椭圆曲线密码细粒度并行计算研究

根据加速经常性设计的原则，提出了一种基于对称运算单元的椭圆曲线密码(ECC)标量乘运算的高效细粒度并行运算架构.为了实现该架构，对ECC标量乘运算展开细粒度并行计算研究，通过标量乘运算的分解和推导，消除了数据相关性，得出运算效率高且适于指令级并行计算的算法形式.对标量乘运算的时间复杂度的分析结果表明，该算法比普通算法的速度提升了66.7%.并可通过并行计算进一步提升标量乘运算的速度性能.在采用3个运算单元的效率最优情况下，比采用1个运算单元时，速度提高了2倍.

Research on fine-grained parallel computation for elliptic curve cryptography

According to the design principles of making common case fast, a kind of fine-grained parallel computation architecture with high efficiency based on symmetrical operation unit was proposed, which was for implementation of scalar multiplication of Elliptic Curve Cryptography （ECC）. To complete the architecture, fine-grained parallel com- putation architecture was focused on. The algorithm form was obtained conforming to instruction parallel computa- tion and avoiding data dependency. The time complexity analysis results showed that, based on the introduced algo- rithm, the speed performance of scalar multiplication can be raised by 66.7% compared with the common method. And it can still be raised more by applying with parallel computation. In the occasion of using three operation units, which owned the best optimal efficiency, the speed can be raised 2 times of that using one operation unit.