Write a Blog >>
Thu 21 Jan 2016 17:20 - 17:45 at Grand Bay North - Track 1: Optimization Chair(s): Mayur Naik

Floating-point numbers are an essential part of modern software, recently gaining particular prominence \r\non the web where floating- point numbers are the exclusive numeric format in Javascript. To use floating point numbers, we require a way to convert the binary machine representations into decimal human readable outputs. Existing conversion algorithm make trade-offs between completeness and performance. The classic Dragon4 algorithm by Steele and White and its later refinements achieve completeness – i.e. produce correct and optimal outputs on all inputs – by using arbitrary precision integer (bignum) arithmetic which leads to a high performance cost. On the other hand, the recent Grisu algorithm by Loitsch shows how to recover performance by using native integer arithmetic but sacrifices optimality for 0.5% of all inputs. We present Errol, a new complete algorithm that is guaranteed to produce correct and optimal results for all inputs, while simultaneously being 2x faster than Grisu and 8x faster than previous complete methods. We formalize and describe Errol and present an implementation and experimental evaluation of our method against the state-of-the-art.

Thu 21 Jan

16:30 - 17:45: Research Papers - Track 1: Optimization at Grand Bay North
Chair(s): Mayur NaikGeorgia Tech
POPL-2016-papers16:30 - 16:55
Somashekaracharya G BhaskaracharyaIndian Institute of Science and National Instruments, Uday BondhugulaIndian Institute of Science, Albert CohenINRIA
Media Attached File Attached
POPL-2016-papers16:55 - 17:20
Wenlei Bao, Sriram KrishnamoorthyPacific Northwest National Laboratories, Louis-Noel PouchetOhio State University, Fabrice RastelloINRIA, France, P. SadayappanOhio State University
Media Attached
POPL-2016-papers17:20 - 17:45
Marc AndryscoUniversity of California, San Diego, Ranjit JhalaUniversity of California, San Diego, Sorin LernerUniversity of California, San Diego
Media Attached