Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-12 Thread Scott Hamilton
On Wed, 2009-10-07 at 22:03 +1100, Scott Hamilton wrote:


   On a final note, I did a 'make distclean' and ./configure for both
SimGear and FlightGear and now
   the change in frame rate is acceptable, it goes from 30 fps to around
22 - 24 fps once 3D clouds are turned on.


   S.




> On Wed, 2009-10-07 at 09:18 +, Stuart Buchanan wrote: 
> 
> > Scott Hamilton wrote:
> > >On Tue, 2009-10-06 at 22:06 -0400, William Harrison wrote:
> > >
> > >>>Maybe it's just me, but has anyone noticed a dramatic performance 
> > >>> decrease with 3d clouds after this patch?
> > >>
> > >>  yep, from 30fps to 2fps...
> 
> 
> 
> OK, I hope this reproducer scenario helps;
> 
> 1. don't turn on 3D clouds from command line, don't specify any
> weather (manual metar), default; scattered at 4000ft and cirrus at
> 19500
> 25-31fps
> 
> 2. go into rendering options, turn on 3D clouds
> 2-3fps
> 
> 3. change 19500 from cirrus to clear
> 2-6fps
> 
> 4. change 4000 from scattered to few
> remains 2-6 fps
> 
> 5. change 4000 from few to clear
> 20-24fps
> 
> 6. turn OFF 3D clouds (so there is no 3D or 2D clouds)
> 23-26fps
> 
> I last built fgfs  2009-10-03 19:23 (Sydney time) 
> I've now done a cvs update and 'make clean' on SimGear and FlightGear
> and updated data/Shaders/
> 
> 1. same as a above. The cmd line is; bin/fgfs --enable-sound
> --enable-hud --aircraft=A380 --airport=YSSY --runway=34L
> --timeofday=afternoon
> 28-32fps
> 
> 2. go into rendering options, turn ON 3D clouds
> 7 fps
> 
> 3. change 19500 from cirrus to clear
> remains 7fps
> 
> 4. change 4000 from scattered to few
> 10-11fps
> 
> 5. change 4000 from few to clear
> 23-27fps
> 
> 6. turn OFF 3D clouds
> remains 23-27fps
> 
> 
> Operating System: openSuSE 11.1 (kernel2.6.27.29-0.1-default )
> Model: "nVidia Quadro FX 580"
> 
> I'm assuming shaders need some OpenGL help;
> 
> # glxinfo 
> name of display: :0.0
> display: :0  screen: 0
> direct rendering: Yes 
> server glx vendor string: NVIDIA Corporation
> server glx version string: 1.4  
> server glx extensions:  
> GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_SGIX_fbconfig, 
> GLX_SGIX_pbuffer, GLX_SGI_video_sync, GLX_SGI_swap_control,
> GLX_EXT_texture_from_pixmap, GLX_ARB_create_context,
> GLX_ARB_multisample, 
> GLX_NV_float_buffer, GLX_ARB_fbconfig_float,
> GLX_NV_swap_group,   
> GLX_EXT_framebuffer_sRGB,
> GLX_NV_multisample_coverage 
> client glx vendor string: NVIDIA
> Corporation  
> client glx version string:
> 1.4
> client glx
> extensions:
> GLX_ARB_get_proc_address, GLX_ARB_multisample,
> GLX_EXT_visual_info,   
> GLX_EXT_visual_rating, GLX_EXT_import_context,
> GLX_SGI_video_sync,
> GLX_NV_swap_group, GLX_NV_video_out, GLX_SGIX_fbconfig,
> GLX_SGIX_pbuffer, 
> GLX_SGI_swap_control, GLX_ARB_create_context,
> GLX_NV_float_buffer,
> GLX_ARB_fbconfig_float,
> GLX_EXT_fbconfig_packed_float,
> GLX_EXT_texture_from_pixmap,
> GLX_EXT_framebuffer_sRGB,
> GLX_NV_present_video,
> GLX_NV_multisample_coverage 
> GLX
> extensions:   
> GLX_EXT_visual_info, GLX_EXT_visual_rating,
> GLX_SGIX_fbconfig,
> GLX_SGIX_pbuffer, GLX_SGI_video_sync,
> GLX_SGI_swap_control,   
> GLX_EXT_texture_from_pixmap, GLX_ARB_create_context,
> GLX_ARB_multisample, 
> GLX_NV_float_buffer, GLX_ARB_fbconfig_float,
> GLX_NV_swap_group,   
> GLX_EXT_framebuffer_sRGB,
> GLX_NV_multisample_coverage,
> 
> GLX_ARB_get_proc_address  
> OpenGL vendor string: NVIDIA
> Corporation  
> OpenGL renderer string: Quadro FX
> 580/PCI/SSE2
> OpenGL version string: 3.0.0 NVIDIA 180.44 
> 
> 
> 
> I hope something here helps, as they do look good, but I can't get off
> the ground to see them :)
> 
> 
> cheers
>   S.
> 
> 
> 
> 
> 
> > 
> > I'm very surprised that performance has decreased so significantly. 
> > However, there is a possible explanation.
> > 
> > The code falls back to a 2D cloud layer if a 3D cloud isn't defined for a 
> > specific cloud type (st, ac etc.). IIRC
> > previously we didn't define a 3D stratus layer, so we'd get a 2D layer 
> > instead.
> > 
> > The updated 3D clouds include stratus layers (which look rather nice IMO), 
> > but require quite a larger number
> > of sprites - something like 40 per cloud. I've seen a small perf hit from 
> > the new stratus layers, but nothing like
> > what you have reported.
> > 
> > I suspect what you are seeing if the difference between a 2D and 3D cloud.
> > 
> > What performance change are you seei

Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-12 Thread Alex Buzin
Stuart Buchanan wrote:
> Scott Hamilton wrote:
>   
>> On Tue, 2009-10-06 at 22:06 -0400, William Harrison wrote:
>>
>> 
Maybe it's just me, but has anyone noticed a dramatic performance 
 decrease with 3d clouds after this patch?
 
>>>  yep, from 30fps to 2fps...
>>>   
>
> I'm very surprised that performance has decreased so significantly. However, 
> there is a possible explanation.
>
> The code falls back to a 2D cloud layer if a 3D cloud isn't defined for a 
> specific cloud type (st, ac etc.). IIRC
> previously we didn't define a 3D stratus layer, so we'd get a 2D layer 
> instead.
>
> The updated 3D clouds include stratus layers (which look rather nice IMO), 
> but require quite a larger number
> of sprites - something like 40 per cloud. I've seen a small perf hit from the 
> new stratus layers, but nothing like
> what you have reported.
>
> I suspect what you are seeing if the difference between a 2D and 3D cloud.
>
> What performance change are you seeing with, say, cumulus clouds, which 
> haven't changed much?
>
> BTW - I've updated Docs/Readme.3Dclouds with information on the new format, 
> if anyone is interested in enhancing
> the clouds.
>
> -Stuart
>   

What I have seen (airplane-ufo, altitude - 2000ft, cloud layer 0 = 
4000ft, overcast, cloud layer 1 = 19500ft, cirrus):
1. 3D clouds OFF from command line: FPS=70, memory usage 360M, layer 0 - 
visible.

2. 3D clouds ON from  command line: memory usage from 800M up to 1.2G
- pitch =0 deg : FPS=10-12
- pitch =90 deg : FPS=70
- pitch =-90 deg : FPS=70

3. 3D clouds ON from  command line, than switch OFF from menu : FPS = 
60-70, memory usage 700-800M

4. 3D clouds ON from  command line,  start at lat,lon = 0, 0 (no scenery 
loaded) : NO clouds in layer 0, both 3D and 2D. If switch 3D clouds OFF 
from menu then 2D clouds appears.

With respect,
Alex



--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-08 Thread James A. Treacy
On Wed, Oct 07, 2009 at 09:17:55PM +, Stuart Buchanan wrote:
> Unfortunately the 3D cloud performance seems to be very
> graphics-card specific. My card is a couple of years old, but
> doesn't seem to have the same problems.

I am curious as to what your graphics card is. It would be nice if we
could get an idea of which cards work well - and those that that have
problems so they could be addressed.

-- 
James Treacy
tre...@debian.org

--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-08 Thread James Turner

On 7 Oct 2009, at 22:21, Stuart Buchanan wrote:

> Unfortunately, last time I tried it, the OSG Impostor implimentation  
> was very
> broken, and I didn't have the skills to fix it.

