Developing Games on Intel Graphics
If you are gaming on graphics integrated in your Intel Processor, this is the place for you! Find answers to your questions or post your issues with PC games
509 Discussions

OpenGL: GL_SELECTION render mode returns wrong Z-Values with Intel 915GM and drivers 7.14.10.4764

delle
Beginner
5,088 Views
Hi, when I use the Selection Buffer in OpenGL I get wrong values for fields "Minimum depth value of primitives" and "Maximum depth value of primitives" (See Selection Buffer in OpenGL section 20.020).

This only with Driver 7.14.10.4764, Driver 4.14.10.4363 works fine.

In particular I get the value "2147483648" (0x80000000) for zMin and zMax for all primitives (always the same value).

With the old driver (4.14.10.4363) and other Cards (NVidia and ATI) my Program gets the correct results.

There is some new driver that I can use that correct this?

Regards

A. Delle
0 Kudos
33 Replies
delle
Beginner
2,822 Views
Same problem with Intel 965GM and drivers 6.14.10.4785.

Regards

A. Delle
0 Kudos
delle
Beginner
2,822 Views
Someone (from Intel) read it?

Thank you.

Delle
0 Kudos
Chris_M_Intel
Employee
2,822 Views

Delle,

(Hey, you get the honor of my first post, ever.) I've started looking into this issue, but Icould use a little more information. Do you have a code snippet that you'd be willing to share that shows exactly what calls you're making? I'd like to make that we can replicate the behavior here, and if necessary try to get the drivers updated to work as expected. Alternatively, maybe there's a different way to accomplish what you're trying to do and we can highlight that as well.

0 Kudos
delle
Beginner
2,822 Views
Hi, my Application is a complex Application (An architectural CAD called "Easy Architect 3D", see here for a screenshot ), it is 8 years old and I never noticed this problem.

It is written in Delphi (Object Pascal), I've tried to isolate some piece of code. here is the first part (it draw the scene in "Selection mode"):

Note: SELECTION_BUFFER_SIZE is a constant: 30000
"Self" in Delphi means "This" in C++

glInitNames;
ZeroMemory(@(Self.FSelBuf^),SELECTION_BUFFER_SIZE*sizeof(TGLUInt));
glSelectBuffer(SELECTION_BUFFER_SIZE,@(Self.FSelBuf^));
glRenderMode(GL_SELECT);
glClear(GL_DEPTH_BUFFER_BIT);
Self.FWorld.DrawFastEx(NIL,FALSE); // Draw the Objects
// Back to render mode (see OpenGL Develop Guide)
Self.FSelCount:=glRenderMode(GL_RENDER);
if Self.FSelCount>0 then
begin
Self.ProcessHints;
Self.FSelCount:=0;
end;

Here is an extract of the code of the Method "ProcessHints" that analize the response:

k:=0;
for i:=0 to Self.FSelCount-1 do
begin
nItems:=Self.FSelBuf;
inc(k);
zMin:=Self.FSelBuf; // Minimum depth value of primitives (range 0 to 2^32-1)
inc(k);
zMax:=Self.FSelBuf[k ]; // Maximum depth value of primitives (range 0 to 2^32-1)
inc(k);
zValue:=zMin; // I use zValue to detect the neares primitive selected
......
......


What I see is that the Items are all detected, but I need to detect the nearest one. So I use "zMin".

Objects are drawn (more or less) in this way:

IList:=Self.FItemsList.FList;
glPushName(GL_ITEM_SELECTED); // This is an Item ...
for i:=0 to IList.Count-1 do
begin
Item:=TGLItem(IList);
glPushName(i);
if not Item.Draw then result:=FALSE;
glPopName;
end;
glPopName;


The problem is that zMin and zMax values are completely wrong for all primitives.

Thank you.

Delle


0 Kudos
Chris_M_Intel
Employee
2,822 Views

Well, sure seems like it ought to work logically, at least as I understand it. (Delphi is a little Greek smiley [:-)]) I'm checking with the validation team to see if they have selection tests that mimic this functionality and find out what their results are. In the end I may have to write a little app to demonstrate this on our side and see if I can duplicate the failure for the driver guys. Either that ormy little testapp will work and we can figure out what the difference between that code and yours is.

Chris

0 Kudos
delle
Beginner
2,822 Views
This is a big trouble for me, I currently cannot use my new Laptop (no way to select objects when they are overlapped) a HP Compaq 6910p (Intel 965GM and drivers 6.14.10.4785)....

Please try to solve it.....

Thank you.

A. Delle

P.S. Note that with driver 4.14.10.4363 works fine (zMin and zMax are returned well) with the same binary (tested with my old Laptop a HP Compaq nc6120). If I update the driver on the nc6120 I get the same error, so I'm forced to use my old PC with an old driver... :(

0 Kudos
sleary78
Beginner
2,822 Views
I've also got this issue. However ReadPixels works fine.

My card is the G33/G31 Express Chipset. I have confirmed this works fine on other cards.

No matter what the max and min distances are returned as 2147483648. Thought it was a bug in my code until I checked it out on other cards.

I have a test application that i can give to anyone who is interested? Would like to know if there has been a fix for this?

