Re: [Flightgear-devel] 777 Model

[Tiago Gusmão]

On Monday 26 July 2004 19:58, Boris Koenig wrote:
> Regarding profiling: what would be necessary to be done ?
> Are there _any_ profiler tools for 3D/openGL applications ?

you might want to take a look at this:
http://www.hawksoft.com/gltrace/

Regards,
Tiago

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
2f585eeea02e2c79d7b1d8c4963bae2d

Re: [Flightgear-devel] 777 Model

[Boris Koenig]

Ampere K. Hardraade wrote:
Be thankful for that 30-36 fps you have.  I usually have about 6-9 fps. ='(
Yes, as I said: I get pretty much the same with the "new" nvidia card,
and regarding the ATI card, I did have to disable several options to
come into the 20+ FPS range, but on the other hand I don't care that
much for the eyecandy stuff, I'd rather have a smooth performance ;-)
But if anybody knows how to profile openGL applications, I really
wouldn't mind to install/configure the necessary stuff if that helps
to track down the most essential problems and make FlightGear become
smoother.
-
Boris
___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
2f585eeea02e2c79d7b1d8c4963bae2d

Re: [Flightgear-devel] 777 Model

[Ampere K. Hardraade]

Be thankful for that 30-36 fps you have.  I usually have about 6-9 fps. ='(

Regards,
Ampere

On July 26, 2004 02:58 pm, Boris Koenig wrote:
> On the other hand, the old ATI R128 card achieves about 70-80 fps
> in 800x600 resolution under _windows_ running stuff like
> counterstrike. While running FlightGear (either under windows or linux)
> leaves me with only about 30-35 fps if I am lucky.

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
2f585eeea02e2c79d7b1d8c4963bae2d

Re: [Flightgear-devel] 777 Model

[Boris Koenig]

Arnt Karlsen wrote:
On Sun, 25 Jul 2004 02:28:46 -0400, Norman wrote in message 
<[EMAIL PROTECTED]>:
I am espescially interested in the profiling results from the newer
higher end cards.  i.e the GForce 4 class or equivalent cards

..which is the low end limit on ATI, 3dfx etc cards, that can do at
least 1fps?
I mentioned this a couple of times already, but the weird thing is
really that I have personally achieved higher/better performance
using an OLD card, rather than using my current nvidia accelerator
on the main machine, I was just recently really about to change
cards ;-)
On the other hand, the old ATI R128 card achieves about 70-80 fps
in 800x600 resolution under _windows_ running stuff like
counterstrike. While running FlightGear (either under windows or linux)
leaves me with only about 30-35 fps if I am lucky.
Regarding profiling: what would be necessary to be done ?
Are there _any_ profiler tools for 3D/openGL applications ?
I know that there do exist some openGL related logging tools
which can monitor the commands that the graphics adapter
receives/processes, but don't have the slightest clue, how
to really PROFILE an openGL app if you want to profile the
openGL instructions.
Don't even know if one could use gprof for such a purpose ?
If there exist any libraries specifically aimed at profiling/debugging
openGL applications, it might indeed make sense to optionally
include such functionality in "developer releases" in order to
first find the bottleneck on most configurations, and then be
able to really address it.
If I remember correctly, SGI did have some kind of openGL debugger,
but don't know if there's any freely available stuff for these
purposes, personally I certainly wouldn't care if I had to install
another 50 meg package in order to have the necessary profiling
capabilities, such a profiling feature could optionally even report
its results automatically back to the FlightGear webpage, that way
one could really get representative data.
-
Boris

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
2f585eeea02e2c79d7b1d8c4963bae2d

Re: [Flightgear-devel] 777 Model

[Arnt Karlsen]

On Sun, 25 Jul 2004 02:28:46 -0400, Norman wrote in message 
<[EMAIL PROTECTED]>:

> Boris Koenig writes:
> > 
> > But, there seems to be a project related to openRT that is dedicated
> > to developing the necessary hardware: http://www.saarcor.de/
> 
> This is a fascinating project but ...  until these chips are as
> prevalent in consumer grade hardware as OpenGL cards are today, I
> think we should content ourselves with just dreaming about programing
> FGFS in OpenRT. 

..OpenGL is mature tech, say IBM hops in full bore, how far away is
useable OpenRT?
 
> Note that FGFS does not utilize many of the features available in the 
> more current generations of OpenGL cards but now that OpenGL 2.0 
> is a reality that may start to change in the not so distant future.

..impact on FG performance with OpenGL 2.0?  
I was thinking X.org 11R6.7.0.  ;-)

> This *might* make a large differance in the rendering performance
> although I suggest that those preoccupied with the rendering speed
> profile the code to see where the time is being spent.
> 
> I am espescially interested in the profiling results from the newer
> higher end cards.  i.e the GForce 4 class or equivalent cards

..which is the low end limit on ATI, 3dfx etc cards, that can do at
least 1fps?

