In this paper, we present a high-performance elliptic curve cryptographic architecture over SCA-256 prime field by introducing a one-cycle full-precision multiplier. Based on the multiplier, we give a thorough bottom-up optimization in algorithm level. The performance of the architecture is boosted by the use of a two-stage pipeline scheme, and our pipeline utilization reaches 100%. It takes only 8 cycles to perform point doubling and 12 cycles to perform point addition. Using NAF recoding of scalar k, it takes about 3333 cycles to complete the point multiplication operation. In the hardware evaluation using a 0.13 mum CMOS standard cell library, our high-performance SM2 architecture executes one point multiplication operation in 20.36 mus, which translates into more than 49000 point multiplications per second. To the best of our knowledge, our architecture offers the highest single-core performance over prime fields reported in literature up to now.