Hi,
We are using the Sitara AM4377. The CPU includes a NEON and vector floating point (VFPv3) coprocessors. We noticed that after a single precision (double precision was not tested) floating point division by 0, the code sometimes hangs (we are using OS), and if the code doesn’t hang, there are cases where after the division by 0, we get 0 as a result of further floating point calculations even though we expect non-zero results, and this behavior is consistent. We have the following questions:
- Is the described behavior normal?
- Which coprocessor executes the floating point calculations, the NEON or the VFP coprocessor?
- Is there a way to capture or to handle floating point division by 0, something like exception handler?
- Is there a document that describes what happens upon floating point division by 0?
Thank you in advance.