penberg/hornet

Support Floating and Fix Point Standards

sirinath opened this issue · 4 comments

Support Floating and Fix Point Standards

@sirinath What does this issue mean exactly?

Hornet obviously needs to implement JVM specification 2.8. ("Floating-Point Arithmetic") which is a subset of IEEE 754 so there's no need to open an issue for that.

What fixed-point standard are you referring to? I'm no aware of any that's relevant here.

JVM does not support full IEEE 754. Also for embedded systems fixed point numbers.

Yes, I just said that the JVM supports a subset of IEEE 754. Why do you need full IEEE 754 support?

As for fixed-point arithmetic, there's no standard to support here.

Decimal numbers have better precision.

Some embedded processors do not have floating point.

Also it would be useful if you are looking to use this for Non Java ByteCode / IR. .Net has some support for decimal. Since these are supported at the processor level at least in intel processors why not support full IEEE 754.