Exact accumulation of floating-point numbers | IEEE Conference Publication | IEEE Xplore