Re: Softfloat on i386

[Greg Troxel <gdt%lexort.com@localhost>]

David Holland <dholland-pkgtech%netbsd.org@localhost> writes:

> The problem is that go is a compiler and it apparently only supports
> codegen for either (a) sse2 and newer or (b) softfloat. This means
> there's a range of non-486SX (and non-486) machines where the only
> working option is softfloat.

I'd call that a bug in go.

> Given the performance cost of softfloat and that we're talking about
> very old machines at this point, i don't think it should be the
> default for "i386" (as in 32-bit x86), but it should be an option and
> somewhere in the setup docs we should tell people to set it if they're
> using such hardware. (Also it might be worth making go builds fail if
> it's not set when on such a machine.)

I agree that we should have an option and that it should default to the
current sse2 code.

> This does to some extent raise the question of what the intended
> purpose of the 32-bit x86 binary builds is these days...

I see multiple cases:

  Machines that haven't been converted over to amd64 even though they
  could be.  I have 1 of these, originally installed in the mid-2000s,
  still not amd64 for reasons I won't explain but are valid if you
  understood the details, and I know of another such system.

  Machines that were modern in 2003 but don't quite run amd64.

  Semi-restricted CPUs for embedded/SBC.  I have an apu2 which has
    cpu0: AMD GX-412TC SOC                               , id 0x730f01
  which runs amd64 but my previous one was a 586-ish geode.

  Then there's true 486 from the 90s, which if desktoppy I sort into
  retrocomputing.