The authors present very accurate and fast algorithms for summation and dot product of floating-point numbers. The main feature of these algorithms is that they use only basic floating-point operations addition, subtraction and multiplication and the same working precision as the data. Higher precision is unnecessary. So, in the future, hardware implementation and standardization of error-free transformations could be possible. Peculiar numerical examples for extremely ill-conditioned problems are discussed in terms of timing and accuracy. Comparisons with other known algorithms show how effective the presented algorithms are.