 |
|
|
|
|
|
| |
| |
|
|
|
|
| |
| |
|
|
nemesis wrote:
> Chambers <Ben### [at] gmail com> wrote:
>> Invisible wrote:
>>> Chambers wrote:
>>>> 1) Support for sophisticated branching
>>> When this happens, the GPU will be exactly the same speed as the CPU.
>>> The GPU is fast *because* it doesn't support sophisticated branching.
>> That's too bad, because POV requires sophisticated branching.
>
> yeah, I wonder how a path tracer, which requires more branching for each new ray
> spawned than a conventional raytracer, did it...
>
If you wonder, I recommend reading their code and finding out. Then you
will see how it does not work past spheres and triangles.
If that was sarcastic, then you just managed a branching statement and
you must not be a GPU.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
> On 16-1-2010 16:42, nemesis wrote:
>> troll
Now, that explains everything.
andrel wrote:
> I am not fed up. A bit tired of repeating the same discussion over and
> over. We try to spread the load by alternating who is answering this
> time*. You just wait until it is Warp's, or even better Thorsten's,
> turn, then it gets really funny.
>
>
> * nothing formal, just the way things develop.
Now I see why the standard reply is so harsh. I have tried, repeatedly,
when this topic comes up to offer actual reasons, enough key words and
phrases that anyone could google for and learn from. Most of the time,
that seems to work. The person who thinks about GPGPU knows enough to
pick out those terms, and find the information themselves. Here, how
ever, we have hit the wall in that it appears nemesis does not want to
know why he is wrong, and only wants to complain. Humor and/or killfiles
are all I have left.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
nemesis wrote:
> Chambers escreveu:
>> I can't recall POV-Ray ever being one of the fastest raytracers around.
>
> I can recall it being used as a useful benchmarking tool once it got
> multicore.
So can I, but what has that to do with whether or not POV is the fastest
raytracer around?
> In any cases, you get it wrong: the amazing speed up is not due to
> native GPU triangle handling, but faster ray intersection calculations
> thanks to the GPU sheer parallel vector processing. You can see that in
> smallptGPU where you get perfect math spheres much faster, no meshes in
> sight.
Yes, I'm familiar with the architecture. However, that was one specific
instance of how other programs compromise quality where POV doesn't.
It's not the only way.
Other programs, for instance, also do some or all of the following:
tesselate objects, make liberal use of floats, use fixed textures rather
than procedural, and handle lighting via precomputed lightmaps. This
list is by no means exhaustive nor definitive.
If POV were to accept such optimizations, then it, too, could be made to
run well on upcoming GPUs (I still doubt the current crop would do well).
...Chambers
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Darren New wrote:
> scott wrote:
>> Because nothing else will come along with a nice SDL like POV has.
>> Almost all other raytracers *require* you to use an external mesh
>> modeller to generate your scene - POV doesn't which IMO is its
>> strongest point.
>
> That, and excellent procedural textures, is why I'm interested in it. :-)
>
Honestly, that's one of my favorite parts :)
...Chambers
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Orchid XP v8 wrote:
> Damn, I don't even want to know how many pages you'd have to scroll
> through to get to the part that you actually want...
Interestingly enough, it was crap in the first few and last few pages that I
wanted to change. (I.e., the commands setting up the tables and such.) The
data itself was cool.
--
Darren New, San Diego CA, USA (PST)
Forget "focus follows mouse." When do
I get "focus follows gaze"?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Sabrina Kilian wrote:
> If you wonder, I recommend reading their code and finding out. Then you
> will see how it does not work past spheres and triangles.
which is why, *again*, I have to point out I was hinting at using the
GPU solely to speed up povray's ray-triangle intersections.
Is there any trolling in that? Wishing to make povray readily more
usable outside its small math geek audience?
Outside that small niche, no one cares about perfect abstract solids or
slow-rendering math surfaces. Is it trolling to ask one to wake up and
smell the coffee?
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
On 17-1-2010 20:08, nemesis wrote:
> Sabrina Kilian wrote:
>> If you wonder, I recommend reading their code and finding out. Then you
>> will see how it does not work past spheres and triangles.
>
> which is why, *again*, I have to point out I was hinting at using the
> GPU solely to speed up povray's ray-triangle intersections.
Because it is like pointing out that there are ways to increase the
speed of a car that only works on highways without realizing that it
would mean that you can not use the car anywhere else. That won't work
because, although there may be people living next to a highway that want
to go to work next to the highway, in general it will make the car as a
general means of transportation useless.
> Is there any trolling in that? Wishing to make povray readily more
> usable outside its small math geek audience?
It is not a small and definitely not a math or a geek audience.
> Outside that small niche, no one cares about perfect abstract solids or
> slow-rendering math surfaces. Is it trolling to ask one to wake up and
> smell the coffee?
It becomes trolling when you do that over and over without taking the
previous remarks into account. If you take your time and research those
remarks and show that they are invalid that would be OK. Just repeating
what you think should be done by others is not OK.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
andrel wrote:
> On 17-1-2010 20:08, nemesis wrote:
>> Sabrina Kilian wrote:
>>> If you wonder, I recommend reading their code and finding out. Then you
>>> will see how it does not work past spheres and triangles.
>>
>> which is why, *again*, I have to point out I was hinting at using the
>> GPU solely to speed up povray's ray-triangle intersections.
>
> Because it is like pointing out that there are ways to increase the
> speed of a car that only works on highways without realizing that it
> would mean that you can not use the car anywhere else.
Sounds about right to me. I'm not complaining that a is to narrow for
my car if I can always take another road.
>> Is there any trolling in that? Wishing to make povray readily more
>> usable outside its small math geek audience?
>
> It is not a small and definitely not a math or a geek audience.
It's certainly nowhere near as large as in the CG industry. Here's a
good place to start:
http://forums.cgsociety.org/
Quite a bit more active than even p.o.t, huh? You may search for
pov-ray there.
And if it's not math or geek focused, why it is most of the images in
p.b.i abstract math, RSOCPs or isosurface terrains? There certainly are
too a lot of truly beauty images depicting real-world objects and scenes
-- mostly as triangle meshes, some as painfully constructed CSG -- but
they are the exception.
Quite like people who wish to go to the workplace on bike rather than
taking the highway...
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
nemesis wrote:
> Sabrina Kilian wrote:
>> If you wonder, I recommend reading their code and finding out. Then you
>> will see how it does not work past spheres and triangles.
>
> which is why, *again*, I have to point out I was hinting at using the
> GPU solely to speed up povray's ray-triangle intersections.
>
If we must resort to car analogies, you are asking "Why, if a turbo
charger running 1 lbs of boost does so much for an engine, can we not
run 25 lbs of boost and get even more."
Moving 'just the triangle code' would result in MORE branches, not
fewer, as each ray would have to know, before being cast, whether it's
potential target was a simple GPU handled object, or a complex CPU one.
But we have covered this same detail in other posts that you seem
content to just ignore. As for a commandline flag, if what you want is a
simple triangle renderer that is accelerated by the GPU, then I suggest
you use one.
In short, take the words we have offered on these arguments, for the
last 50-odd posts, and google them. Learn something instead of sitting
there and talking about something you have admitted you do not have even
a bare understanding of. Then, when you have completed even a cursory
study of any part of it, come back and we can talk like grown-ups.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
| |
|
|
Sabrina Kilian <ski### [at] vtedu> wrote:
> content to just ignore. As for a commandline flag, if what you want is a
> simple triangle renderer that is accelerated by the GPU, then I suggest
> you use one.
I do. It's not my loss.
Post a reply to this message
|
|
| |
| |
|
|
|
|
| |
|
|
