PDA

View Full Version : It's raining bugs!



McBain
12-31-2002, 12:22 AM
Just found another possible ati driver bug.

state.light[0].position does not seem to be updated with the catalyst 3.0 drivers

I made a simple phong vertex program with a rotating light. When running exactly the same executable, the light rotates with 2.5 drivers, but seems to be static with 3.0 drivers.

If i have the time, i will further investigate and maybe post sample bug code, but now i'm kinda busy. For every bug i encounter in my program, i have to figure out if it could be a driver bug, or if I just made a mistake.

Thank you ATI (again).

NitroGL
12-31-2002, 10:37 AM
I just tried this with my own program, and it works just fine.

dorbie
01-01-2003, 01:32 AM
Have you ever thought of emailing devrel@ati.com instead of posting suspected bugs here. It's a more appropriate place to report this and you're more likely to get a response. We're all supposed to chase this stuff down before reporting, since usually it's developer error, that's just the way things work out.

[This message has been edited by dorbie (edited 01-01-2003).]

MZ
01-01-2003, 06:43 AM
Originally posted by dorbie:
It's a more appropriate place to report this
I'm afraid it is still not appropriate _time_ to report this http://www.opengl.org/discussion_boards/ubb/wink.gif (hint: New Year!)

McBain
01-02-2003, 05:40 AM
I wrote some test code for the bug
the bug seems to manifest only with some ATI driver versions. NVIDIA hardware does not seem to have the bug.

See source file for more information.

c code can be found at: http://www.cs.kuleuven.ac.be/~ares/atibug2/atibug.cpp

an executable can be found at: http://www.cs.kuleuven.ac.be/~ares/atibug2/bin.zip

a vc6 project can be found at: http://www.cs.kuleuven.ac.be/~ares/atibug2/vc6.zip


BTW: Are bug reports are not appreciated here ? If so, i wont post them here anymore. Also, i thourougly test all bugs before mailing to devrel, and sometimes i need some help to do that. You guys say it are mostly developer bugs, but up to now, i have got confirmation from *each* bug report i sent to ATI.

best regards,
Ares Lagae

zeckensack
01-02-2003, 07:18 AM
Originally posted by McBain:
BTW: Are bug reports are not appreciated here ? If so, i wont post them here anymore. Also, i thourougly test all bugs before mailing to devrel, and sometimes i need some help to do that. You guys say it are mostly developer bugs, but up to now, i have got confirmation from *each* bug report i sent to ATI.

best regards,
Ares LagaeCan't speak for others, but I do appreciate it. It helps avoiding multiple reports of the same thing, if people can see recent bug reports. It may also help others in avoiding the bug, or understanding why some code doesn't work as expected.

And of course it's good for the quality of bug reports too, as you may have overlooked something and it's nice to have a few more eyes look at it before issuing the report.

I say go on with it.

It would also be very nice if ATI representatives were more visibly active here but that's something ATI have to decide for themselves.

ehart
01-02-2003, 07:58 AM
Bug reports are always appreciated by IHV's, especially when they are well isolated like yours.

I forwarded your earlier issues to the driver team and a fix has been implemented and is being tested. I don't think it will make the next major release, because that may have already begun qualification.

As for bug reports, what is most helpful to us (ATI) for tracking them is to send your issue to devrel@ati.com. This ensures the issue gets tracked, and generally you will also get a notification when a driver that fixes it is available. They may ask you to fill out a form to register, but this allows them to give you a pre-release driver with the fix when possible.

I think I need to make a New Year's resolution to do a better job of reading and responding to this forum.

-Evan

pbrown
01-02-2003, 04:59 PM
I agree with Evan -- I really appreciate this kind of feedback as well, particularly when the problem is isolated. The more information you can provide, the better -- "my app doesn't work on your hardware" isn't too helpful.

Posting this info can be helpful to other developers -- I got four separate bug reports within a week or two after I botched the interaction of ARB_vertex_program and NV_register_combiners (in particular, the handling of secondary color) in our driver. After the first report, I looked like a genius when I immediately came up with a root cause and fix on the other three. :-)

I'm on this board on a "when I can get to it" basis, as are guys like mcraighead, cass, and jra101. "When I can get to it" sometimes doesn't happen, although we will usually catch these bug reports. But we might still miss a few. If think you've found an NVIDIA driver bug and haven't gotten a response within a couple days, feel free to drop me an email and I can hopefully find someone to look at your issue.

And thanks to all of you that have helped us with bugs in our driver -- I really appreciate it.

Pat

Eric
01-02-2003, 08:30 PM
Well, if I think I have found a bug (in NVIDIA's drivers as this my main development platform), I usually build a little GLUT app showing the bug. I then ask some other people to test it (we have plenty of NV-hardware in the office but I am sure I asked Nutty and others from this board on some clipping planes issues a while ago) to make sure that I am not totally crazy.

In the end, I usually report to NVIDIA using their bug report address (can't remember which one it is......).

I kind of agree with zeckensack that it can be useful to be aware of driver bugs. Of course, it would start to be bothering if people reported bugs that were actually in their code but that doesn't seem to be the case very often here...

Regards.

Eric