lucidium said:
Most modern physx enabled cards can handle 20 billion instructions per second, which for a game like this would be solely applied to water physics and object physics rather than extensive game wide physics, as such 20 billion instructions on a modern pc gpu would equate on a game such as this to around 700fps precision, dropping to 300-400 on other types of games, dropping further still down to fixed frame (one calculation per video refresh) on open world sandbox games employing a physics model. To put it bluntly, 240fps resolution on physics modeling for a game like this isn't really a huge deal, i suspect its only being calculated and trumpeted for marketing purposes due to the kerfuffle of framerates and resolutions currently rife in the industry media at current, underlined by their focus on advertising the texture resolution and native video resolution too. Frankly, if they dropped the physics precision down to 60fps and locked video to 60fps you would see very little difference because you're still getting a calculation per each of the frames that calculation have an effect anyway. |
So in your experience there is little benefit to calculating the physics more times than you have frames being shown? That's surprising to me since I know with the basic modeling my lab does even on simple problems the step size matters pretty massively up to a point. Simulating the kinetics of a chemical reaction with 1000 steps versus 100 can have a large effect in certain conditions. I had assumed that most games had more physics refresh frames than the displayed frames per second to make the physics at the point of display as accurate as possible, but I guess since the differences in most games would be totally imperceptible that's not worth the effort.
Interesting to know, thank you for te reply.
...
