Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
Gerard Robin wrote: Being Nvidia and X installed , i continu to search a good answer : After many experimentations, I did not notice any change between 24bpp and 32 bpp. There is no difference between 24 and 32 bpp on NVidia hardware. Both of them give you a 32 bit 8:8:8:8 RGBA front and backbuffer, a 32 bit Z depth and (now) an 8 bit stencil buffer, for a grand total of 104 real bits per pixel. You can inspect the list of OpenGL visuals available using the glxinfo command line tool if you like. The real choice underneath the (glut or SDL) abstraction layer is much more complicated than a single number. The reason that this suddenly breaks with the new drivers is that the new drivers have a new feature: they can now support 16 bit color buffers even when the desktop is at 32bpp. But these 16 bit modes do *not* support 8 bit stencil, which is required for the shadow implementation. So it used to by that when FlightGear asked for a 16bpp stencil framebuffer on a 32bpp desktop, it got a 32 bit mode anyway. But now, the driver can actually fulfill the request, so it provides a mode that won't work with shadows. FlightGear asks for a default color depth of 16bpp, but it also asks for stencil; this is essentially a bug. These are not compatible requests on any modern GPUs, which only support 8 bit stencil in true color modes. Andy ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
Andy Ross wrote: Gerard Robin wrote: Being Nvidia and X installed , i continu to search a good answer : After many experimentations, I did not notice any change between 24bpp and 32 bpp. There is no difference between 24 and 32 bpp on NVidia hardware. Both of them give you a 32 bit 8:8:8:8 RGBA front and backbuffer, a 32 bit Z depth and (now) an 8 bit stencil buffer, for a grand total of 104 real bits per pixel. You can inspect the list of OpenGL visuals available using the glxinfo command line tool if you like. The real choice underneath the (glut or SDL) abstraction layer is much more complicated than a single number. The reason that this suddenly breaks with the new drivers is that the new drivers have a new feature: they can now support 16 bit color buffers even when the desktop is at 32bpp. But these 16 bit modes do *not* support 8 bit stencil, which is required for the shadow implementation. So it used to by that when FlightGear asked for a 16bpp stencil framebuffer on a 32bpp desktop, it got a 32 bit mode anyway. But now, the driver can actually fulfill the request, so it provides a mode that won't work with shadows. FlightGear asks for a default color depth of 16bpp, but it also asks for stencil; this is essentially a bug. These are not compatible requests on any modern GPUs, which only support 8 bit stencil in true color modes. Andy And I'll add that we don't ask for a stencil buffer when in 16 bits mode, perhaps we should omit this restriction. Then the driver would have a chance to make the right choice. Harald. ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
Le mardi 02 août 2005 à 09:53 -0700, Andy Ross a écrit : Gerard Robin wrote: Being Nvidia and X installed , i continu to search a good answer : After many experimentations, I did not notice any change between 24bpp and 32 bpp. There is no difference between 24 and 32 bpp on NVidia hardware. Both of them give you a 32 bit 8:8:8:8 RGBA front and backbuffer, a 32 bit Z depth and (now) an 8 bit stencil buffer, for a grand total of 104 real bits per pixel. You can inspect the list of OpenGL visuals available using the glxinfo command line tool if you like. The real choice underneath the (glut or SDL) abstraction layer is much more complicated than a single number. The reason that this suddenly breaks with the new drivers is that the new drivers have a new feature: they can now support 16 bit color buffers even when the desktop is at 32bpp. But these 16 bit modes do *not* support 8 bit stencil, which is required for the shadow implementation. So it used to by that when FlightGear asked for a 16bpp stencil framebuffer on a 32bpp desktop, it got a 32 bit mode anyway. But now, the driver can actually fulfill the request, so it provides a mode that won't work with shadows. FlightGear asks for a default color depth of 16bpp, but it also asks for stencil; this is essentially a bug. These are not compatible requests on any modern GPUs, which only support 8 bit stencil in true color modes. Andy Many Thanks, i begin to understand :=) -- Gerard ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
