[ODE] Intel/Msvc Compiler

Tom De Muer tom.demuer at skynet.be
Tue Jan 6 11:16:49 MST 2004

Hi Nguyen,

Ok, now I did my homework better and went through a bit of the archives...
it seems that there is already some work done around SIMD'ing the dSetZero
and some functions have been identified as taking up alot of computing power
(the actual solving).  Has this been fed back to the CVS or are there some
results available - I didn't find them in the archives?

> Hi Tom,
> TDM> Well the FPS I'm stating is the theoretically maximum of the
> TDM> nothing else, thus in other words it says that with the same code,
the msvc
> TDM> compiler produces an ode.dll which makes the physics part of my sim
> TDM> at an average of 1/700 s and with the intel at 1/870 s.
>      Well, it's an average on THIS specific case. Could you make sure
>      that your case( how many bodies, how they connected, how many
>      types each, etc...) can represent "normal" cases?
>      Then, I still suggest you profile it. :)

Ok fine, I'll profile it ;o)

My test case is a car :
* chassis is one body : 8 joints (for the suspension)
* 4 wheels : each one body, spheres
* double wishbones: 3 bodies, 2 joints (connecting the wishbones) + 1 joint
for the wheel (depending on steering wheel it is a hinge or hinge2)
* some basic spring / damper computations in the wishbone
Total is 22 hinge-like joints, 17 bodies, 17 geometries
At normal simulation only 4 contacts.
Collision detection: all bodies have geometry connected but the whole car is
one space and the other geometry is basically a cube (5 planes)

Now I'm busy: how would one simulate a flexing chassis?  I thought of using
a ball joint between the chassis parts (fi front and end) and then using
(virtual) springs connecting it at the sides (where the doors are).
Although it may cause the chassis "flexing" it doesn't feel right.  What I
would like to do is based on the relative angle measured along the ball
joint between the two bodies vary the "correcting force".  Where would I
apply that force/torque?

I'm also thinking about simulating a tubular frame with ball joints that's
able to bend just to see how far I could take this... any suggestions there?
Do I face stability problems there?  I remember reading something about too
stiff springs/rigid joints being bad for stability...


> -- 
> Best regards,
> ---------------------------------------------------------------------
>    Nguyen Binh
>    Software Engineer
>    Glass Egg Digital Media
>    E.Town Building
>    7th Floor, 364 CongHoa Street
>    Tan Binh District,
>    HoChiMinh City,
>    VietNam,
>    Phone : +84 8 8109018
>    Fax   : +84 8 8109013
>      www.glassegg.com
> ---------------------------------------------------------------------

More information about the ODE mailing list