SemiAnalysis deeply dissects Huawei Kirin 9030 Pro: architecture and performance significantly improved, China's advanced process development not hindered According to BlockBeats, on June 15th, SemiAnalysis Steellab released a report on the in-depth disassembly of the Kirin 9030 Pro equipped on Huawei Mate 80 Pro, confirming that it uses SMIC's N+3 process and has a minimum metal spacing of only 32.5nm, which is about 10% tighter than Intel's 36nm. Through radical DUV multi patterning and DTCO optimization, N+3 achieved a logic density comparable to TSMC N6 (approximately 113.4 MTr/mm ², slightly higher than N6's 107.7), but at the cost of higher process complexity and more difficult cost and yield control. The Kirin 9030 Pro chip architecture and performance improvement, which is an evolved version of Kirin 9020, has a similar die area but higher utilization: adding 1 core, increasing GPU CU from 4 to 6, increasing NPU Tiny cores, and expanding cache. The CPU/GPU/NPU core area has significantly reduced (core reduction of 22%, GPU CU reduction of 28%), and the performance has improved significantly compared to the previous generation (GPU 3DMark improvement of 70-79%), but overall it still lags behind current flagships (Apple, Snapdragon 8 Elite, etc.), and the energy efficiency gap is even greater, approaching the level of Android flagships three years ago. The SemiAnalysis report points out that export controls have not prevented China from advancing advanced processes, but have forced SMIC to take a more complex DUV route, relying on tighter design rules, backside power supply, and Huawei's LogicFolding stacking technology in the future. N+3 demonstrates the resilience of China's semiconductor industry, but there is still a generation gap compared to leading nodes, with maturity and cost being the main limitations. ---------------------------------Click on the original link below to join the Beating · Feishu AI news channel, which continuously monitors global AI hotspots and news 24/7. [Original link]