Current Status of Windows Development

[travists]

OK, after cleaning my system, I ran the game again. Still have the bottem droping out on the fps. Didn't grab exact figures, but system memory was running a little over 50% used, and cpu was bouncing around but usualy between 60% and 75%. Looks like Klauss was right. It'll have to wait a bit so I can order a graphics card. I'll still play, just have to live with it though. Don't see a speed, but if anyone is trying to do a system rquirements it looks like my graphics card is at the low end, 768 MB memory and a 3ghz processor is about as low as you can get... Need more configurations playing the svn to report to know for sure. A list of known good graphics adaptors/chipsets would be nice too.

[klauss]

It's weird... the X600 should be able to handle the lowest shaders.

Can you post stdout/stdext output highlighting anything relating to techniques please?

(to take a look at what shaders are being used)

[travists]

It does, mostly, just bogs way down.

Kinda long, as I've played a bit since my las delete, but perhaps it gives you more to work with.

[klauss]

Holy f**k man, that's gold powder.

Link successful. The GLSL vertex shader will run in hardware. The GLSL fragment shader will run in hardware.

Warning: built-in varying gl_TexCoord[7] has mismatched access semantics between the vertex and fragment shader

Link successful. The GLSL vertex shader will run in software due to the GLSL fragment shader running in software. The GLSL fragment shader will run in software - available

Warning: built-in varying gl_TexCoord[2] has mismatched access semantics between the vertex and fragment shader

Link successful. The GLSL vertex shader will run in software - available number of ALU instructions exceeded. The GLSL fragment shader will run in hardware.

Warning: built-in varying gl_TexCoord[2] has mismatched access semantics between the vertex and fragment shader

Link successful. The GLSL vertex shader will run in hardware. The GLSL fragment shader will run in hardware.

ERROR: 0:34: '+' : wrong operand types no operation '+' exists that takes a left-hand operand of type 'in 4-component vector of float' and a right operand of type '3-component vector of float' (or there is no acceptable conversion)

ERROR: 0:34: 'assign' : cannot convert from '4-component vector of float' to '3-component vector of float'

ERROR: 2 compilation errors. No code generated.





Fragment Program Error: Failed to compile planet/atmo_ps1.4

Compilation of technique earth failed... trying 0_fixed_gl/earth

Sadly, I don't have the full path for successful techniques:

Compilation of technique fixed successful

Compilation of technique default successful

Compilation of technique default_simple successful

Compilation of technique 0_fixed_gl/earth successful

I guess I'll have to enhance logging a bit.

[pheonixstorm]

Thats what I did with units.csv when it kept telling me it couldn't load for reason blah blah.. but that still doesn't sit well as its finding units.csv to have a size of -1.. but it still gets loaded anyway...

[travists]

Thats what I did with units.csv when it kept telling me it couldn't load for reason blah blah.. but that still doesn't sit well as its finding units.csv to have a size of -1.. but it still gets loaded anyway...

Told you guys I played it for a bit without erasing the output files. That's been my experience. Occasionally something will display as “load failed”, but it plays quite well. Just bogs down, and I can't run the higher shaders. "Gold powder"? Is that good, bad, or bad for me, but a valuable resource? Glad all this means somthing to somebody. My path to bin and data is: C:\Program Files\Vega Strike\Desktopvsnew, if that helps.

Ah well; I may not be good, but I am a programmer. I'm sure that any debug dumps that I do are similarly incomprehensible to anyone that doesn't know the code.

[pheonixstorm]

The shader dumps give good references to why it failed and where.. at least good for klauss since he was one of the ones who wrote the horrid things

[log0]

Some more shader errors(Ubuntu+fglrx). Most of them are due to mixing vec3, vec4 and floats in calculations, should be pretty easy to fix.

[klauss]

The shader dumps give good references to why it failed and where.. at least good for klauss since he was one of the ones who wrote the horrid things

Not only that, it's a clue that some shaders are compling but running on software which is horrible for performance.

I've seen an nVidia extension once to control whether you wanted to allow software emulation or not, but I couldn't find it before the lights went out at home (power outage... dammit).

In any case, besides fixing those vec3/vec4 inconsistencies, I'll have to make VS make sure all shaders really run with hardware acceleration.

[travists]

Hummmmm.... Thanks, I'll check directx and my settings. There may be a switch...


Catilist doesn't like the new drivers, but got it to launch and turned off their AI thing....



I'll earse my files and play a bit. Post files again later.

[travists]

Here goes! No change, ag base and Diplomatic center are two that kick me when getting close. Alwase running into lots of ships, several sozen + ringed planet = BAD fps


Started with avrage shader to get that crash in there. Then played it for a bit, it crashed a few time on me there. It also worked ok (just for some varity)

[klauss]

I'm at work now, but I'll sift through those logs and create tracker items for them so I can address them when I get home

[klauss]

Just committed some fixes in revision 13018 and 13019.

It should solve some of the problems in those logs, and the performance drop, but please confirm (and attach fresh logs).

[travists]

May be a day or too, but will do!

[travists]

Had some time when I got home. Current through 13020, still crashes at dip. ctr. Ring systems around planets bog it way down still, many ships seems to as well. Rings don't appear thick enough to have to dodge rocks, why arn't they just a transparancy maped disk/torus? Anyway... here are the files. Don't crash it for everyone to get it working on mine. I did dump the old ones before starting this time.

[klauss]

Rings don't appear thick enough to have to dodge rocks, why arn't they just a transparancy maped disk/torus?

They are indeed just transparency mapped disks.

There are plans to make them better, but atm (and for a while yet) they're simple transparency mapped disks.

BTW: did you rebuild VS? Because I see no indication that the software emulation detection code kicked in.

[travists]

rebuild? I've heard of it, but mostly in referance to the devel files. ie loading the visual studio C files and compileing it. so I guess no... How would I do that?

[klauss]

Ouch... windows... sorry, yes, windows is hard to build.

Can someone with windows-building-powers attach an updated binary so we can test this thing please?

[travists]

Where I not on satilite I'd finish the downloads to run vs on my linux stick. As it is I have to be kinda carfull, only so much bandwidth per month. But where I live, it's that or dialup. At 28.8 on a good day none the less!

[log0]

Highend still fails on fglrx.

Some info on texture..Lod: http://lists.freedesktop.org/archives/m ... 02024.html

[klauss]

Highend still fails on fglrx.

Some info on texture..Lod: http://lists.freedesktop.org/archives/m ... 02024.html

Good... I've been googling for that bastard with little success.

Now the ARB version of the extension: ARB_shader_texture_lod

[pheonixstorm]

Yeah, I shall post one here and comitt a new build to the svn as well. Been falling a little behind here havent I

[travists]

Your plate is quite full, pheonix, it could easaly be a full time job to keep this going. Several even! Do it when you can, don't feel rushed.

[pheonixstorm]

Grab the VegastrikenonPGO.exe off the svn. The zip file is too large.

[travists]

Ok, I've upgraded the svn. It no longer crashes on high shader settings. I'll Post logs shortly. I also will have my new graphics card comming soon and maybe then I can see the shader rendered planet textures. As it is right now, ocean worlds are still big bals tinted only by the global lights.