Implication for zvfh

[eopXD]

Just noticed that 18.5. Zvfh: Vector Extension for Half-Precision Floating-Point mentions

The Zvfh extension depends on the Zve32f and Zfhmin extensions.

Should there be an implication for Zvfh to zvfhmin?

[aswaterman]

It doesn’t need to. It’s already fully specified without that implication. Instructions can appear in multiple unrelated extensions, so either depending on Zvfhmin or not depending on it is valid.

Note also that Zfh doesn’t depend on Zfhmin; those two are analogously independent.

[eopXD]

I was thinking of this because the Zve* extensions seems to have the implications to its subset (e.g. zve32f implies zve32x), and Zvfhmin is a proper subset of zvfh.

Is there a reason behind this asymmetry?