The lack of access to EUV paraphernalia in any way, shape, or form has forced China’s SMIC to come up with ingenious ways to leverage its existing DUV equipment to gradually improve its 7nm manufacturing process, and with Huawei’s Kirin 9030, the semiconductor manufacturer has proven that it can stand ‘toe to toe’ with TSMC, or can it?
An in-depth look at the chipset powering the Mate 80 flagship series shows some impressive feats, such as achieving higher density than Intel’s 18A node through a smaller metal pitch, but that’s not the only metric worth comparing. In fact, SMIC and Huawei have ignored various properties when designing the Kirin 9030 that actually matter in a comparison, which has led us to believe that China still has years of catching up to do.
SMIC’s 7nm N+3 metal pitch is smaller than Intel’s 18A when examining the Kirin 9030, but this achievement isn’t without a boatload of trade-offs
The smallest metal pitch that SemiAnalysis and High Yield found on the Kirin 9030 was 32.5nm, making it 10 percent tighter than Intel’s 18A, measuring 36nm on Panther Lake CPUs. What’s also astonishing to see is that SMIC’s 7nm N+3 metal pitch is tighter than TSMC’s N6, with MediaTek’s older Helio G99 shown to measure 40nm.
Keep in mind that N6 uses EUV, summarizing that SMIC successfully designing chipsets with a higher density is clear-cut proof that China’s largest semiconductor manufacturer can progress without the latest chipmaking tools. Unfortunately, this is where the cherry-picking ends as we begin to provide details on all the sacrifices that SMIC made to reach this point.
Through the use of multi-patterning, DTCO (Design Technology Co-Optimization), and complex integration, the Kirin 9030 successfully achieved a tighter metal pitch, resulting in improved area, while SMIC outright ignored or was forced to ignore the efficiency and performance aspects of the silicon.
In fact, Apple’s efficiency cores manage to deliver 20 percent higher integer performance than the Kirin 9030’s prime core, while only drawing 1W compared to the Huawei SoC’s 4.5W usage. While SMIC can obtain bragging rights that its 7nm N+3 process and TSMC’s N6 are identical, the latter is multiple generations old. The Kirin 9030’s prime core matches Cortex-X2 efficiency and performance, but this ARM design was introduced back in 2021, making it five years old.
Looking at this disparity, the Kirin 9030 cannot match the voltage-frequency curve, and the transistor budget allowed to companies like Apple and Qualcomm. With the first 2nm SoCs arriving later this year, that budget will increase further, showcasing TSMC’s superiority over SMIC. Using multi-patterning and DTCO might only allow future Kirin models to surpass the competition in density while sacrificing power and performance.
In short, SMIC is nowhere near close to Intel, Samsung, or TSMC, but Huawei’s LogicFolding packaging might hold the key toward boosting competitiveness, even if it uses a stacking approach to recover density, shorten signal paths, and improve performance.
News Source: SemiAnalysis and High Yield
About the author: Omar Sohail is a reporter and analyst for Wccftech's mobile section, specializing in the technology and business of the mobile industry. His expertise lies in the intricate hardware supply chain, covering developments in semiconductor manufacturing, chip lithography, and camera sensor technology.
Follow Wccftech on Google to get more of our news coverage in your feeds.
