Abstract: Whether the high-speed growth of patent calculated knowledge capital really have the potential to boost long-run economic growth and whether it has significant and positive effect on real economic growth are two main problems concerned by scholars and discussed in this paper. Based on assumptions of knowledge production function promoted by Romer and Jones in endogenous growth theory, this paper estimates parameters of the knowledge production function using Chinese knowledge capital calculated by granted patents in 30 provinces during 1996 and 2014. We find that the endogenous accumulation of knowledge capital has not been strong enough to breakthrough the long-run economic growth rate constrains and support Chinese economy to grow at such high speed in the long run. In addition, this paper estimates the actual promotion effect of knowledge capital on real GDP growth using fixed-effect semi-parametric method in different time periods. We find that with patent system perfecting knowledge capital will effectively boost economic growth. And the high speed growth of knowledge capital has limited promotion effect. Once the knowledge capital grow too fast for the patent transformation capacity, knowledge capital do little to promote economic growth. Knowledge capital calculated by invention patents, which represent high quality part of knowledge set, appears to boost economic growth more significantly than patents in the whole. We suggest that, the intellectual property policies should emphasis more on the quality rather than the quantity of patents to reinforce the effect of knowledge endogenous accumulation. The government should reduce intervention policy which focus on short-term patent quantity growth and emphasis more on market resource allocation mechanism. We should highlight the importance of patents commercialization and industrialization by formulating the co-operation innovation network.