Re: DRM removal soon

On Thu, Feb 28, 2019 at 03:53:53PM -0600, Doug Kirk wrote:
> On Thu, Feb 28, 2019 at 3:07 PM Steve Kargl <
> [email protected]> wrote:
> 
> >
> > Seems our experiences are exact opposites.  :(
> >
> > I suppose it is the bane of those of who cannot afford
> > new hardware every 2 or 3 years.
> >
> >
> Hmm, because amd64 has only been available for the last 2 or 3 years? It
> was added to FreeBSD as experimental in 5.1-RELEASE (June 2003) and
> standard in 5.2-RELEASE (January 2004).

I have a amd64 system with

CPU: AMD FX(tm)-8350 Eight-Core Processor  (4018.33-MHz K8-class CPU)
Origin="AuthenticAMD"  Id=0x600f20  Family=0x15  Model=0x2  Stepping=0

Guess what?  Yep, it uses drm-legacy-kmod because the radeon video
card (drmn0: <Caicos [Radeon HD 6450]>) I have is sort of not supported
by drm-current-kmod.  When I tried to use drm-current-kmod, I watched
/var/log/message fill up with some error message about waiting for a
mfence memory barrier.  I was advised by the graphics team to use
drm-legacy-kmod.

Sure, I can just run out an buy a shiny new video card, which will
be obsolescence in a few years and fall off the support video list. 
At which point, I can rinse and repeat.

> Jeff Atwood commented on desktop adoption of x86-64 in 2007 here:
> https://blog.codinghorror.com/is-it-time-for-64-bit-on-the-desktop/ , and
> it even mentions that graphics cards don't like 32-bit limits.
> 
> If you bought an x86-32 system within the last 2-3 years, why is that
> anybody else's fault that your graphics-based computer was obsolete when
> you bought it?

You have completely missed the point.  I didn't buy the laptop
in the last 2 or 3 years.  It is several years old.  The system
was working just fine until the PAE vs non-PAE merge was committed
(which surprise effects i386 systems).

It is the system on which I work out the bits for libm, where the i387
FPU is set such that long double only has 53-bits of precision.  Yep,
I know, no one uses i386-class systems for numerical work.  Given the
condition of libm and the crickets on freebsd-numerics@, it seems very
few people do any numerical research with freebsd.  

> Please stop whining on the mailing list, you're just adding noise by
> setting up straw-man arguments.

I suppose pointing out that "the emperor has no clothes" can
be considered whining.  When I commit a patch to the gcc tree,
if it affects others, I work to fix the problem.  I don't shoot
the messenger.  At one point in time, FreeBSD developers cared
about the quality of their work.

-- 
Steve
_______________________________________________
[email protected] mailing list
https://lists.freebsd.org/mailman/listinfo/freebsd-x11
To unsubscribe, send any mail to "[email protected]"