-- 
..med vennlig hilsen = with Kind Regards from Arnt... ;-)
...with a number of polar bear hunters in his ancestry...
  Scenarios always come in sets of three: 
  best case, worst case, and just in case.


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

RE: [Flightgear-devel] 777 Model

[Norman Vine]

Boris Koenig writes:
> 
> But, there seems to be a project related to openRT that is dedicated to
> developing the necessary hardware: http://www.saarcor.de/

This is a fascinating project but ...  until these chips are as prevalent
in consumer grade hardware as OpenGL cards are today, I think we 
should content ourselves with just dreaming about programing FGFS
in OpenRT. 

Note that FGFS does not utilize many of the features available in the 
more current generations of OpenGL cards but now that OpenGL 2.0 
is a reality that may start to change in the not so distant future.

This *might* make a large differance in the rendering performance
although I suggest that those preoccupied with the rendering speed
profile the code to see where the time is being spent.

I am espescially interested in the profiling results from the newer
higher end cards.  i.e the GForce 4 class or equivalent cards

Cheers

Norman

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] 777 Model

[Boris Koenig]

On July 20, 2004 03:23 am, Jim Wilson wrote:
Hmmm... that 777 Model page didn't mention a GPU.  In any case, I gather
from reading just the first paragraph on the OpenRT page you'd be 
looking
at having plib utilize the OpenRT API in lieu of OpenGL's.
I may be wrong, but from what I've read, openRT is indeed supposed to
make use of a specific GPU - or alternatively a (cluster of) CPU(s)
@ ~40 GHZ :-)
But, there seems to be a project related to openRT that is dedicated to
developing the necessary hardware: http://www.saarcor.de/
Seemingly, also a project of the same German university.
Ampere K. Hardraade wrote:
All I am saying is that it will be a good idea to look deeper into it instead 
of pushing it aside.  After all, from what I have read on their site, the 
OpenRT library seems to offer some pretty neat capabilities that aren't in 
the current version of plib.
plib makes use of openGL while openRT is a different "rendering" 
approach which doesn't utilize rasterization anymore - so I think Jim
is right in saying that one would really have to drop the entire openGL
approach to make use of something like that.
I don't know the details, but I doubt that it would be really "simple"
to convert to such a new approach, which wouldn't rely on openGL anymore.

In the long term the openRT folks probably hope to replace the openGL 
implementation in many 3D applications with openRT, because it could 
provide a performance boost in many cases, particularly because it would
no longer be necessary to process all graphics data just in order to
determine which parts are really visible or not ...

At the very least, we should keep this real-time-raytracing technology in 
mind.  The idea of Microsoft come out with games that utilize real time 
raytracing while Linux has nothing equivilent is... freightening.
I agree, the whole idea is extremely fascinating: Having had a look at
some of the screenshots or even videos, the stuff seems really to be
pretty amazing and powerful, but currently it's probably really a bit
beyond the scope of any "game", to care too much for openRT, simply
because of the lack of hardware support, I think - be it a relevant
GPU board or the 40 GHZ requirement for openRT :-)
And then I am not even sure if there's really yet an OPEN
implementation available !?
As long as FlightGear keeps getting modularized even more, it should not
be that much of a problem, to consider new technologies - even though
this unlikely to become an issue within the next 3-5 years ;-)
But on the other hand, at http://graphics.cs.uni-sb.de/RTGames
you can read:
"We are very much interested in evaluating new ways for computer games
and therefore like to cooperate with the gaming industry. Thus if you
are in such a position, please send us an email
!"
So, now it depends if FlightGear is considered part of "the gaming
industry" :-)
But if the openRT developers are really also looking for opensource
projects, it would probably be not that bad for FlightGear to at least
indicate some willingness ;-)

-
Boris
___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] 777 Model

[Ampere K. Hardraade]

All I am saying is that it will be a good idea to look deeper into it instead 
of pushing it aside.  After all, from what I have read on their site, the 
OpenRT library seems to offer some pretty neat capabilities that aren't in 
the current version of plib.

At the very least, we should keep this real-time-raytracing technology in 
mind.  The idea of Microsoft come out with games that utilize real time 
raytracing while Linux has nothing equivilent is... freightening.

Regards,
Ampere


On July 20, 2004 03:23 am, Jim Wilson wrote:
> Hmmm... that 777 Model page didn't mention a GPU.  In any case, I gather
> from reading just the first paragraph on the OpenRT page you'd be looking
> at having plib utilize the OpenRT API in lieu of OpenGL's.
>
> Best,
>
> Jim

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] 777 Model

[Jim Wilson]

"Ampere K. Hardraade" said:

> On July 10, 2004 08:25 pm, Norman Vine wrote:
> > Ampere K. Hardraade writes:
> > > Anyway we can get the plib group to look into their method for rendering?
> >
> > Have at it !
> How do I reach them?
> 
> >
> > Note PLib's scenegraph is SSG < Simple Scene Graph >
> > Since this model is anything but simple IMO it doesn't really
> > qualify for SSG < Simple Scene Graph > :-)
> Well, their method of rendering is capable of rendering that 350 millions 
> triangle monster in under 10 seconds.  If using that method means that the 
> framerates of FlightGear goes up plus more detail scenery, then it certainly 
> worths look into in my opinion.

