Variance is available in Intel Summary Statistics Library. The library also contains algorithms for computation of moments up to 4th order (central and algebraic), skewness, kurtosis, quantiles, and variance-covariance/correlation matrices. Support of missing values, algorithms for robust estimation of mean/covariance, and detection of outliers are also included into the library. The estimates are computed in single and double precision. The package is available for downloading at http://software.intel.com/en-us/articles/intel-summary-statistics-library

For sum of squares, use BLAS level 1 functions ?dot. Dot product of a vector with itself will give you sum of squares.

Writing it with standard language features is likely to be more efficient: dot_product(), inner_product, explicit loops.