I kreate a distribution specific rpm-package. But the message is: Any problems during the buildproces are registred.

so i can't install the fglrx-driver

If I install the driver automaticly, opensuse does not work later.

?????????????????

I am working on the implementation of HD audio bitstream through the last radeon 5xxx.

There is no available documentation on this whereas the implementation is different from one constructor to another.

Could you bring some technical details about it ?

For each format (dolby True HD, DTS HD, DD+,AAC, LPCM...) how should be fed the media structures and which audio rendering mode should be used (waveout, directsound, wasapi exclusive, asio ?) ?

Thank you for your help

Regards,

Damien Bain-Thouverez

Does anybody know the exact procedure to enable adaptive deinterlacing on M9?

i'am expericening heavy video tearing. When will there be vsync for xv?

http://www.jotschi.de/wp-content/uploads/2009/05/crappy_video.png

Greetings

Jotschi

i just upgraded to fglrx 9-10 and found out that the GL_TEXTURE_2D_ARRAY_EXT is no longer working as before.

The extension is enabled:

jotschi@NeXuS:~$ glxinfo | grep texture_array
    GL_EXT_texgen_reflection, GL_EXT_texture3D, GL_EXT_texture_array,

I can also access the sampler2DArray uniform variable within my shader but i just get 0 floats returned for any texcoord. (Black Image)

If i use GL_TEXTURE_2D everything works as usual.

void main(){ gl_FragColor = vec4(1.0); } then i get error "Hardware compilation failed" and no statistic. but when i restart GSA then i get assambly code from previous shader. but still none statistic. when i change target card or shader i get error again. so i can get disassembly only when i start GSA with some shader in. compilation lead to error. sometimes i get "unknown error save shader and restart GSA"

when i make syntax error in code i get regular error message about.

i try it on windows XP and windows 7. older version work fine.

operating system to window7 .

i m very courious about your latest demo so can't wait to try .

but when i try ing to run the pingpong on win7,it says can only be ran on vista sp1.what a pity .everyone konws vista isn't that polular.so this way ,

your 10.1 tech no showing to audinces around the world .

so please fix it.     

There is a post somewhere in here about running the NVidia Emulator, but that doesn't seem to allow changing the default driver (always reverts back).  Will this only work in a boot camp configuration?  Not sure where logs are located, so my attempts have been limited to swapping various libES drivers to no avail.

By the way, ES2.0 examples from various sources run fine under this configuration.  Like the Open GL ES 2.0 Programming Guide ones and the PowerVR SDK ones...

Config:

MAC Snow Leopard 10.6.1 - NVidia GeForce 8800GT (sorry...)

RenderMonkey 1.8.2

Parallels 5

Windows 7

Thanks in advance,  Patrice

I am in the process of doing some GPGPU research and need some questions about my GPU answered if possible.

I am running an ATI Mobility RADEON HD 4530/4570 card on a Dell Studio laptop using Catalyst 9.2. I was wondering if anyone has any information on the size of texture cache on this card?  Also, am I correct in assuming that if memory goes outside the defined bounds of on board memory the memory outside that bounds are mapped to main memory via PCIe?

One last question, does anyone know if there have been any successful implementations of CFD software using the GPU? I have found a few items on the iterative conjugate gradient solver but not its actual use inside of application code.  My quess is that many iterative applications like FEM are not good performers on GPU due to data element reuse?

Thank you.

We are trying to run the DXVA HD sample from Media Foundation SDK in Windows 7 running on ATI 4600. It fails  showing DX DEBUG Error - DXVA HD not exported in CAPS.

We downloaded the DXVA Checker exe and the DXVA HD tab is showing nothing. Is there something that we missing or DXVA HD is not supported ?

Thanks and Regards

Vishu

There do appear to be some registry entries DALOvlYUVBrightness etc changing these seems to have no effect but they do apear to be modified by the tool Ati Tray Tools. 

What is the correct method for progamatically setting the overlay colour controls on recent Radeon cards, also can this be exposed via the Display SDK.

