iBase today exposed a new motherboard 'MBN806', the processor labeled 'AMD EPYC Embedded 3000 Series', or rather a BGA integrated package on the motherboard SoC chip System, model EPYC 3201, 8 core 8 threads.
Before I heard of a EPYC 3251, but the specification is unknown, it is estimated that the core number more.
And this embedded EPYC 3000 series actually we already heard, in the first half of 2016 was exposed, was also classified as Opteron Opteron sequence.
Designed for high-end embedded markets such as communications and networking, the same code-named Snowy Owl, is based on 14nm Zen and is a modular design that is divided into SCM BGA and MCM BGA, 16 core 32 threads (32MB L3 cache), and another 12 core, 8 core, 4 core version, thermal design power 35-100W.
In contrast, ThreadRipper up to 16 core 32 threads, but the thermal design power consumption reached 180W, of course, the frequency of the EPYC 3000 series will certainly be much lower.
The EPYC 3000 Series also features four channels of DDR4 memory (RDIMM / UDIMM / LRDIMM / NVDIMM / Flash / 3DS), up to 64 PCI-E 3.0 buses, up to 16 SATA or NVMe devices, up to eight 10GbE 10 Gigabit Ethernet and Integrate independent security subsystem, server controller center.
Unlike the top-of-the-line Naples EPYC 7000 Series for data centers, the EPYC 3000 Series is a BGA integrated package that plugs directly into the motherboard.
One of the big benefits of an embedded platform is stability and durability. AMD will provide up to 10 years of technical support for the EPYC 3000 Series.
Its rival is obviously the Intel Xeon D series, which is currently the most high-end Xeon D-1587 specifications for 16 core 32 threads, 24MB L3 cache, the highest frequency of 2.1GHz, 32 PCI-E 3.0, thermal design power 20- 65W. Fast technology
5. NVIDIA or release of mining graphics, code-named 'Turing'
According to foreign media Digital Trends reported that NVIDIA is expected to be in March this year, the GTC for the gaming market launch of a new video card, code-named may be Turing (Turing).
NVIDIA announced a new generation of Volta graphics chip architecture in May last year, a framework that will also first appear on the GV100 chip for use in the Tesla V100 graphics card. The Tesla V100 is used primarily in the data center market and will not be used in the gaming product line in.
It is speculated that NVIDIA will release its new graphics card architecture, Ampere, in April of this year. Ampere is expected to replace the existing GeForce GTX 10 series graphics card based on Pascal architecture and will first be used on the GV104 chipset for the high-end market. The video card is expected to be sold in April of this year, which is in line with the rhythm that NVIDIA will release a new game card at the GTC.
In addition, the GV104 is also expected to be used on the GeForce GTX 2080/2070, while the GV102 will target high-end enthusiasts such as the GTX 2080 Ti, while the GV106 may target the mid-range and the GV107 / 108 for the low-end.
Here are three symbols: Volta, Ampere, Turing.
What we know now is that the Volta architecture will be used in the Tesla V100. And, NVIDIA does not usually develop different architectures for different markets, and if there is a market differentiation, it will develop a single generation architecture and produce different chips In order to serve different markets.
The latest generation of architecture is undoubtedly Volta, and Ampere and Turing are likely to refer to chips for both markets: gaming and cryptocurrency mining. Previously rumored to be Ampere gaming graphics, Turing refers to mining graphics.
From the name, one can see some clues: Turing is from the British computer scientist Alan Turing, who helped to crack the German password during World War 2. Given the work of Turing in cryptography, the name is used to encrypt currency Mine GPU is also reasonable.
Turing first appeared in NVIDIA last week's earnings conference, when NVIDIA also said that the game graphics business at a record low while the mining industry is snatching cards with the game, so that the price increase, in short supply.