On 2026-01-10 14:20, Richard Henderson wrote:
> I suggest you pass that as a bool parameter to bfloat16_to_float8_e4m3 etc.
> This would then be handled as part of round-and-pack, maybe a separate step,
> maybe via float_round_nearest_even_max.
> 

Thanks for the suggestion!
Will replace the format saturate flag by a bool parameter of convert
function at v2.

> I'm not sure what to do with arm_althp vs ocpfp.  It seems like they have a
> couple of things in common.  Perhaps we should decompose these to separate
> behavior flags.
> 
> 
> r~

I agree that we should decompose these to seperate behavior flags. I
will try to introduce some behavior flas (liked no infinity, maximum
normal pattern, etc.) at v2.

rnax

Reply via email to