Thanks

John

Is this feature also available from the command-line?

My current argument is +fourCC DXT1.

Does anyone perhaps know where to find the BIOS file for the FireGL v8700? It seems to me that there are no such thing as a BIOS for a FireGL card, no BIOS support / updates...? So, let's say you've screwed up your FireGL v8700 card, then it means you've screwed up $1500, but it you've screwed up your Radeon, cheapo GFX card, sure, no problem, just get the BIOS {all-over-the-internet}. Why is it thatthe "Professional" card solutions have less support for these High-Tech stuff, whuile it also costs much more? Please, if anyone konws where I can get hold of a FireGL v8700 BIOS, PLEASE reply to this topic {not that I have any hope}...?

Im getting this error using the debug runtime in the August 2009 DX SDK.

"Direct3D9: (ERROR) :ASSERTION FAILED! File d:\vista_rtm\MultiMedia\DirectX\dxg\d3d9\fw\texture.hpp Line 71: GetBufferDesc()->Format == GetUserFormat()"

Running on an ATI 5870

Its only when I use a depth texture. It passes the CheckDeviceFormat() call and renders correctly, just gives this error. I havent tried on W7 or XP will try tomorrow.

I dont recall getting this error before, but I havent tested on an ATI product in a year or so.

I'm particularly interested in Queriyng a Radeon HD 4890 GPU, specially using OpenCL.

Since I could have several GPUs in my system, what i want is to activate/deactivate a software lock depending on the serial number (or some other unique internal hardware identifier) on the GPUs depending on which device is going to execute a piece of a program.

So, is something like a "GPUID" available to the mentioned GPUs?

Thanks

So is Stream Kernel Analyzer for OpenCL in development?

I am developing an OpenGL application and encountered some problems, caused by Windows and my hardware configuration.

My system consists of 3 RadeonHD 4870 and 6 connected monitors. On every monitor, an OpenGL Window is created and should render some full-hd content. I figured out, that windows send all OpenGL commands to all graphic cards.

To get over this problem, I have tried to use the AMD_gpu_association extension to bind the specific commands to a render context, but I was not able to create a render context like described in the extension specification.

wglCreateAssociatedContextAMD( 1 ) returned 0 and GetLastError does also return 0

Here is an example of my code:

hDC=GetDC(windowHandle);

pixelFormat=ChoosePixelFormat(hDC,&pfd);

SetPixelFormat(hDC,pixelFormat,&pfd);

hRC=wglCreateContext(hDC);

wglMakeCurrent(hDC,hRC));

gpuID = wglGetContextGPUIDAMD( hRC );

wglMakeCurrent( NULL, NULL );

wglDeleteContext( hRC );

hRC = wglCreateAssociatedContextAMD( gpuID );

I have tried this with and without 'wglDeleteContext( hRC )' and 'wglMakeCurrent( NULL, NULL )' with no effect.

Could you please send me some sample code or tell me whats my fault?

Michael

I see that OpenCL drivers will not be developed for 3xxx family, but what about DirectCompute ?

I need OpenCL or DirectCompute support for my notebook, and I don't want to buy a new notebook, I've just bought this one 7 months ago...

Thanks.

--Steve

http://developer.nvidia.com/object/nexus.html

I think it can be useful for ATI Stream developers.

require windows 7 and they won't even install on vista. Why is it so?

I have a 5770 and I am able to run DX11 stuff included with DXSDK in vista and also the new DX11 benchmark 'Heaven' released by Unigine.

First, this is  my first post, hello!