Le mardi 02 août 2005 à 19:17 +0200, Harald JOHNSEN a écrit : Andy Ross wrote: FlightGear asks for a default color depth of 16bpp, but it also asks for stencil; this is essentially a bug. These are not compatible requests on any modern GPUs, which only support 8 bit stencil in true color modes. Andy And I'll add that we don't ask for a stencil buffer when in 16 bits mode, perhaps we should omit this restriction. Then the driver would have a chance to make the right choice. Harald. Well, i don't know if it is any relationship, but, FG needs absolutely to be runned on 24 depth Xserver, i mean we must start X with 24 depth. If not, the command:fgfs --bpp=24 --Aircraft= ... --Airport= gives the message error RenderTexture Error: Couldn't find a suitable pixel format. Shadow is working but we get an ugly texture scenery, May we concluded: --bpp=24 do not fully operate on FG. Only partly ? -- Gerard ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
Oops, you are quite right. I should have thought about it before I had sent this mail. Matthias On Monday 01 August 2005 23:21, Gerard Robin wrote: Le lundi 01 août 2005 à 23:08 +0200, Matthias Boerner a écrit : Hi, also NVIDIA is not working with 32bpp: You will get following error message if you switch to 32bpp in the section Screen, SubSection Display,...: (II) Setting vga for screen 0. (EE) NVIDIA(0): Given color depth (32) is not supported (EE) NVIDIA(0): *** Aborting *** (II) UnloadModule: nvidia (EE) Screen(s) found, but none have a usable configuration. The man pages of xorg.conf say at DISPLAY SUBSECTION: Depth depth This entry specifies what colour depth the Display subsection is to be used for. This entry is usually specified, but it may be omitted to create a match-all Display subsection or when wishing to match only against the FbBpp parameter. The range of depth values that are allowed depends on the driver. Most driver support 8, 15, 16 and 24. Some also support 1 and/or 4, and some may support other values (like 30). Note: depth means the number of bits in a pixel that are actually used to determine the pixel colour. 32 is not a valid depth value. Most hardware that uses 32 bits per pixel only uses 24 of them to hold the colour information, which means that the colour depth is 24, not 32. Matthias Are you confusing both depth and pixel definition: here an extract from NVIDIA readme DEPTH, BITS PER PIXEL, AND PITCH While not directly a concern when programming modes, the bits used per pixel is an issue when considering the maximum programmable resolution; for this reason, it is worthwhile to address the confusion surrounding the terms depth and bits per pixel. Depth is how many bits of data are stored per pixel. Supported depths are 8, 15, 16, and 24. Most video hardware, however, stores pixel data in sizes of 8, 16, or 32 bits; this is the amount of memory allocated per pixel. When you specify your depth, X selects the bits per pixel (bpp) size in which to store the data. Below is a table of what bpp is used for each possible depth: Depth BPP ------ 8 8 15 16 16 16 24 32 ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
Le lundi 01 août 2005 à 00:18 +0200, Arnt Karlsen a écrit : On Sat, 30 Jul 2005 16:40:58 +0200, Oliver wrote in message [EMAIL PROTECTED]: On Saturday 30 July 2005 16:25, Dave Martin wrote: I don't know if anyone has brought this up yet but the 1.0-7667 driver from NVIDIA for linux breaks the drawn shadows as in they don't appear at all. This tested and confirmed on a FX5800U and 6600GT PCIE Dave Martin No, it works here. You just need to start flightgear in 24 bit mode. fgfs --bpp=24 ...does --bpp=32 work any better than 24bpp for you? (Assuming X run at 32 on Nvidia cards) Being Nvidia and X installed , i continu to search a good answer : After many experimentations, I did not notice any change between 24bpp and 32 bpp. I am not an expert in graphics development, may be the differences depends on the GPU itself and the capability to handle both definitions, The main question could be about CPU: does CPU time used and is it any losses with one or the other ? Does somebody can give an answer ? -- Gerard ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
On Mon, 01 Aug 2005 15:13:54 +0200, Gerard wrote in message [EMAIL PROTECTED]: Le lundi 01 août 2005 à 00:18 +0200, Arnt Karlsen a écrit : On Sat, 30 Jul 2005 16:40:58 +0200, Oliver wrote in message [EMAIL PROTECTED]: On Saturday 30 July 2005 16:25, Dave Martin wrote: I don't know if anyone has brought this up yet but the 1.0-7667 driver from NVIDIA for linux breaks the drawn shadows as in they don't appear at all. This tested and confirmed on a FX5800U and 6600GT PCIE Dave Martin No, it works here. You just need to start flightgear in 24 bit mode. fgfs --bpp=24 ...does --bpp=32 work any better than 24bpp for you? (Assuming X run at 32 on Nvidia cards) Being Nvidia and X installed , i continu to search a good answer : After many experimentations, I did not notice any change between 24bpp and 32 bpp. ..glxgears, FlightGear etc f/s? I