Page 1 of 1

Physics anomaly in iRacing: Interesting read.

Posted: Mon Feb 07, 2011 11:54 pm
by Phathry25
For JLV at least.

http://www.iracing.com/news/blog/q-circ ... f-iracing/

Really though, super impressive that it was found by the end user, exploited, and then solved. How amazing do you have to be at playing video games to notice that every odd numbered lap is faster than an even numbered lap....

Is this something that is happening in MXS? Not that it matters, there's no way anyone can run close enough lap times every lap to ever notice. :mrgreen:

Re: Physics anomaly in iRacing: Interesting read.

Posted: Tue Feb 08, 2011 12:28 am
by Mikey
tl;dr

Re: Physics anomaly in iRacing: Interesting read.

Posted: Tue Feb 08, 2011 12:36 am
by jlv
MX Simulator did have an issue with interpolating between quaternions with opposite signs. Remember when the rider would occasionally flip? That was because it would interpolate around the long way because the sign of the two quaternions being interpolated didn't match.

I'm not sure I get the iRacing bug though. It sounds like they're rounding towards negative infinity. The default mode is generally round to nearest or even. It's the default specifically to prevent problems like that.

MX Simulator uses double precision, so instead of 23 bits of precision you get 53 bits. The round off error is about one millionth of what it would be with single precision. The rounding mode is nearest or even.

Re: Physics anomaly in iRacing: Interesting read.

Posted: Thu Feb 17, 2011 2:12 am
by surfboyrio
Whats your name at iRacing Phatry25?
My name is Daniel S Gomes add me lets race together.. Couple months i dont go for spin, have been busy at work.