So, I'm trying to get the Temperature of my X1350 with ati2edx.dll, my Sourcecode:

// main.cpp
//

#include <stdio.h>
#include <windows.h>

#define GetTemperatureFunctionName "AtiEDUGetThermalRemoteTemperatureOffset"

typedef UINT (CALLBACK* LPFNDLLTERMO)(UINT, UINT, UINT);

int main(int argc, char** argv)
{
    HINSTANCE hDll;               // Handle to DLL
    LPFNDLLTERMO lpfnDllTermo;    // Function pointer
    UINT uReturnVal;

    printf("Loading Dll..\n");
    hDll = LoadLibrary(TEXT("ati2edxx.dll"));
    if (hDll != NULL)
    {
        printf("Dll loaded!\n");

        printf("Loading Function (%s)...\n", GetTemperatureFunctionName);
        lpfnDllTermo = (LPFNDLLTERMO)GetProcAddress(hDll, GetTemperatureFunctionName);

        if (!lpfnDllTermo)
        {
            printf("Error Loading Function!\n");
            FreeLibrary(hDll);
        }
        else
        {
            printf("Function loaded, calling...\n");
            UINT a = 0;
            UINT b = 0;
            UINT c = 0;
            uReturnVal = lpfnDllTermo(a, b, c);
            printf("Returned Value: %d", uReturnVal);
        }
    }
    else
    {
        printf("Error loading Dll!\n");
    }

    getchar();
}

It always outputs 5, what am i dooing wrong?

markus111

There seems to be a real void in outputing actual NTSC/PAL/HD video via a graphics card for people that want to write straight out to the port so to speak. 

I have left posts here but no replies. So where do you get help in these areas there is little information on?

Like how does field timing work with HDMI or DVI+ output to actual video monitors that output 50i or 60i?

How do you know when a video frame is sent so you can time the fill of another buffer.

What is the field order on the output?

Can you just by time to get answers to questions video output via an ATI video cards?

Thanks

Writting a multiple input mixer for HD output. I need some advice. My biggest issue is being able to synchronously sending out video frames via directX and knowing it was sent. So far trying to do this I have not been able to time a frame output during a frame sync.

With video capture cards there are callbacks which make it simple to know when to output but the I/O latency has been too large. (6 to 10 frames). So I started to look at video cards to act as an output device. The first problem I have run into is timing the output. Seems there is nothing provided in Windows to know when a frame has been sent.

Is this true or I have not been able to find the function I need.

Thanks

If I switch to an NV card in the exact same machine the debug speed issue goes away, so I feel that I've isolated the problem to the ATI driver. I'm running XP x64 on an old HP xw9300 dual Opteron machine.

I know I'm not being very specific on the problem but here are some numbers: stepping over a relatively simple line of code is taking about 6 seconds for each line - normally (i.e with Nvidia hardware) the response is nearly-instantaneous. Also, at application exit, there is a 45-second delay or more, whereas normally the teardown only takes a few seconds.

Incidentally, I'm seeing many messages like this in the VS output window:

fnCallReturnMessage::WM_WINDOWPOSCHANGING: uID = 0

Any help or advice would be appreciated, as my AMD/ATI developer experience is  less than stellar right now.

Thanks,
Jon   

We have an application written in wxWidgets which seems to be behaving badly with HydraVision's preserve app size feature, which unfortunately for us, is turned on by default.

We have a main application window, but the user may open other persistent floating windows later on that show app logs or other tools... HydraVision is resizing those windows to the same size as our main app window, which is very annoying for users! 

Any hints, suggestions, explanations?

Thanks,
Jon

ps. Anyone have a better suggestion for where to post these and other problems (esp OpenGL probs?)

In order to separate DVI settings (i.e VTotal, Htotal, Vsync...) from the crtc2 engine, I tried to use the flat panel crtc shadow registers (the ones with fp_ prefix). I observed that they work fine for the crtc engine. However, I could not find an fp crtc shadow register to change the VTotal value which I was hoping that would be the key for the solution. In xf86 code I found an undocumented  register (FP_CRTC2_V_TOTAL_DISP  0x354). However, writing to this register had no effect when shadow registers are enabled. Therefore, I'm not sure if this register really exists on M9.

Anybody has an idea on how to display tvout and dvi on the second head (display)?

Note that we need to use a specific LCD output (DVI) device that strictly obeys VESA modes.

