3 ms·
Maybe the parent poster is talking about double-double arithmetic: https://en.wikipedia.org/wiki/Quadruple-precision_floating-point_format#Double-double_arithm
by johndough 6y ago
Maybe the parent poster is talking about double-double arithmetic:
https://en.wikipedia.org/wiki/Quadruple-precision_floating-point_format#Double-double_arithmetic https://en.wikipedia.org/wiki/Quadruple-precision_floating-p...
Paper:
http://web.mit.edu/tabbott/Public/quaddouble-debian/qd-2.3.4-old/docs/qd.pdf http://web.mit.edu/tabbott/Public/quaddouble-debian/qd-2.3.4...
- guyomes 6y agoDouble-double and quad-double arithmetic have even been implemented for the GPU as research prototype libraries (gpuprec [0,1] and campary [2,3] among other). [0] https://github.com/lumianph/gpuprec https://github.com/lumianph/gpuprec [1] https://event.cwi.nl/damon2010/gpuprecision.pdf https://event.cwi.nl/damon2010/gpuprecision.pdf [2] https://homepages.laas.fr/mmjoldes/campary/ https://homepages.laas.fr/mmjoldes/campary/ [3] https://hal.archives-ouvertes.fr/hal-01312858/document https://hal.archives-ouvertes.fr/hal-01312858/document
- tomsmeding 6y agoInteresting! Didn't know about that, makes sense.