This page was machine-translated and may differ from the original. View original
Samsung Electronics Reveals Plans for Mass Production of Baidu's High-Performance AI Chips
Collaboration on the production and development of the 14nm process-based AI chip 'KUNLUN'
Comprehensive foundry solutions covering design, process, and packaging
Samsung Electronics announced that it plans to mass-produce the 'Kunlun' AI chip, based on the 14nm process of Baidu, a major Chinese internet search engine company, in early 2020.(1).jpg)
▲ Baidu's AI chip KUNLUN <Image=Samsung Electronics>
With this first foundry collaboration, Samsung Electronics is now able to expand its foundry business scope to include AI chips that can be used in cloud and edge computing.
Baidu's Kunlun is an artificial intelligence chip that can be used in various fields of AI ranging from cloud to edge computing. It is a product realized with high performance of 512 GBps bandwidth and 260 TOPS (Tera Operations Per Second, 150W) by applying Baidu's proprietary XPU architecture, Samsung Electronics' 14nm process, and I Cube (Interposer Cube) packaging technology.
Samsung Electronics has improved power (PI) and electrical signal (SI) quality by more than 50% compared to existing solutions by applying a foundry solution optimized for High Performance Computing (HPC).
This improves the noise generated when a signal is transmitted to the chip, and the voltage It is a solution that maintains a constant level to enable the circuit to operate more stably.
I Cube is Samsung Electronics' unique differentiated 2.5D packaging technology that integrates SoC chips and HBM (High Bandwidth Memory) chips onto a silicon interposer (Si-Interposer). This technology has the advantage of increasing transfer speeds and reducing package area by placing each chip within a single package.
Ouyang Jian, Chief Architect and Head of AI Semiconductor Development at Baidu, stated, “We are pleased to lead the HPC industry with the successful development of Kunlun. Kunlun was a challenging project aiming for high performance and reliability, and we were able to achieve the desired results through Samsung’s HPC foundry solutions.”
Lee Sang-hyun, Senior Vice President of the Marketing Team at Samsung Electronics’ DS Division Foundry Business Unit, said, “Starting with mobile products, we have now expanded our foundry scope to the HPC field,” adding, “In the future, we will continue to provide comprehensive foundry solutions, including design support through the ecosystem, 5/4 nanometer advanced processes, and next-generation packaging technology.”
Comprehensive foundry solutions covering design, process, and packaging
Samsung Electronics announced that it plans to mass-produce the 'Kunlun' AI chip, based on the 14nm process of Baidu, a major Chinese internet search engine company, in early 2020.
(1).jpg)
▲ Baidu's AI chip KUNLUN <Image=Samsung Electronics>
With this first foundry collaboration, Samsung Electronics is now able to expand its foundry business scope to include AI chips that can be used in cloud and edge computing.
Baidu's Kunlun is an artificial intelligence chip that can be used in various fields of AI ranging from cloud to edge computing. It is a product realized with high performance of 512 GBps bandwidth and 260 TOPS (Tera Operations Per Second, 150W) by applying Baidu's proprietary XPU architecture, Samsung Electronics' 14nm process, and I Cube (Interposer Cube) packaging technology.
Samsung Electronics has improved power (PI) and electrical signal (SI) quality by more than 50% compared to existing solutions by applying a foundry solution optimized for High Performance Computing (HPC).
This improves the noise generated when a signal is transmitted to the chip, and the voltage It is a solution that maintains a constant level to enable the circuit to operate more stably.
I Cube is Samsung Electronics' unique differentiated 2.5D packaging technology that integrates SoC chips and HBM (High Bandwidth Memory) chips onto a silicon interposer (Si-Interposer). This technology has the advantage of increasing transfer speeds and reducing package area by placing each chip within a single package.
Ouyang Jian, Chief Architect and Head of AI Semiconductor Development at Baidu, stated, “We are pleased to lead the HPC industry with the successful development of Kunlun. Kunlun was a challenging project aiming for high performance and reliability, and we were able to achieve the desired results through Samsung’s HPC foundry solutions.”
Lee Sang-hyun, Senior Vice President of the Marketing Team at Samsung Electronics’ DS Division Foundry Business Unit, said, “Starting with mobile products, we have now expanded our foundry scope to the HPC field,” adding, “In the future, we will continue to provide comprehensive foundry solutions, including design support through the ecosystem, 5/4 nanometer advanced processes, and next-generation packaging technology.”
본 기사에 대한 정정·반론·추후보도 청구는 보도 청구 안내를, 그간 게재된 보도문은 정정·반론보도 모아보기를 참고해 주세요.














