A Fast Implementation of Matrix-matrix Product in Double-double Precision on NVIDIA C2050 and Application to Semidefinite Programming | IEEE Conference Publication | IEEE Xplore