As a general point, if impostors worked well in OSG, there's many  
other things we could be using them for - up to, and including,  
complete scenery tiles. This would enable the view distance to be  
pushed up quite high, without impacting framerate the way it currently  
does.

I agree, though, that fixing it 'properly' requires some OGS skills -  
but then I'm surprised more people in the OSG community aren't also  
needed decent impostors.

James

--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-07 Thread Stuart Buchanan
Heiko Schulz wrote:aeitsch...@yahoo.de>
> I coulden't try the new cloud patch yet- I'm waiting that someone makes a 
> binary 
> snapshot for win32-users ;-)
> 
> So much I understood, there is a certain level of sprites numbers needed, so 
> that the single clouds is correctly shaded.
> But as transpareny is one of the strongest factor which slows down, it is no 
> wonder that the fps might decreases...
> 
> As our technic is very similar to the one of X-Plane and MSFS, it is quite 
> interesting how they got managed to have a bit better perfomance:

I think a big difference is that MSFS uses Impostors. These effectively 
pre-render a set of sprites onto a texture, and use the resulting texture until 
the view angle changes significantly. This saves a lot of transparency 
rendering.

Unfortunately, last time I tried it, the OSG Impostor implimentation was very
broken, and I didn't have the skills to fix it.

I may have another look to see if it has been improved in recent OSG releases.

-Stuart



  

--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-07 Thread Stuart Buchanan
Durk Talsma wrote:

> On Wednesday 07 October 2009 11:18:07 am Stuart Buchanan wrote:
> >
> > I'm very surprised that performance has decreased so significantly.
> > However, there is a possible explanation.
> >
> 
> FWIW, I've run into one situation where the new cloud code significantly 
> slowed down my system. I found that reducing the range setting from 20km to 
> approximately 11-12 kilometers gave me back decent performance, without too 
> mach sacrificing of detail. It seemed to be a rather sharp transition, so I'm 
> wondering whether texture memory use has something to do with it. 
> 
> Just thinking out loud, Would it be feasable to impement some sort of LOD 
> scheme, where sprites are dynamically added as they are getting closer? That 
> way, it might be possible to have clouds across a much larger area with as 
> much of an impact on performance.

That's a very interesting idea - I'll add it to my list of things to 
investigate.

Unfortunately the 3D cloud performance seems to be very graphics-card specific. 
My
card is a couple of years old, but doesn't seem to have the same problems.

-Stuart



  

--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-07 Thread Heiko Schulz
Hi,

 
> I also noticed a huge frame rate drop when I was in or
> close to being in the cloud layer, so that the clouds
> consumed the greatest percentage of the screen real estate
> ... that is when the frame rates were the worst.  If we are
> drawing zillions of transparent quads and suddenly they are
> sized to cover nearly the whole screen, then we are
> basically rendering the whole screen (or most of the screen)
> for each cloud quad, or enough of them that are near by to
> pull the frame rates down to very low levels.
> 
> 
> Curt.

I coulden't try the new cloud patch yet- I'm waiting that someone makes a 
binary snapshot for win32-users ;-)

So much I understood, there is a certain level of sprites numbers needed, so 
that the single clouds is correctly shaded.
But as transpareny is one of the strongest factor which slows down, it is no 
wonder that the fps might decreases...

As our technic is very similar to the one of X-Plane and MSFS, it is quite 
interesting how they got managed to have a bit better perfomance:

X-Plane: they use one sprite for one single cloud- that looks sometimes funny 
viewed from above (rotation of the clouds), but seems to give a good 
perfomance. But they don't give this realistic shading from white to grey as we 
can see on the real clouds in the sky or here in FGFS...

MSFS: depending if default or Add-on. The recent Add-On called REX looks really 
good, but also seems to use only sprite per single cloud.
The default clouds tried in the demo once a time uses some more sprites.
Their clouds system shows a quite nice shading similar to our. 

I prefer our shading from white to grey of the clouds, as it looks really 
realistic,  but as I understood Stuart it only works good and without any 
errors with a certain number of sprites. 

