'Improve floating point determinism physx
For PhysX, I want to improve float determinism. One way of doing this is using soft floats or more precise math. I did not see whether other people tried this before.
Turns out there is a PhysX compile flag 'PX_FLOAT_POINT_PRECISE_MATH' for VS16 (windows) platform. This will set the '/fp:percise' flag for vs16. However, for Linux targets, they use 'clang' as a compiler which doesn't support this/or a similar flag.
Both VS16 and GGC seem to be able to be configured with some improved float precision. For GGC there is a whole bunch of flags here.
How do I get something like /fp:precise from VS16 1 for the clang compiler? Also, do you think this approach can enhance floating point determinism?
Sources
This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.
Source: Stack Overflow
| Solution | Source |
|---|