glCapsViewer (and online database)

[phibermon]

Thanks for this Sascha! an incredibly useful tool, even more so as the database grows.

I'm afraid I get an error upon 'Detect caps' :

Invalid object path .

--

XP 64bit SP2, local NTFS volume, No file/folder security issues (admin user), GTX 460 with driver 270.61.

I've extracted the content to a sub-folder and the root of a drive with the same results.

EDIT :

http://zrusin.blogspot.com/2011/04/apitrace.html

This tool allows one to trace GL / D3D calls that are made by an application by utializing pass a through OpenGL32.dll, d3d9.dll etc

Perhaps the output from this tool or somthing similar could be parsed in a such a way that it can be cross-referenced against your database in order to show a developer the cards that support the features they use? statistical techniques could then be employed to highlight extensions that when avoided, provided the greater additional number of suported cards, or to select a card and to highlight unsupported calls and possibly one day provide contextual hints as to unused extensions that may be preferable to those utilized by the app.

If the information is available to use, perhaps statistics from 3D Mark on used cards would provide a useful data-feed to sort functions/extensions by their availability in the wild.

Regarding the searching of the database, even if it contained every card in existance, it probably wouldn't be very big. You could retain an online database but have the app download updated data now and then and do the intensive searches client side, this would help alleviate server load and help avoid possible security holes.

[Sascha Willems]

Could you please post the whole error message (unless "invalid object" is really all that's displayed)? Sounds kinda odd.

But btw. I see you're using XP 64-Bit? Or is that a typo? 64-Bit XP was never a real release and only intended for testing usage, so the problem might come from my function that uses WMI for getting the OS-Name under windows, maybe it's not working with XP 64 bit?

[phibermon]

Yes full error text (other than Press Ok to ignore etc)

You could very well be right. Since I'm the only one that has this issue, it's most likely related to XP 64bit. As it's effectivly not in use it's probably safe to ignore my bug.

I would like to defend XP 64bit however if I may, it's still fully supported (32bit xp had it's support ended, 64bit is still in support cycle) and has a very low memory footprint when compared to vista/7 64bit while still allowing >4gb memory.

It doesn't have the improved scheduling of vista/7 but is used widely in academic institutions as the 64bit windows platform due to it's slight edge in performance and again, low memory footprint. Also a lot of render farms use it for the same reasons.

---

Having said that, yes I know I'm a nutter I despise the 7 interface so much, that I only use it for >= DX10 games. Hopefully ReactOS will be my Windows OS one day.

[vgo]

I get the same "Invalid object path" error on Windows 2003 Server x64 too, the GLView tool works so at least it's nothing OpenGL related...

[Sascha Willems]

Well, as I said it's propably related to WMI, which I'm using to retrieve the OS string under windows. It's possible that the needed WMI object is not defined under your os'es.

So I've compiled an exe without the WMI-capturing of the OS name (just commented out), please try and tell me if it works : http://www.saschawillems.de/misc/glcapsviewer_nowmi.exe

If it works for you two I'll see how I can get around this, as I'd like to include the OS string in order to differentiate same driver versions on same GPUs for different OSes and need the OS name for that.

[phibermon]

EDIT : started posting before you posted.

I can confirm that the new executable works on XP 64bit, you were spot on with your WMI theory.

XP64bit and 2003 Server 64bit are so nearly identical that differentiating between the two is probably not needed. You can probably get away with determining 64bit from directives and then checking the system drive for 'Documents and Settings' which I believe is not present on later 64bit Windows.

A bit of a cheap hack I'll admit..

Oh and thank you very much for your time! it is most appriciated

[Sascha Willems]

Thanks for the information phibermon. If possible please refrain from uploading reports until I get this issue fixed, or otherwise your report will be shown with an unkown OS. I'll try to find a way to detect windows OS information on all available systems, and will release an updated version then.

[phibermon]

I actually just had exactly the same thought as my mouse hovered over upload so I didn't.

Windows API function GetVersionEx will do the trick :

http://msdn.microsoft.com/en-us/libr...(v=vs.85).aspx

EDIT :

Page has a table towards the bottom showing factors of windows version from 2000 to windows 7 :

http://msdn.microsoft.com/en-us/libr...(v=vs.85).aspx

Page has C example covering everything from 2000 to 7 (half way down there are updates, bug fix)

http://msdn.microsoft.com/en-us/libr...(v=vs.85).aspx


Thanks again, will wait until next release before I upload.

[Sascha Willems]

I just made a small update to the database :

• Added extension search
  You can now display only reports that support a given extensions. So if you want to use some special extension, just enter it into the search box and all reports that support it are listed.
• Basic stats
  Number of reports divided by vendor and os are now displayed.

@Phibermon :
Thanks for the links. Though I'd prefer to have just a simple string with OS and version info returned I'll have a look at those functions to see what I can do to get away from WMI. Though WMI is a very nice thing, cause it even allows you to read out the videocard's memory and much much more.

[Sascha Willems]

Win32-Version updated to 0.4a

I noticed on my second PC that the error message even showed up for a 32-Bit XP. So I dumped the WMI thing completly and replaced it with a simple method that reads out the OS product name from the registry. This should workt better, and at least worked on Vista 64 and XP 32. It woulbe be great if everyone that got the error with the old version could test out the new one and tell me if it works!

Download : http://www.saschawillems.de/download...ewer-win32.zip