I wonder if it would be possible to have this shading from white to grey from 
top to  bottom without many sprites. This would help a lot, and we already can 
define the number of sprites uisng the slider  in the rendering menu.

Correct me if I'm wrong in any points here

Regards
HHS




  

--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-07 Thread Curtis Olson
On Wed, Oct 7, 2009 at 10:33 AM, Durk Talsma  wrote:

> On Wednesday 07 October 2009 11:18:07 am Stuart Buchanan wrote:
> >
> > I'm very surprised that performance has decreased so significantly.
> > However, there is a possible explanation.
> >
>
> FWIW, I've run into one situation where the new cloud code significantly
> slowed down my system. I found that reducing the range setting from 20km to
> approximately 11-12 kilometers gave me back decent performance, without too
> mach sacrificing of detail. It seemed to be a rather sharp transition, so
> I'm
> wondering whether texture memory use has something to do with it.
>
> Just thinking out loud, Would it be feasable to impement some sort of LOD
> scheme, where sprites are dynamically added as they are getting closer?
> That
> way, it might be possible to have clouds across a much larger area with as
> much of an impact on performance.
>

I also noticed a huge frame rate drop when I was in or close to being in the
cloud layer, so that the clouds consumed the greatest percentage of the
screen real estate ... that is when the frame rates were the worst.  If we
are drawing zillions of transparent quads and suddenly they are sized to
cover nearly the whole screen, then we are basically rendering the whole
screen (or most of the screen) for each cloud quad, or enough of them that
are near by to pull the frame rates down to very low levels.

Curt.
-- 
Curtis Olson: http://baron.flightgear.org/~curt/
--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-07 Thread Durk Talsma
On Wednesday 07 October 2009 11:18:07 am Stuart Buchanan wrote:
>
> I'm very surprised that performance has decreased so significantly.
> However, there is a possible explanation.
>

FWIW, I've run into one situation where the new cloud code significantly 
slowed down my system. I found that reducing the range setting from 20km to 
approximately 11-12 kilometers gave me back decent performance, without too 
mach sacrificing of detail. It seemed to be a rather sharp transition, so I'm 
wondering whether texture memory use has something to do with it. 

Just thinking out loud, Would it be feasable to impement some sort of LOD 
scheme, where sprites are dynamically added as they are getting closer? That 
way, it might be possible to have clouds across a much larger area with as 
much of an impact on performance.

Cheers,
Durk

--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-07 Thread Curtis Olson
Hi Scott,

Something else to try would be to turn off the other shaders (water
reflection/ground cover).  When I get all of these + clouds + rain/snow, my
performance really goes in the toilet too.  The snow/elevation shader seems
to be especially hard on frame rates, but I notice the farm field shader has
a big affect too if there is a lot of farm fields in the view.

I don't quite get the farm field shader purpose.  Can we not do pretty much
the same thing with a basic texture?

I see what the terrain elevation/slope shader is doing and that's
interesting, but it's a huge frame rate hit for what it offers.

Regards,

Curt.


On Wed, Oct 7, 2009 at 6:03 AM, Scott Hamilton  wrote:

