Santosh NagarakatteJust committed new versions of correctly rounded sinf, cosf, tanf, asinf, acosf, and atanf that are about 20-30% faster than our previous versions. #RLIBM https://github.com/rutgers-apl/The-RLIBM-Project
Here is a talk summarizing our RLIBM project, where we are building correctly rounded math libraries. https://www.youtube.com/watch?v=vAcf6d26kiM #RLIBM #RutgersCS
A Case for Correctly Rounded Math Libraries
रञ्जित (Ranjit Mathew)Interesting:
“High Performance Correctly Rounded Math Libraries For 32-Bit Floating Point Representations”, Jay Lim and Santosh Nagarakatte, SIGPLAN (https://blog.sigplan.org/2021/08/26/high-performance-correctly-rounded-math-libraries-for-32-bit-floating-point-representations/).
Via HN: https://news.ycombinator.com/item?id=28318425
#Math #Precision #NumericalAnalysis #FloatingPoint #Rounding #RLIBM
High Performance Correctly Rounded Math Libraries for 32-bit Floating Point Representations
Everyone uses math libraries. Surprisingly, mainstream math libraries do not produce correct results for several thousands of inputs. Developers are seldom aware of them, which affects reproducibil…