[ODE] dt computation

[gl]

The seed & FPU flags (also FYI D3D changes the flags behind your back so the 
usual fix was to set them at the beginning of every simulation step).

But IIRC it's impossible to get 100% identical results even on the same 
platform necessarily, see this old mail:
http://q12.org/pipermail/ode/2005-March/015451.html
--
gl

----- Original Message ----- 
From: steve1011
To: ODE Mailing List (E-mail)
Sent: Tuesday, April 04, 2006 5:23 PM
Subject: Re: [ODE] dt computation


There was discussion a while ago about math difference between PowerPC and 
Intel processors that showed up between OSX and others. On the same cpu, I 
would hope there were no differences due to linux/win. It should be the same 
FPU; unless there are differences between Intel and AMD or within cpu 
families. What about the random number seed? Could that account for the 
behavior?

Steve

Bram Stolk wrote:
J. Perkins wrote:

  It doesn't go through the basket?
      No, it bounces off the top.

Ah...

Well, I put that down to slight numerical deviation between
linux and win32 implementation of libmath funcs I guess.
Or is ODE fully ieee math compliant on both platforms?

I suppose you have already too much to do, but in the
unlikely case you have some spare time, we could opt for
finding a proper release height (sz) that enables a score
for win32. Then we can simple cheat and do a ifdef WIN32
on the fall-height, and have the score anyway. My guess is
that the fall-height should be just a little bit lower for
win32, if it bounces off the back of the basket.

   bram

_______________________________________________
ODE mailing list
ODE at q12.org
http://q12.org/mailman/listinfo/ode



--------------------------------------------------------------------------------


_______________________________________________
ODE mailing list
ODE at q12.org
http://q12.org/mailman/listinfo/ode