DuoCore

[imbaczek]

for the 8 series nvidias, we'd need a free OpenGL profiler and someone willing to run it, since I believe none of the devs actually has a 8xxx.

I heard somewhere that a profiler is in the works, but don't remember the details, does anyone have more knowledge?

[clericvash]

Multicore, one more advantage of the OSRTS project.

OSRTS seems to be going no where.

[AF]

Indeed the expected time to completion imo has risen from 2 years to 20, with the probability of an announcement confirming tis demise rising from 20% to 90%.

Put it simply I could do a Command Engine clone quicker, and that includes learning all the encessary stuff and designing it and the 5-10 years of stuff todo it.

[Neddie]

AF, this is the sort of statement from you that people seem to dislike. Not because it is arrogant and offensive, but because it is arrogant and you have no intent to do that anyway.

[AF]

I have yet to make a commitment that I have no intention of keeping in that post.

1 : exaggerating or disposed to exaggerate one's own worth or importance often by an overbearing manner <an arrogant official>

Weighing up the facts, I don't think that a 5-10 year time frame is unrealistic or exaggerating my self worth at all. Look at my work ethic, the projects I have worked on already, and the degree I'm currently working through and I'd say its possible. Of course it's obvious I to anyone I wouldn't actually do it, I doubt most people would even consider it, and I did say it was an opinion, possibly humour?

What's more how is it offensive? Is simply stating that Command Engines near total lack of progress in the last 4-5 months increases the likelihood of the project ending before anything usable is produced an offensive statement?

Or is it an opinion based on experience and common sense? If I took a snail and calculated how long at its current speed it would take to travel 1000km and found it was a gigantic number, should I apologize to the snail? Or should I factor in the fact that water could theoretically travel at the speed of light given infinite momentum or zero mass and recalculate the time for the snail based upon the value of c?


What you said there had no business in this thread. I admit my post was not angelic, nor will it win me any nobel prizes, but such a reply belonged in a private message. In future please follow your own guidelines.

[SwiftSpear]

I didn't say CE was doing awesome and everyone should be on the edge of their seats waiting for it to release, I'm simply pointing out that there are still advantages to what was originally conceived there. The project lost a lot of momentum with lua being ported into spring and people realizing there were alot more gameplay options all of a sudden.

I'd still love to see CE done some day, something that can really use multicore processing, the best of OGL graphics and an incredibly procedural game development system. It's still a project I'd love to see, even if for all intents and purposes it's currently stalled worse than a windstar in a tarpit

[LordMatt]

for the 8 series nvidias, we'd need a free OpenGL profiler and someone willing to run it, since I believe none of the devs actually has a 8xxx.

I heard somewhere that a profiler is in the works, but don't remember the details, does anyone have more knowledge?

I would be happy to run it, though I don't really know what it is.

[Tobi]

I used the trial version of this one for figuring out (part of) the cause of the white grounddecals. I think it is a profiler too (I used it only for state debugging).

[imbaczek]

Cool. Now I only need an 8800

Seriously now, does it understand debugging information so we know where those slow OpenGL calls come from, or can the binary be stripped?

LordMatt, you could try using this piece of software and tell us what calls take a lot of time, this will surely help.

[Tobi]

IIRC it did show stacktraces (and breakpoints in any OpenGL function), but again, I didn't test the profiler part of it, so I don't know if that links to debugging information too.

The feature I found most useful was diffing of OpenGL states: by diffing the state with black decals with the state with white decals I easily figured out which OpenGL state was wrong. (I still couldn't find where the wrong state was set though, I think kloot figured that out )

[jcnossen]

The project lost a lot of momentum with lua being ported into spring and people realizing there were alot more gameplay options all of a sudden.

Yeah.. when the CE project idea started, I had no idea a canadian called trepan was about to add like 700k of lua binding code :)

Maybe the project can get started again if I get a terrain renderer working so there is some kind of world to experiment with... But unfortunately I have basically no time these days.

[AF]

Ill run it with my 8800GTS 640MB later...

[AF]

Im getting an error:

Process event - OpenGL error:
Function Name: glTexParameterf
Error code: GL_INVALID_ENUM.
Error description:
An unacceptable value is specified for an enumerated argument. The offending command is ignored, and has no other side effect than to set the error flag

followed by masses of:

Process event - OpenGL error:
Function Name: glPopMatrix
Error code: GL_STACK_UNDERFLOW.
Error description:
This function would cause a stack underflow. The offending function is ignored, having no side effect other than to set the error flag

[AF]

Theres major lag when runnign ti when moving th camera around, at least several seconds worth of freeze when pressing tab to view the whole map.

I notice in past games on thsi PC Interface Draw takes a lot of the resurces.

At the end these were the top 3 entries int eh function call statistics window:

glCallList, 63.35%, 121
glTranslatef, 3.14%, 6
glMatrixMode, 2.62%, 5

http://rapidshare.com/files/58884334/stats.zip.html

[AF]

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

Installing perfhood 5 and running the previous test with a vs2005 debug build.

[DemO]

Just like to say thanks to AF and the others who are actually attempting to determine the issues with 8x00 series cards in Spring. Hope you get somewhere with it.

Really would like to at least have the option to play Spring again on my new PC :)

[AF]

Debug string: Detected error: The debugged process asked for an extension function pointer (wglChoosePixelFormatARB) from one render context, but called this function pointer in another render context (context #0)

This line is followed by a gigantic slew of the following error:

GLExpert: OpenGL Error report; A provided enum value is out of range
OGL Error Code = 0x00000500

For some reason I cant get ti to show OGL in th little graph bit.

Anyway the GLExpert error re-occurs repeatedly



This is the first instance of the error.

Here is another from the third instance fo the error taken from context 2 (the other contexts were 'no cotnext' and 'delete1')




Note that at this point spring has yet to even get past the initial menu with 'commanders' or the AIs on it.


I also figured out howto get it to show file names and line numbers. Goto tools->options then check the source fodler and paste the path to your rts folder in the textbox.

[AF]

And mroe errors, this time with spring line numbers:

grassdrawer.cpp line 316
http://img517.imageshack.us/img517/2217 ... er3vt7.jpg

luaopengl.cpp line 753
http://img515.imageshack.us/img515/4673 ... er4dl3.jpg

That last 1 repeats a heck of a lot coupled with stack Underflow warnings




This is all using svn source

[imbaczek]

keep up the good work, mate.

[AF]

There's not much more I can do as the last issue I mentioned happens so often It would take me weeks to get a minutes worth of game play going.