Hmmm... that 777 Model page didn't mention a GPU.  In any case, I gather from
reading just the first paragraph on the OpenRT page you'd be looking at having
plib utilize the OpenRT API in lieu of OpenGL's.

Best,

Jim


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] 777 Model

[Jim Wilson]

Jon Stockill said:

> Wow!
> 
> http://graphics.cs.uni-sb.de/MassiveRT/boeing777.html
> 
> How long until we're using models with that level of detail then? ;-)
> 

WAG: 8 years on high end retail hardware. ;-)

..fwiw...which isn't much.

Best,

Jim


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] 777 Model

[Martin Spott]

"Ampere K. Hardraade" wrote:

> Well, their method of rendering is capable of rendering that 350 millions 
> triangle monster in under 10 seconds.  If using that method means that the 
> framerates of FlightGear goes up plus more detail scenery, then it certainly 
> worths look into in my opinion.

Please keep in mind that the Boeing apparently doesn't contain textures
but siple shadowing instead,

Martin.
-- 
 Unix _IS_ user friendly - it's just selective about who its friends are !
--

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] 777 Model

[Durk Talsma]

The plib project homepage is at 

http://plib.sf.net

and the main plib developers mailinglist is here:

[EMAIL PROTECTED]

Cheers,
Durk

On Monday 12 July 2004 04:35, Ampere K. Hardraade wrote:
> On July 10, 2004 08:25 pm, Norman Vine wrote:
> > Ampere K. Hardraade writes:
> > > Anyway we can get the plib group to look into their method for
> > > rendering?
> >
> > Have at it !
>
> How do I reach them?
>
> > Note PLib's scenegraph is SSG < Simple Scene Graph >
> > Since this model is anything but simple IMO it doesn't really
> > qualify for SSG < Simple Scene Graph > :-)
>
> Well, their method of rendering is capable of rendering that 350 millions
> triangle monster in under 10 seconds.  If using that method means that the
> framerates of FlightGear goes up plus more detail scenery, then it
> certainly worths look into in my opinion.
>
> Regards,
> Ampere
>
> ___
> Flightgear-devel mailing list
> [EMAIL PROTECTED]
> http://mail.flightgear.org/mailman/listinfo/flightgear-devel


___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] 777 Model

[Ampere K. Hardraade]

On July 10, 2004 08:25 pm, Norman Vine wrote:
> Ampere K. Hardraade writes:
> > Anyway we can get the plib group to look into their method for rendering?
>
> Have at it !
How do I reach them?

>
> Note PLib's scenegraph is SSG < Simple Scene Graph >
> Since this model is anything but simple IMO it doesn't really
> qualify for SSG < Simple Scene Graph > :-)
Well, their method of rendering is capable of rendering that 350 millions 
triangle monster in under 10 seconds.  If using that method means that the 
framerates of FlightGear goes up plus more detail scenery, then it certainly 
worths look into in my opinion.

Regards,
Ampere

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] 777 Model

[Ampere K. Hardraade]

http://graphics.cs.uni-sb.de/RTGames/

Regards,
Ampere

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

RE: [Flightgear-devel] 777 Model

[Norman Vine]

Ampere K. Hardraade writes:
> 
> Anyway we can get the plib group to look into their method for rendering?

Have at it !

Note PLib's scenegraph is SSG < Simple Scene Graph >
Since this model is anything but simple IMO it doesn't really 
qualify for SSG < Simple Scene Graph > :-)

Norman

> 
> On July 10, 2004 06:32 pm, Jon Stockill wrote:
> > Wow!
> >
> > http://graphics.cs.uni-sb.de/MassiveRT/boeing777.html
> >
> > How long until we're using models with that level of detail then? ;-)
> 
> ___
> Flightgear-devel mailing list
> [EMAIL PROTECTED]
> http://mail.flightgear.org/mailman/listinfo/flightgear-devel
> 

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] 777 Model

[Ampere K. Hardraade]

Anyway we can get the plib group to look into their method for rendering?

Regards,
Ampere

On July 10, 2004 06:32 pm, Jon Stockill wrote:
> Wow!
>
> http://graphics.cs.uni-sb.de/MassiveRT/boeing777.html
>
> How long until we're using models with that level of detail then? ;-)

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Re: [Flightgear-devel] 777 Model

[Ampere K. Hardraade]

We can use that... now, but you'll have to set it so that the highest level of 
detail of each component only shows up when the camera is less than 1 meter 
from it.

LOL.

Regards,
Ampere

On July 10, 2004 06:32 pm, Jon Stockill wrote:
> Wow!
>
> http://graphics.cs.uni-sb.de/MassiveRT/boeing777.html
>
> How long until we're using models with that level of detail then? ;-)

___
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel