Baidu and Samsung release AI chip, designed based on Samsung’s 14nm process and I-Cube package technology

Chinese-language Internet search provider Baidu and Samsung Electronics announced that Baidu’s first cloud-to-edge AI accelerator, Baidu KUNLUN, has completed its development and will be mass-produced early next year.

Baidu KUNLUN chip is built on the company’s advanced XPU, a home-grown neural processor architecture for cloud, edge, and AI, as well as Samsung’s 14-nanometer (nm) process technology with its I-Cube (Interposer-Cube) package solution.

The chip offers 512 gigabytes per second (GBps) memory bandwidth and supplies up to 260 Tera operations per second (TOPS) at 150 watts. In addition, the new chip allows Ernie, a pre-training model for natural language processing, to infer three times faster than the conventional GPU/FPGA-accelerating model.

Leveraging the chip’s limit-pushing computing power and power efficiency, Baidu can support a variety of functions including large-scale AI workloads, such as search ranking, speech recognition, image processing, natural language processing, autonomous driving, and deep learning platforms like PaddlePaddle.

Through the first foundry cooperation between the two companies, Baidu will provide advanced AI platforms for maximizing AI performance, and Samsung will expand its foundry business into high performance computing (HPC) chips that are designed for cloud and edge computing.

As higher performance is required in diverse applications such as AI and HPC, chip integration technology is becoming more and more important. Samsung’s I-Cube technology, which connects a logic chip and high bandwidth memory (HBM) 2 with an interposer, provides higher density/ bandwidth on minimum size by utilizing Samsung’s differentiated solutions.

Compared to previous technology, these solutions maximize product performance with more than 50 percent improved power/signal integrity. It is anticipated that I-Cube technology will mark a new epoch in the heterogeneous computing market. Samsung is also developing more advanced packaging technologies, such as redistribution layers (RDL) interposer and 4x, 8x HBM integrated package.

“We are excited to lead the HPC industry together with Samsung Foundry,” said OuYang Jian, distinguished architect of Baidu. “Baidu KUNLUN is a very challenging project since it requires not only a high level of reliability and performance at the same time, but is also a compilation of the most advanced technologies in the semiconductor industry. Thanks to Samsung’s state of the art process technologies and competent foundry services, we were able to meet and surpass our goal to offer superior AI user experience. ”

“We are excited to start a new foundry service for Baidu using our 14nm process technology,” said Ryan Lee, vice president of Foundry Marketing at Samsung Electronics. “Baidu KUNLUN is an important milestone for Samsung Foundry as we’re expanding our business area beyond mobile to datacenter applications by developing and mass-producing AI chips. Samsung will provide comprehensive foundry solutions from design support to cutting-edge manufacturing technologies, such as 5LPE, 4LPE, as well as 2.5D packaging.”

Intel releases new class of AI hardware from cloud to edge to boost AI development, deployment and performance

Intel has updated its artificial intelligence (AI) offerings with new products designed to accelerate AI system development and deployment from cloud to edge. Intel demonstrated its Intel Nervana Neural Network Processors (NNP) for training (NNP-T1000) and inference (NNP-I1000) — Intel’s initial purpose-built ASICs for complex deep learning with incredible scale and efficiency for cloud and data center customers. 

Intel also revealed its next-generation Intel Movidius Vision Processing Unit (VPU) for edge media, computer vision and inference applications.

These products further strengthen Intel’s portfolio of AI solutions, which is expected to generate more than $3.5 billion in revenue in 2019. The broadest in breadth and depth in the industry, Intel’s AI portfolio helps customers enable AI model development and deployment at any scale from massive clouds to tiny edge devices, and everything in between.

Now in production and being delivered to customers, the new Intel Nervana NNPs are part of a systems-level AI approach offering a full software stack developed with open components and deep learning framework integration for maximum use.

The Intel Nervana NNP-T strikes the right balance between computing, communication and memory, allowing near-linear, energy-efficient scaling from small clusters up to the largest pod supercomputers. 

The Intel Nervana NNP-I is power- and budget-efficient and ideal for running intense, multimodal inference at real-world scale using flexible form factors. Both products were developed for the AI processing needs of AI customers like Baidu and Facebook.

“We are excited to be working with Intel to deploy faster and more efficient inference compute with the Intel Nervana Neural Network Processor for inference and to extend support for our state-of-the-art deep learning compiler, Glow, to the NNP-I,” said Misha Smelyanskiy, director, AI System Co-Design at Facebook.

Additionally, Intel’s next-generation Intel Movidius VPU, scheduled to be available in the first half of 2020, incorporates unique, highly efficient architectural advances that are expected to deliver leading performance — more than 10 times the inference performance as the previous generation — with up to six times the power efficiency of competitor processors. 

Intel also announced its new Intel DevCloud for the Edge, which along with the Intel Distribution of OpenVINO toolkit, addresses a key pain point for developers — allowing them to try, prototype and test AI solutions on a broad range of Intel processors before they buy hardware.

With most of the world running some part of its AI on Intel Xeon scalable processors, Intel continues to improve this platform with features like Intel Deep Learning Boost with Vector Neural Network Instruction (VNNI) that bring enhanced AI inference performance across the data center and edge deployments. 

While that will continue to serve as a strong AI foundation for years, the most advanced deep learning training needs for Intel customers call for performance to double every 3.5 months, and those types of breakthroughs will only happen with a portfolio of AI solutions like Intel’s. Intel is equipped to look at the full picture of computing, memory, storage, interconnect, packaging and software to maximize efficiency, programmability and ensure the critical ability to scale up distributing deep learning across thousands of nodes to, in turn, scale the knowledge revolution.