It would be nice if the source to this tool was released so we could get around problems like these without delay.

Barring that, when were you planning on releasing 64 bit versions of the library/application?  Say, compiled with VS2008.

Thanks,

Adrian

Sucker Punch Productions

http://developer.amd.com/Downloads/APL_SampleConvert.zip

>Page Load - Invalid Download. Please return to the page containing the original download link and try again.(System.NullReferenceException: Object reference not set to an instance of an object.
   at AMD.DevCentral.UI.MembershipPages.LicenseAgreement.Page_Load(Object sender, EventArgs e))

i just developed an opengl application that uses texture arrays.

I need to clamp my textures to the border so i definded the GL_TEXTURE_WRAP T and S to CLAMP_TO_BORDER.

glTexParameterf(GL_TEXTURE_2D_ARRAY_EXT, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_BORDER);
    glTexParameterf(GL_TEXTURE_2D_ARRAY_EXT, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_BORDER);

But this does not work with ati cards/drivers. Nvidia cards/drivers work just fine.

I tested ATI 4850 and 4890.

An example source code file can be downloaded here:

http://www.jotschi.de/download/opengl_texture_array_glsl_example.tgz

You can switch from GL_TEXTURE_2D to GL_TEXTURE_2D_ARRAY by enabling or disbling:

#define NOARRAY

Here can you see that the clamping is not applied when texture arrays are used:

http://www.jotschi.de/wp-content/uploads/2009/10/GL_TEXTURE_2D_ARRAY_CLAMP_TO_BORDER-ATI.png

In contrast to clamping when texture_2d's are used:

http://www.jotschi.de/wp-content/uploads/2009/10/GL_TEXTURE_2D_CLAMP_TO_BORDER-ATI.png

Is there a way to workaround this issue? When will the opengl implementation be fixed?

Greetings,

Jotschi

I am developing a software for windows. This software runs over a maximized window. I use OpenGL for drawing textures, lines and other graphic primitives, but I also draw menus, panels wich are drawn with GDI.

I have found this problem in vista and HD3200. I have tested HD 1600, HD4650 and NVIDIA 8800GTS both in XP and Vista and they all work fine.

Please, because HD3200 is a very extended graphic card in cheap laptops it is very interesting for us to solve this problem.

Does anybody know how can we contact ATI driver support to solve this issue?

Thanks

We're working on a game, which use only OpenGL 3.x (Forward Compatible Context), so I have one question. When will be avaiable CrossFire support for OpenGL apps, without a profile? This is big disadvantage for us, because we can't test scalability our game performance on a Radeon cards. We tried use Doom 3 profile for turn on crossfire mode, but this cause crash our game.

Best regards,

I don't know where to post this topic so I post here

I am a directshow developer, and I am looking for implementation details of audio bitstream.

The new radeon 5xxx are announced as supporting HD audio bitstream, including : Dolby True HD/MLP, DTS HD, AAC,WMA Pro

Microsoft brought some standards encapsulation in windows 7 (see http://msdn.microsoft.com/en-us/library/dd316761(VS.85).aspx)

But this only applies to windows 7

My question is : does your driver follow this implementation for new radeon 5xxx (and also maybe for radeon 4xxx) or is there another approach ?

Can you give details ?

For example commercial players such as arcsoft has brought this bitstream support to their software on your new HD5xxx series and for all operating systems.

I am looking forward to get some details about this

Thank you

Regards,

Damien BT

Any pointers would be appreciated.

Simple effects are easy enough, but is it possible to take a complex effect like Order-Independent Transparency.rfx, export the .FX and .X and textures, and then from the information provided write a sequence of API calls to produce the identical effect?

If so, how?

Another thing that seems to be going wrong is that calling glClear with the layered framebuffer bound is only clearing the first layer of the texture array. I would assume that glClear would clear the whole framebuffer, but maybe I'm wrong (and/or Nvidia's driver is not following the spec.)

Has anyone else done this successfully on ATI hardware yet?

Thanks,
Jon