>  On Wed, 2009-10-07 at 09:18 +, Stuart Buchanan wrote:
>
> Scott Hamilton wrote:
> >On Tue, 2009-10-06 at 22:06 -0400, William Harrison wrote:
> >
> >>>Maybe it's just me, but has anyone noticed a dramatic performance 
> >>> decrease with 3d clouds after this patch?
> >>
> >>  yep, from 30fps to 2fps...
>
>
>
> OK, I hope this reproducer scenario helps;
>
> 1. don't turn on 3D clouds from command line, don't specify any weather
> (manual metar), default; scattered at 4000ft and cirrus at 19500
> 25-31fps
>
> 2. go into rendering options, turn on 3D clouds
> 2-3fps
>
> 3. change 19500 from cirrus to clear
> 2-6fps
>
> 4. change 4000 from scattered to few
> remains 2-6 fps
>
> 5. change 4000 from few to clear
> 20-24fps
>
> 6. turn OFF 3D clouds (so there is no 3D or 2D clouds)
> 23-26fps
>
> I last built fgfs  2009-10-03 19:23 (Sydney time)
> I've now done a cvs update and 'make clean' on SimGear and FlightGear and
> updated data/Shaders/
>
> 1. same as a above. The cmd line is; bin/fgfs --enable-sound --enable-hud
> --aircraft=A380 --airport=YSSY --runway=34L --timeofday=afternoon
> 28-32fps
>
> 2. go into rendering options, turn ON 3D clouds
> 7 fps
>
> 3. change 19500 from cirrus to clear
> remains 7fps
>
> 4. change 4000 from scattered to few
> 10-11fps
>
> 5. change 4000 from few to clear
> 23-27fps
>
> 6. turn OFF 3D clouds
> remains 23-27fps
>
>
> Operating System: openSuSE 11.1 (kernel2.6.27.29-0.1-default )
> Model: "nVidia Quadro FX 580"
>
> I'm assuming shaders need some OpenGL help;
>
> # glxinfo
> name of display: :0.0
> display: :0  screen: 0
> direct rendering: Yes
> server glx vendor string: NVIDIA Corporation
> server glx version string: 1.4
> server glx extensions:
> GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_SGIX_fbconfig,
> GLX_SGIX_pbuffer, GLX_SGI_video_sync, GLX_SGI_swap_control,
> GLX_EXT_texture_from_pixmap, GLX_ARB_create_context,
> GLX_ARB_multisample,
> GLX_NV_float_buffer, GLX_ARB_fbconfig_float,
> GLX_NV_swap_group,
> GLX_EXT_framebuffer_sRGB,
> GLX_NV_multisample_coverage
> client glx vendor string: NVIDIA
> Corporation
> client glx version string:
> 1.4
> client glx
> extensions:
> GLX_ARB_get_proc_address, GLX_ARB_multisample,
> GLX_EXT_visual_info,
> GLX_EXT_visual_rating, GLX_EXT_import_context,
> GLX_SGI_video_sync,
> GLX_NV_swap_group, GLX_NV_video_out, GLX_SGIX_fbconfig,
> GLX_SGIX_pbuffer,
> GLX_SGI_swap_control, GLX_ARB_create_context,
> GLX_NV_float_buffer,
> GLX_ARB_fbconfig_float,
> GLX_EXT_fbconfig_packed_float,
> GLX_EXT_texture_from_pixmap,
> GLX_EXT_framebuffer_sRGB,
> GLX_NV_present_video,
> GLX_NV_multisample_coverage
> GLX
> extensions:
> GLX_EXT_visual_info, GLX_EXT_visual_rating,
> GLX_SGIX_fbconfig,
> GLX_SGIX_pbuffer, GLX_SGI_video_sync,
> GLX_SGI_swap_control,
> GLX_EXT_texture_from_pixmap, GLX_ARB_create_context,
> GLX_ARB_multisample,
> GLX_NV_float_buffer, GLX_ARB_fbconfig_float,
> GLX_NV_swap_group,
> GLX_EXT_framebuffer_sRGB,
> GLX_NV_multisample_coverage,
>
> GLX_ARB_get_proc_address
> OpenGL vendor string: NVIDIA
> Corporation
> OpenGL renderer string: Quadro FX
> 580/PCI/SSE2
> OpenGL version string: 3.0.0 NVIDIA 180.44
>
>
>
> I hope something here helps, as they do look good, but I can't get off the
> ground to see them :)
>
>
> cheers
>   S.
>
>
>
>
>  I'm very surprised that performance has decreased so significantly. However, 
> there is a possible explanation.
>
> The code falls back to a 2D cloud layer if a 3D cloud isn't defined for a 
> specific cloud type (st, ac etc.). IIRC
> previously we didn't define a 3D stratus layer, so we'd get a 2D layer 
> instead.
>
> The updated 3D clouds include stratus layers (which look rather nice IMO), 
> but require quite a larger number
> of sprites - something like 40 per cloud. I've seen a small perf hit from the 
> new stratus layers, but nothing like
> what you have reported.
>
> I suspect what you are seeing if the difference between a 2D and 3D cloud.
>
> What performance change are you seeing with, say, cumulus clouds, which 
> haven't changed much?
>
> BTW - I've updated Docs/Readme.3Dclouds with information on the new format, 
> if anyone is interested in enhancing
> the clouds.
>
> -Stuart
>
>
>
>
> 

Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-07 Thread Scott Hamilton
On Wed, 2009-10-07 at 09:18 +, Stuart Buchanan wrote:

> Scott Hamilton wrote:
> >On Tue, 2009-10-06 at 22:06 -0400, William Harrison wrote:
> >
> >>>Maybe it's just me, but has anyone noticed a dramatic performance 
> >>> decrease with 3d clouds after this patch?
> >>
> >>  yep, from 30fps to 2fps...



