See their AI strategy from Intel's new Xeon processor and FPGA card

Intel is demonstrating its two new Xeon processors and new FPGA cards that support deep learning at the SupercompuTIng 2016 conference in the US; from the company's technology showcase, you can see the complete machine learning solution that it is preparing to launch. One corner.

The industry has been looking forward to Intel's machine learning program, which is one of the hottest technology areas in the semiconductor industry in recent years; this is a newcomer to artificial intelligence (AI) processor design, Graphcore co-founder and CEO Nigel Toon said: "I think Brian Krzanich (Intel CEO) will be very big on related technologies." Graphcore recently raised $30 million in funding from investors including Intel rival Samsung.

Intel also has some investments in machine learning technology, such as the acquisition of two processor startups Nervana and Movidius, which specialize in neural network technology, and Altera, which was acquired for $16.7 billion. The latter's FPGA has been Baidu (Baidu). ), Microsoft (Microsoft) and other large companies are used in data centers to speed up search engines, networking and other tasks.

At the same time, Intel continues to view its large multi-core x86 processor Xeon Phi as a weapon against Nvidia and AMD's graphics processing gas; at the IDF Intel Technology Forum in August, the company said its codenamed Knights Mill's Phi The processor will be the first product to simultaneously play the main processor and accelerator, and is expected to ship in 2017.

Machine learning plays the role that Intel’s former CEO Andy Grove might call a “key turning point”. The company has a high market share in the server processor market, but machine learning needs to have higher computing performance than those chips. Support for highly parallelized tasks.

Google has used its own ASIC to accelerate this reasoning task in machine learning, while Intel's goal is to take advantage of the new PCIe card combined with Altera Arria FPGAs; Facebook has designed its own GPU server with Nvidia chips to support training. The computationally intensive task of neural networks.

As for Nvidia, it released its own GPU server earlier this year and partnered with IBM on another version of the Power processor; AMD launched an open software program for its GPU earlier this year.

All of the above manufacturers want to compete in the machine learning hardware market - according to market research firm TracTIca, the market size will be 436 million US dollars in 2015, and can further grow to 41.5 billion US dollars by 2024; huge business opportunities have prompted Graphcore Semiconductor startups such as Wave CompuTIng and Cornami have strategically increased their investments, as well as the two companies Intel recently acquired.

Graphcore's Toon said: "Intel's acquisition of Nervana is for software -- I don't think its hardware technology is strong enough;" but an Intel executive said that the company will lock Nervana's hardware to the private cloud enterprise application market. As for Movidius, Toon said its expertise is consistent with Intel's RealSense family of 3D camera targets for low-power system applications at the network/IEF edge nodes.

"To be honest, I am not sure what role Altera will play;" Toon joined Altera in 1988 and is responsible for the promotion of its European operations: "Intel may try to use FPGA as a machine learning accelerator in some way, but I don't think That's the right approach.” He pointed out that Graphcore's multi-core chips will be a higher performance solution than FPGAs or GPUs in deep learning applications.

Compilation: Judith Cheng

(Reference: Intel Preps its AI Strategy, by Rick Merritt)

Ground Screw For Other Buildings

High Dip Galvanized Ground Screw Pile made of carbon steel :Q235 ,according to standard ;DIN EN ISO1461-1999 ,not only used on the Solar Mounting projects,but also can be use for many other fields ,such as some temporary building,for timber,flage and banner,traffice sign ,for fencing construction ,for container house,for vineyard fencing ,for bridge & . it is easy to install ,no need digging ,no-concrete ,save time and cost .no harm to environment .

Due to their design and ease to installation,they are most commonly used wheneversoil conditons preventstandard foundation solutions,instead of requiring large excavationwork,they thread into the ground.

Construction Ground Screw,Hdg Ground Screw,Screw Anchor,Helical Pier,Ground Screw Anchor

BAODING JIMAOTONG IMPORT AND EXPORT CO., LTD , https://www.chinagroundscrew.com