🚀 go-pugleaf

Spalls Hurgenson  <spallsh...@gmail.com> wrote:
>These days we have (almost) achieved this goal, so these fevered
>imaginings seem almost prescient to modern eyes. Flying across Los
>Santos in Grand Theft Auto V and it's difficult to believe we
>/haven't/ achieved this promised level of fidelity. Of course, most of
>the buildings are empty shells and the AI fades in and in an
>existential bubble around the protagonist, but it's a beautiful
>illusion. Of course, in 1996 I would believed we would have achieved
>all this by 2005...

Well, it was achieved by then.  Grand Theft Auto III did all that in 2001,
except with lower detail textures and models.  In 1996 the graphics
in GTA 3 would've looked amazing, and probably would have been called
"photorealistic" by most people.  You don't hear that term being used
much used more, as it's become obvious what looks like we've almost
achieved in the latest AAA game is in fact a moving target.  In 1996 in
that moving target was extreamly low by today's standards.  If you were
shown GTA 3 in 1996 you'd be saying similar things to how you described
GTA V above.

>[...] Similarly, I completely missed out on the
>potential of 3D-rendering, a technology still in its infancy at the
>time (seriously, at the time 3D graphics cards were a lot like VR tech
>is today; nifty graphics but who knows if they were going to really
>catch on?). It was - I thought - the ever increasing CPU speeds that
>were going to create this gaming revolution, and while fast CPUs can
>structure detailed worlds, they aren't that great at pushing pixels.

Well, the difference between 3D graphics cards in 1996 and VR tech today,
is that in the former case the demand was clearly there.  3D graphics
itself clearly wasn't a fad, it was everywhere.  Games that weren't 3D
were already starting to look primitive. (Consider 1995's Stonekeep,
a dungeon crawler in the style Wizardry an Bard's Tale, but with much
better graphics.  Graphically it was the pinnacle MS-DOS era "2D" dungeon
crawlers, but in 1995 it looked dated compared to Ultima Underworld.)

The question was whether or not 3D hardware accelerators would be able
to meet that demand, if they could out perform CPUs enough to make them
worthwhile.  If you were familiar enough with the technology of the time,
like the 1995 graphics chip in the original PlayStation and 3dfx's Voodoo
chip used in arcade games in 1996 (but not in PCs until the end of 1996),
then you'd know that they had a huge advantage over CPU rendering.
Most PC gamers though would have to make something of leap of faith to
say that 3D acceleration was the future, although not anywhere near as
much as someone saying VR is future today.

>So, yeah; my predictions were a bit off target. Then again, so were
>pretty much everyone's in 1996. 

I was pretty good at making predictions.  I remember saying that PC games
were a couple of years away from being regularly released on CD-ROM
and then later saying the same thing about PC games and DVD-ROMs.
It was all about the price of the optical drives.  They had to get
into the $100-$200 range before they'd be standard in all new PCs and
it wouldn't until then that market penatration would be high enough for
PC game publishers to make the switch.

>But I don't make predictions about where gaming technology is headed
>anymore.

There's much less need to these days.  Computer technology is becoming
more like airplane technology, its getting close to its limits.
Modern commercial airplanes all cruise at the around same speed, about
Mach 0.85, because they can't be made to go faster without becoming a
lot less efficient.  CPU speeds have already hit a wall around 4GHz,
and even Intel has had to admit Moore's Law has slowed down.  We're not
too far away from the point were it won't be possible to add more gates
in order to squeak out 5% IPC (instructions per cycle) gains or put more
cores on the same die that most software can't effectively use anyways.
3D accelerators.

About the only computer technology that's still scaling like it used
to is 3D graphic accelerators.  That's partly because they're behind
CPU techology (eg. NVIDIA only started making 14 nm chips in October
2016, while Intel first 14 nm CPU was released in August 2015), but
also because they do actually benefit by craming ever more cores into
the the same die.  However even that benefit isn't as big as it appears
on paper because games already have trouble making full use of the GPU
resources.  (Games that cause video cards to overheat during their menus
but not during the actual game demonstrate this problem.  The menus are
simple enough that the GPU can reach almost 100% utilization and stay
there. During game play average utilization is significantly lower,
so the card doesn't get as hot.)

-- 
 l/  //	  Ross Ridge -- The Great HTMU
[oo][oo]  rri...@csclub.uwaterloo.ca
-()-/()/  http://www.csclub.uwaterloo.ca/~rridge/ 
 db  //