OK, I hope this reproducer scenario helps;

1. don't turn on 3D clouds from command line, don't specify any weather
(manual metar), default; scattered at 4000ft and cirrus at 19500
25-31fps

2. go into rendering options, turn on 3D clouds
2-3fps

3. change 19500 from cirrus to clear
2-6fps

4. change 4000 from scattered to few
remains 2-6 fps

5. change 4000 from few to clear
20-24fps

6. turn OFF 3D clouds (so there is no 3D or 2D clouds)
23-26fps

I last built fgfs  2009-10-03 19:23 (Sydney time) 
I've now done a cvs update and 'make clean' on SimGear and FlightGear
and updated data/Shaders/

1. same as a above. The cmd line is; bin/fgfs --enable-sound
--enable-hud --aircraft=A380 --airport=YSSY --runway=34L
--timeofday=afternoon
28-32fps

2. go into rendering options, turn ON 3D clouds
7 fps

3. change 19500 from cirrus to clear
remains 7fps

4. change 4000 from scattered to few
10-11fps

5. change 4000 from few to clear
23-27fps

6. turn OFF 3D clouds
remains 23-27fps


Operating System: openSuSE 11.1 (kernel2.6.27.29-0.1-default )
Model: "nVidia Quadro FX 580"

I'm assuming shaders need some OpenGL help;

# glxinfo 
name of display: :0.0
display: :0  screen: 0
direct rendering: Yes 
server glx vendor string: NVIDIA Corporation
server glx version string: 1.4  
server glx extensions:  
GLX_EXT_visual_info, GLX_EXT_visual_rating, GLX_SGIX_fbconfig, 
GLX_SGIX_pbuffer, GLX_SGI_video_sync, GLX_SGI_swap_control,
GLX_EXT_texture_from_pixmap, GLX_ARB_create_context,
GLX_ARB_multisample, 
GLX_NV_float_buffer, GLX_ARB_fbconfig_float,
GLX_NV_swap_group,   
GLX_EXT_framebuffer_sRGB,
GLX_NV_multisample_coverage 
client glx vendor string: NVIDIA
Corporation  
client glx version string:
1.4
client glx
extensions:
GLX_ARB_get_proc_address, GLX_ARB_multisample,
GLX_EXT_visual_info,   
GLX_EXT_visual_rating, GLX_EXT_import_context,
GLX_SGI_video_sync,
GLX_NV_swap_group, GLX_NV_video_out, GLX_SGIX_fbconfig,
GLX_SGIX_pbuffer, 
GLX_SGI_swap_control, GLX_ARB_create_context,
GLX_NV_float_buffer,
GLX_ARB_fbconfig_float,
GLX_EXT_fbconfig_packed_float,
GLX_EXT_texture_from_pixmap,
GLX_EXT_framebuffer_sRGB,
GLX_NV_present_video,
GLX_NV_multisample_coverage 
GLX
extensions:   
GLX_EXT_visual_info, GLX_EXT_visual_rating,
GLX_SGIX_fbconfig,
GLX_SGIX_pbuffer, GLX_SGI_video_sync,
GLX_SGI_swap_control,   
GLX_EXT_texture_from_pixmap, GLX_ARB_create_context,
GLX_ARB_multisample, 
GLX_NV_float_buffer, GLX_ARB_fbconfig_float,
GLX_NV_swap_group,   
GLX_EXT_framebuffer_sRGB,
GLX_NV_multisample_coverage,

GLX_ARB_get_proc_address  
OpenGL vendor string: NVIDIA
Corporation  
OpenGL renderer string: Quadro FX
580/PCI/SSE2
OpenGL version string: 3.0.0 NVIDIA 180.44 



I hope something here helps, as they do look good, but I can't get off
the ground to see them :)


cheers
  S.