am not an expert in graphics development, may be the differences depends on the GPU itself and the capability to handle both definitions, The main question could be about CPU: does CPU time used and is it any losses with one or the other ? Does somebody can give an answer ? ..pass, what I learned from my own research on gpu's before buying an ATI 9250 clone, is ATI are native 24bpp and 24bpp only, where Nvidia is 1x32bpp or 2x16bpp, suggesting ATI would suck at 16bpp doing less than 3x8bpp and at 32bpp not being able to see or make any use of the top 8 bits. My understanding of Nvidea is their cards should work better at 32bpp and 16bpp than at 24bpp, because 24bpp wastes half a 16bpp engine. -- ..med vennlig hilsen = with Kind Regards from Arnt... ;o) ...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 Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
Le lundi 01 août 2005 à 21:53 +0200, Arnt Karlsen a écrit : Being Nvidia and X installed , i continu to search a good answer : After many experimentations, I did not notice any change between 24bpp and 32 bpp. ...glxgears, FlightGear etc f/s? Ouaf. glxgears isn't a representative benchmark, with it we cannot get a good performance analysis. I have played to demonstrate that my old ati 9200 and my other old nvidia 5200 is better than the NVIDIA 6600GT. assuming we use the Nvidia 7xxx driver (not the 6xxx) FG says 6600GT is x2.5 more (32 bpp or 24 seem the same performance ) Celestia says (depending on the render choice) from x3 to x4 more (probably 32bpp, my Xserver is permanently 32bpp) I am not an expert in graphics development, may be the differences depends on the GPU itself and the capability to handle both definitions, The main question could be about CPU: does CPU time used and is it any losses with one or the other ? Does somebody can give an answer ? ...pass, what I learned from my own research on gpu's before buying an ATI 9250 clone, is ATI are native 24bpp and 24bpp only, where Nvidia is 1x32bpp or 2x16bpp, suggesting ATI would suck at 16bpp doing less than 3x8bpp and at 32bpp not being able to see or make any use of the top 8 bits. My understanding of Nvidea is their cards should work better at 32bpp and 16bpp than at 24bpp, because 24bpp wastes half a 16bpp engine. Ok , i will try to analyse it. -- Gerard ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
Arnt Karlsen wrote: ..pass, what I learned from my own research on gpu's before buying an ATI 9250 clone, is ATI are native 24bpp and 24bpp only, where Nvidia is 1x32bpp or 2x16bpp, suggesting ATI would suck at 16bpp doing less than 3x8bpp and at 32bpp not being able to see or make any use of the top 8 bits. My understanding of Nvidea is their cards should work better at 32bpp and 16bpp than at 24bpp, because 24bpp wastes half a 16bpp engine. From what I understand, 24bpp is the same amount of data as 32bpp. It just signifies that there is a separate alpha channel. Since this is not strictly 'color' the last 8 alpha bits are not counted in the color depth. Still, each pixel takes up 32 bits of memory. ATI cards do 16bpp just the same as all the other cards, 16 bits of color and nothing else. (red and blue get 5bpp, green I think is the one that gets 6bpp) Josh ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
On Mon, 01 Aug 2005 23:08:03 +0200, Gerard wrote in message [EMAIL PROTECTED]: Le lundi 01 août 2005 à 21:53 +0200, Arnt Karlsen a écrit : Being Nvidia and X installed , i continu to search a good answer : After many experimentations, I did not notice any change between 24bpp and 32 bpp. ...glxgears, FlightGear etc f/s? Ouaf. glxgears isn't a representative benchmark, with it we cannot get a good performance analysis. ..I said etc. ;o) I have played to demonstrate that my old ati 9200 and my other old nvidia 5200 is better than the NVIDIA 6600GT. assuming we use the Nvidia 7xxx driver (not the 6xxx) FG says 6600GT is x2.5 more (32 bpp or 24 seem the same performance ) Celestia says (depending on the render choice) from x3 to x4 more (probably 32bpp, my Xserver is permanently 32bpp) ..benchmark start-up commandline ideas will help benchmark apples and oranges, as such, rather than as bananas and pineapples. ;o) I am not an expert in graphics development, may be the differences depends on the GPU itself and the capability to handle both definitions, The main question could be about CPU: does CPU time used and is it any losses with one or the other ? Does somebody can give an answer ? ...pass, what I learned from my own research on gpu's before buying an ATI 9250 clone, is ATI are native 24bpp and 24bpp only, where Nvidia is 1x32bpp or 2x16bpp, suggesting ATI would suck at 16bpp doing less than 3x8bpp and at 32bpp not being able to see or make any use of the top 8 bits. My understanding of Nvidea is their cards should work better at 32bpp and 16bpp than at 24bpp, because 24bpp wastes half a 16bpp engine. Ok , i will try to analyse it. -- ..med vennlig hilsen = with Kind Regards from Arnt... ;o) ...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 Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
