You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Replace the uint/int types with fixed point types. Have them behave like regular int types when the exponent is 0. This will cut down on the number of types we have to support.
Have 2 sets of math operators, one that expands bits to keep all precision (floating-point emulation behavior), and one that keeps bits the same (this should be the default probably).
For now, if fixed point exponent doesn't match, just fail in typechecking.
We may also want another variant, which is saturating ops?
The text was updated successfully, but these errors were encountered:
Replace the uint/int types with fixed point types. Have them behave like regular int types when the exponent is 0. This will cut down on the number of types we have to support.
Have 2 sets of math operators, one that expands bits to keep all precision (floating-point emulation behavior), and one that keeps bits the same (this should be the default probably).
For now, if fixed point exponent doesn't match, just fail in typechecking.
We may also want another variant, which is saturating ops?
The text was updated successfully, but these errors were encountered: