syenkoc commented on PR #232:
URL: https://github.com/apache/commons-math/pull/232#issuecomment-1540153183

   OK thanks for the review, I should have time this weekend to make the 
required code changes. 
   
   In terms of performance, I tested this using a hacked up version of the old 
FastMathTestPerformance class. This new implementation definitely faster 
without the "fast" case code, about 30%.
   
   Here is what I get, comparing StrictMath, Math and the two AccurateMath 
implementations across 200000000 runs:
   
   Without "fast" case      
   StrictMath 2.797007    1.0 
   AccuratMathOld 3.641469 1.3019 
   Math 2.750160 0.9833 
   AccurateMathNew 1.4190 0.5073
   
   With "fast" case
   StrictMath 2.778882    1.0 
   AccuratMathOld 3.590121 1.2919 
   Math 2.709783 0.9751 
   AccurateMathNew 2.4186 0.8704
   
   These tests were done on a 1.2GHz Intel Broadwell processor from circa 2015. 
I dumped the JIT-compiled code, but it's a bit hard to make sense of why it is 
faster.
   
   I am unfamiliar with JMH, but I will try this weekend to get it going. 


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to