Samsung tapes out 3nm mobile processor, closing gap with TSMC

Find the Latest SoC Solutions
for... Automotive... IoT... Security... Audio... Video

Samsung Electronics has completed the design of its 3-nanometer mobile application processor (AP) and successfully achieved its first tape-out through its foundry division. Tape-out is the final step in preparing for mass production, transitioning from design completion to manufacturing. With this advancement, Samsung has reduced the process transition time. With this advancement, Samsung has reduced the process transition time gap to just six months behind TSMC, which supplied the 3nm A17 chip to Apple's iPhone 15 Pro last year.

www.chosun.com/english, May. 08, 2024 – 

According to industry sources on May 8, Samsung Electronics has successfully taped out its 3nm system-on-chip (SoC), which integrates mobile CPUs, GPUs and others, in collaboration with global electronic design automation (EDA) company Synopsys. The company is set to begin mass production later this year, aiming to feature these chips in the next generation of its Galaxy smartphone series by early next year.

The mobile AP is a core semiconductor that serves as the brain of a smartphone, accounting for about 20% of the total bill of materials (BoM) of a smartphone. The 3nm mobile APs are significantly more expensive than previous generations as they utilize cutting-edge technology. The price of TSMC's mobile APs produced using the 3nm process is reportedly around $20,000 (about 27 million won) per wafer.

Samsung Electronics is the first in the industry to apply gate-all-around (GAA) technology in the 3nm process. GAA technology enhances performance and reduces the operating voltage of the semiconductor by ensuring the transistor's channel and gate contact on all four sides, compared to the three sides in the traditional fin field-effect transistor (Fin-FET) approach.

However, doubts remain about the performance improvement of Samsung's 3nm mobile AP. According to a company staffer, while it has achieved a 10% performance increase over previous models, details such as the number of transistors are confidential, and it is unclear how much the power efficiency has improved–a major industry focus with the rollout of on-device AI.

click here to read more...

 Back