> 
> I'm very surprised that performance has decreased so significantly. However, 
> there is a possible explanation.
> 
> The code falls back to a 2D cloud layer if a 3D cloud isn't defined for a 
> specific cloud type (st, ac etc.). IIRC
> previously we didn't define a 3D stratus layer, so we'd get a 2D layer 
> instead.
> 
> The updated 3D clouds include stratus layers (which look rather nice IMO), 
> but require quite a larger number
> of sprites - something like 40 per cloud. I've seen a small perf hit from the 
> new stratus layers, but nothing like
> what you have reported.
> 
> I suspect what you are seeing if the difference between a 2D and 3D cloud.
> 
> What performance change are you seeing with, say, cumulus clouds, which 
> haven't changed much?
> 
> BTW - I've updated Docs/Readme.3Dclouds with information on the new format, 
> if anyone is interested in enhancing
> the clouds.
> 
> -Stuart
> 
> 
>   
> 
> --
> Come build with us! The BlackBerry(R) Developer Conference in SF, CA
> is the only developer event you need to attend this year. Jumpstart your
> developing skills, take BlackBerry mobile applications to market and stay 
> ahead of the curve. Join us from November 9 - 12, 2009.

Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-07 Thread Stuart Buchanan
Scott Hamilton wrote:
>On Tue, 2009-10-06 at 22:06 -0400, William Harrison wrote:
>
>>>Maybe it's just me, but has anyone noticed a dramatic performance 
>>> decrease with 3d clouds after this patch?
>>
>>  yep, from 30fps to 2fps...

I'm very surprised that performance has decreased so significantly. However, 
there is a possible explanation.

The code falls back to a 2D cloud layer if a 3D cloud isn't defined for a 
specific cloud type (st, ac etc.). IIRC
previously we didn't define a 3D stratus layer, so we'd get a 2D layer instead.

The updated 3D clouds include stratus layers (which look rather nice IMO), but 
require quite a larger number
of sprites - something like 40 per cloud. I've seen a small perf hit from the 
new stratus layers, but nothing like
what you have reported.

I suspect what you are seeing if the difference between a 2D and 3D cloud.

What performance change are you seeing with, say, cumulus clouds, which haven't 
changed much?

BTW - I've updated Docs/Readme.3Dclouds with information on the new format, if 
anyone is interested in enhancing
the clouds.

-Stuart


  

--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-07 Thread Victhor Foster
Same here too. And I don't even have a good graphics card! ;)

--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-06 Thread Vadym Kukhtin
Nope, same fps...

But clouds look much better; thanks Stuart!


-- 
---
WBR, Vadym.

--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-06 Thread Scott Hamilton
On Tue, 2009-10-06 at 22:06 -0400, William Harrison wrote:

> Maybe it's just me, but has anyone noticed a dramatic performance
> decrease with 3d clouds after this patch?


  yep, from 30fps to 2fps...


  S.




> 
> 
> On Fri, Oct 2, 2009 at 1:51 AM, Durk Talsma 
> wrote:
> 
> On Thursday 01 October 2009 06:22:15 am Durk Talsma wrote:
> > Hi Stuart,
> >
> > Unless somebody else beats me to it, I'll try to commit this
> patch this
> > weekend.
> 
> > > Once it is applied, I'll update the appropriate README
> documentation
> > > myself.
> > >
> 
> 
> 
> 
> Okay done. Have fun with the documentation :-)
> 
> Cheers,
> Durk
> 
> 
> --
> 
> 
> Come build with us! The BlackBerry® Developer Conference
> in SF, CA
> is the only developer event you need to attend this year.
> Jumpstart your
> developing skills, take BlackBerry mobile applications to
> market and stay
> ahead of the curve. Join us from November 9-12, 2009.
> Register now!
> http://p.sf.net/sfu/devconf
> ___
> Flightgear-devel mailing list
> Flightgear-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/flightgear-devel
> 
> 
> 
> 
> --
> Come build with us! The BlackBerry(R) Developer Conference in SF, CA
> is the only developer event you need to attend this year. Jumpstart your
> developing skills, take BlackBerry mobile applications to market and stay 
> ahead of the curve. Join us from November 9 - 12, 2009. Register now!
> http://p.sf.net/sfu/devconference
> ___ Flightgear-devel mailing list 
> Flightgear-devel@lists.sourceforge.net 
> https://lists.sourceforge.net/lists/listinfo/flightgear-devel



**
This message is intended for the addressee named and may contain
privileged information or confidential information or both. If you
are not the intended recipient please delete it and notify the sender.
**
--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-06 Thread William Harrison
Maybe it's just me, but has anyone noticed a dramatic performance decrease
with 3d clouds after this patch?