On Tuesday 02 Aug 2005 00:01, Arnt Karlsen wrote: On Mon, 01 Aug 2005 18:14:16 -0400, Josh wrote in message [EMAIL PROTECTED]: Arnt Karlsen wrote: ..pass, what I learned from my own research on gpu's before buying an ATI 9250 clone, is ATI are native 24bpp and 24bpp only, where Nvidia is 1x32bpp or 2x16bpp, suggesting ATI would suck at 16bpp doing less than 3x8bpp and at 32bpp not being able to see or make any use of the top 8 bits. My understanding of Nvidea is their cards should work better at 32bpp and 16bpp than at 24bpp, because 24bpp wastes half a 16bpp engine. From what I understand, 24bpp is the same amount of data as 32bpp. It just signifies that there is a separate alpha channel. Since this is not strictly 'color' the last 8 alpha bits are not counted in the color depth. ..yes, but does this impact 32bpp performance relative to 24bpp and not 24bpp relative to 16bpp like it should on ATI's and should not on Nvidea and vice versa? Still, each pixel takes up 32 bits of memory. ..my understanding is ATI cannot do 32bpp math at all, their gpus are 24bpp only, while Nvidea gpus does both 16bpp and 32bpp but not 24bpp. Strategic gpu HW design choises made a decade or so back. ATI cards do 16bpp just the same as all the other cards, 16 bits of color and nothing else. (red and blue get 5bpp, green I think is the one that gets 6bpp) ..true, at the same speed as they will do 24bpp, 15bpp and possibly also 8bpp, I doubt ATI gpu's has a 3x8bpp mode, Nvidea however talks about a 2x16bpp and an 1x32bpp mode. As Josh said, in a 32 bpp mode 8 bits are used for an alpha channel so there isn't really any 32 bpp maths to worry about. It probably makes more sense to think in terms of 3x8 bpp for 24 bit modes or 4x8 bpp for 32 bit display modes, with each 8 bit channel giving 256 levels of intensity (brightness). 8 bit and lower colour modes work differently and use an indexed palette and a look-up table of absolute rgb values. The actual colours in the palette can have any value but you're limited to a total of 256 of 'em. An 8 bit greyscale mode is essentially the same as one of the 24 bit colour modes channels except you're only dealing with absolute brightness - all colour info is ignored. 24 bpp data can be displayed on a 15 or 16 bit mode simply by discarding the least significant bits of each channel. This can produce some colour banding and other undesired artifacts but it's economical as the data requires no conversion. After 24 bpp the next commonly used colour mode is 36 bpp - 3x12 bpp. This is mainly used for print stuff. LeeE ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
On Sat, 30 Jul 2005 16:40:58 +0200, Oliver wrote in message [EMAIL PROTECTED]: On Saturday 30 July 2005 16:25, Dave Martin wrote: I don't know if anyone has brought this up yet but the 1.0-7667 driver from NVIDIA for linux breaks the drawn shadows as in they don't appear at all. This tested and confirmed on a FX5800U and 6600GT PCIE Dave Martin No, it works here. You just need to start flightgear in 24 bit mode. fgfs --bpp=24 ..does --bpp=32 work any better than 24bpp for you? (Assuming X run at 32 on Nvidia cards) -- ..med vennlig hilsen = with Kind Regards from Arnt... ;o) ...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 Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
[Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
I don't know if anyone has brought this up yet but the 1.0-7667 driver from NVIDIA for linux breaks the drawn shadows as in they don't appear at all. This tested and confirmed on a FX5800U and 6600GT PCIE Dave Martin ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
On Saturday 30 July 2005 16:25, Dave Martin wrote: I don't know if anyone has brought this up yet but the 1.0-7667 driver from NVIDIA for linux breaks the drawn shadows as in they don't appear at all. This tested and confirmed on a FX5800U and 6600GT PCIE Dave Martin No, it works here. You just need to start flightgear in 24 bit mode. fgfs --bpp=24 Best Regards, Oliver C. ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
Re: [Flightgear-devel] NVIDIA 1.0-7667 breaks shadows entirely.
On Saturday 30 July 2005 15:40, Oliver C. wrote: No, it works here. You just need to start flightgear in 24 bit mode. fgfs --bpp=24 Best Regards, Oliver C. Thanks for that :) Dave Martin. ___ Flightgear-devel mailing list Flightgear-devel@flightgear.org http://mail.flightgear.org/mailman/listinfo/flightgear-devel 2f585eeea02e2c79d7b1d8c4963bae2d
