[Libre-soc-isa] [Bug 1057] split out all int/fp min/max ops into their own RFC ls013

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=1057

--- Comment #38 from Jacob Lifshay <programmerjake at gmail.com> ---
(In reply to Luke Kenneth Casson Leighton from comment #37)
> (In reply to Jacob Lifshay from comment #36)
> > I updated openpower-isa.git to use the new fminmax instruction, and removed
> > fminmaxs while I was at it.
> 
> oh! err hang on - that needed some discussion rather than an
> arbitrary decision.

i thought we already discussed it...

the *s ops are only necessary where they behave differently than the
corresponding bfp64 ops, min/max do not behave differently therefore are
redundant.

for a somewhat more detailed explanation see the programming note in the right
column of page 522 (page 548) of powerisa v3.1b

> looking at pages 734 onwards in Public v3.1
> there are a considerable number of different precision min/max
> operations.

only double/quad precision -- those are not redundant because their
input/outputs are encoded differently (bfp64 vs. bfp128). the corresponding
single ops would use the bfp64 encoding so can be eliminated due to redundancy.

-- 
You are receiving this mail because:
You are on the CC list for the bug.