On Fri, Oct 2, 2009 at 1:51 AM, Durk Talsma  wrote:

> On Thursday 01 October 2009 06:22:15 am Durk Talsma wrote:
> > Hi Stuart,
> >
> > Unless somebody else beats me to it, I'll try to commit this patch this
> > weekend.
> > > Once it is applied, I'll update the appropriate README documentation
> > > myself.
> > >
>
>
> Okay done. Have fun with the documentation :-)
>
> Cheers,
> Durk
>
>
> --
> Come build with us! The BlackBerry® Developer Conference in SF, CA
> is the only developer event you need to attend this year. Jumpstart your
> developing skills, take BlackBerry mobile applications to market and stay
> ahead of the curve. Join us from November 9-12, 2009. Register now!
> http://p.sf.net/sfu/devconf
> ___
> Flightgear-devel mailing list
> Flightgear-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/flightgear-devel
>
--
Come build with us! The BlackBerry(R) Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9 - 12, 2009. Register now!
http://p.sf.net/sfu/devconference___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-10-01 Thread Durk Talsma
On Thursday 01 October 2009 06:22:15 am Durk Talsma wrote:
> Hi Stuart,
>
> Unless somebody else beats me to it, I'll try to commit this patch this
> weekend.
> > Once it is applied, I'll update the appropriate README documentation
> > myself.
> >


Okay done. Have fun with the documentation :-)

Cheers,
Durk

--
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


Re: [Flightgear-devel] [PATCH] 3D Clouds update

2009-09-30 Thread Durk Talsma
Hi Stuart,

Unless somebody else beats me to it, I'll try to commit this patch this 
weekend.

Thanks for this fine contribution,

Cheers,
Durk


On Wednesday 30 September 2009 11:15:21 pm Stuart Buchanan wrote:
> Hi All,
>
> I've been doing some work to improve the 3D clouds.
>
> As well as a variety of bug fixes, improvements include:-
> - A new XML format
> - Texture indexing based on the position of the sprite in the cloud-mass,
> allowing more control over the texture set. - Improved fog and shading
> - Better sprite distribution
> - A more natural distribution of clouds, so no more obvious grids.
>
> Most of the improvements can be seen in the Cu and Ac cloud types, but the
> stratus clouds have been improved as well.
>
> A patch is available here:
> http://www.nanjika.co.uk/flightgear/clouds.tar.gz
>
> Once it is applied, I'll update the appropriate README documentation
> myself.
>
> Comments are welcome as always.
>
> -Stuart
>
>
>
>
>
> ---
>--- Come build with us! The BlackBerry® Developer Conference in SF, CA
> is the only developer event you need to attend this year. Jumpstart your
> developing skills, take BlackBerry mobile applications to market and stay
> ahead of the curve. Join us from November 9-12, 2009. Register now!
> http://p.sf.net/sfu/devconf
> ___
> Flightgear-devel mailing list
> Flightgear-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/flightgear-devel


--
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel


[Flightgear-devel] [PATCH] 3D Clouds update

2009-09-30 Thread Stuart Buchanan
Hi All,

I've been doing some work to improve the 3D clouds.

As well as a variety of bug fixes, improvements include:-
- A new XML format
- Texture indexing based on the position of the sprite in the cloud-mass, 
allowing more control over the texture set.
- Improved fog and shading
- Better sprite distribution
- A more natural distribution of clouds, so no more obvious grids.

Most of the improvements can be seen in the Cu and Ac cloud types, but the 
stratus clouds have been improved as well.

A patch is available here: http://www.nanjika.co.uk/flightgear/clouds.tar.gz

Once it is applied, I'll update the appropriate README documentation myself.

Comments are welcome as always.

-Stuart



  

--
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay 
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
___
Flightgear-devel mailing list
Flightgear-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/flightgear-devel