Stephen




0 Kudos
sleary78
Beginner
2,822 Views
For clarity.. My driver version is 6.14.10.4964 released 6/27/2008

Stephen
0 Kudos
Chris_M_Intel
Employee
2,822 Views

Stephen,

If you've got a test case, send it on over! I'll make sure I can see the same problem here and get it on over to the development team.

Chris

0 Kudos
delle
Beginner
2,822 Views
Hi Stephen,

can you send them (Intel) the Test Case? I cannot select two objects if they overlap...

Delle
0 Kudos
ersfu
Beginner
2,822 Views
Has there been any updates to this problem, I get the same error.
I'm using build 7.14.10.4837.
Thanks.
0 Kudos
delle
Beginner
2,822 Views
Quoting - ersfu
Has there been any updates to this problem, I get the same error.
I'm using build 7.14.10.4837.
Thanks.

As expected :( the problem is present in driver 14.36 XP too....

I begin to think that nobody @ Intel care about OpenGL driver quality....

Delle

1414.36.36

0 Kudos
delle
Beginner
2,822 Views

Hi Intel people...

This is the source code that shows the problem....

Focus on "ZMin" value....

On NVidia and ATI does work, but not with Intel drivers

0 Kudos
delle
Beginner
2,822 Views
Quoting - Fendriss

Delle,

(Hey, you get the honor of my first post, ever.) I've started looking into this issue, but Icould use a little more information. Do you have a code snippet that you'd be willing to share that shows exactly what calls you're making? I'd like to make that we can replicate the behavior here, and if necessary try to get the drivers updated to work as expected. Alternatively, maybe there's a different way to accomplish what you're trying to do and we can highlight that as well.

Hi Fendriss, I've just upload the code snipped (see my previous post: file "picking.zip")...

....so what I've to say ... it's up to you!!!! (six months have passed)

Delle.

0 Kudos
delle
Beginner
2,822 Views
Quoting - delle

Hi Fendriss, I've just upload the code snipped (see my previous post: file "picking.zip")...

....so what I've to say ... it's up to you!!!! (six months have passed)

Delle.

Hi Intel Guys, Is Fendriss still working for Intel or Not?

Thank you.

Delle

0 Kudos
jimbojones
Beginner
2,822 Views
Hi, we're seeing this as well. Is there any update to this? This is a pretty big issue, it basically makes our app unusable on intel chips.
At the moment our only option is to drop all support for intel GPUs and require our customers to use nVidia or ATI only.
0 Kudos
cow9
Beginner
2,822 Views
Hi guys,

We are facing the exact same problem. We have had our product running without problems on nVidia and ATi and it's hard explaining to our customers that their Intel-based laptops won't work 'cos of a driver bug.

Can someone from Intel please respond to this?
0 Kudos
delle
Beginner
2,822 Views
Quoting - cow9
Hi guys,

We are facing the exact same problem. We have had our product running without problems on nVidia and ATi and it's hard explaining to our customers that their Intel-based laptops won't work 'cos of a driver bug.

Can someone from Intel please respond to this?

Hi, more than six month ago I've received thie e-mail from Intel connected to this thread:

>
> Greetings from Intel Software Network Support.
>
> We have been following your technical thread on the Intel IIG forum, and
> we were wondering whether you are doing your development work for a software
> company
(and if so, which one), or whether you are an individual developer.
> If you work for a software company, you might be able to receive some
> additional engineering assistance (as well as marketing assistance) through
> membership in the Intel Software Partner Program (www.intel.com/partner
> )
.
>
> If you'd like us to have an ISPP recruiter contact you with some more
> information, please let us know.
>
> Regards,
>
> Lexi S.
> Intel Software Network Support

After I've answered that I'm a freelance --> Silence....

Bye

Delle
0 Kudos
cow9
Beginner
2,822 Views
Quoting - delle

Hi, more than six month ago I've received thie e-mail from Intel connected to this thread:

>
> Greetings from Intel Software Network Support.
>
> We have been following your technical thread on the Intel IIG forum, and
> we were wondering whether you are doing your development work for a software
> company
(and if so, which one), or whether you are an individual developer.
> If you work for a software company, you might be able to receive some
> additional engineering assistance (as well as marketing assistance) through
> membership in the Intel Software Partner Program (www.intel.com/partner
> )
.
>
> If you'd like us to have an ISPP recruiter contact you with some more
> information, please let us know.
>
> Regards,
>
> Lexi S.
> Intel Software Network Support

After I've answered that I'm a freelance --> Silence....

Bye

Delle

Thanks for the tip. I've just registered. But I still do not know who to contact regarding this issue.
I remember filing a bug report some time ago under some general feedback page but there was no response.
Now let's hope there's someone reading this forum

0 Kudos
delle
Beginner
2,788 Views
Maybe someone at Intel is waiting the day he can write a similar post for the 965GM...

Quoting

Hi Matias,

I hate to be the bearer of bad news, but 945G and 945GM are end of maintenance and will not be receiving any additional driver updates beyond the 15.8 baseline for Vista.

I also would like to apologize for the delay in response from Intel. We have made some changes to ensure that posts will have a response in a reasonable amount of time moving forward.

Regards,
Arzhange


Behaving like that I can do Driver development too...

Delle
0 Kudos
Reply