I guess it’s not a big surprise but now it’s official…
This entry was posted
on Monday, March 10th, 2008 at 8:29 am and is filed under Physics.
You can follow any responses to this entry through the RSS 2.0 feed.
Both comments and pings are currently closed.
I’ve been following the whole Physics Demos => Novodex => Ageia => Nvidia adventure, and even if it is -stupidly- easy to say that now, I never believed in the Ageia Hardware.
But I strongly believed (and I still) in the software library (except that you are no longer part of the team)
When ageia acquired novodex some years ago, I was thinking “oh no, those clever guys of novodex have now to deal with business guys imposing wrong decisions…, they must have hard times trying to use this piece of hardware and trying to find a way to show a convincing hardware speedup”
Perhaps I’m wrong, but life seems sweeter in barcelona
Ps: following your work since years, congratulations, and thanks !
You’re pretty much right about the business guys and the wrong decisions. More than once I worked on improvements & new features that never made it into the Software SDK because the hardware would not support them. Like, I don’t know, support for large worlds, better CCD, etc. It was very frustrating. The hardware also introduced a lot of bloat in the code: extra layers of indirection, double buffering for some data, the notorious “mirror manager”, etc. Worse: some features like Featherstone joints got removed from the codebase, simply because the code was too complex for the hardware. Instead we switched to a unified, HW-friendly “6 DOF” joint, which is a major PITA to work with if you ask me. On the other hand it was nice to work with the guys from Mathengine/Meqon, and share the knowledge.
March 11th, 2008 at 2:32 pm
…finally the closing of the loop
I’ve been following the whole Physics Demos => Novodex => Ageia => Nvidia adventure, and even if it is -stupidly- easy to say that now, I never believed in the Ageia Hardware.
But I strongly believed (and I still) in the software library (except that you are no longer part of the team)
When ageia acquired novodex some years ago, I was thinking “oh no, those clever guys of novodex have now to deal with business guys imposing wrong decisions…, they must have hard times trying to use this piece of hardware and trying to find a way to show a convincing hardware speedup”
Perhaps I’m wrong, but life seems sweeter in barcelona
Ps: following your work since years, congratulations, and thanks !
March 12th, 2008 at 2:39 am
You’re pretty much right about the business guys and the wrong decisions. More than once I worked on improvements & new features that never made it into the Software SDK because the hardware would not support them. Like, I don’t know, support for large worlds, better CCD, etc. It was very frustrating. The hardware also introduced a lot of bloat in the code: extra layers of indirection, double buffering for some data, the notorious “mirror manager”, etc. Worse: some features like Featherstone joints got removed from the codebase, simply because the code was too complex for the hardware. Instead we switched to a unified, HW-friendly “6 DOF” joint, which is a major PITA to work with if you ask me. On the other hand it was nice to work with the guys from Mathengine/Meqon, and share the knowledge.