Page 2 of 2

Posted: Wed Jan 04, 2006 7:56 pm
by QuestOfDreams
Ok, I have 2 solutions to the problem: 1 is using DrawIndexedPrimitive + D3DPT_TRIANGLELIST, the other is using DrawPrimitive + D3DPT_TRIANGLEFAN (the original genesis3d version)
Now I'm not quite sure which one to use because they run differently on my 2 test environments. On one system the 1st solution is faster on the other system the 2nd solution is faster.

I've attached a zip with both versions. It would be great if a few people (the more the better :wink: ) could test them and report back which one runs better. (And of course if it really fixes the actor rendering problem/ other problems occur... )

Posted: Thu Jan 05, 2006 3:20 am
by steven8
I will grab it in the morning when I get home from work and give it a try. Thanks, QOD!

Posted: Thu Jan 05, 2006 2:54 pm
by steven8
They both work. Awesome. Now, which was is better is a good question. I would say DrawIndexedPrimitive + D3DPT_TRIANGLELIST gave me roughly 3 to 4 FPS higher and the cutscene ran quicker (don't know if that is caused by the d3d driver or not), but I'd say it is quicker by a nose. I have not updated to 075 yet. I'll do that soon. Should that make difference with the driver?

My system is:

AMD Thunderbird 1.2 GHZ
512 MB SDRam
Geforce 5200 FX 128MB DDR
81.95 Forceware

Thank you QOD!!

Posted: Fri Jan 06, 2006 12:29 pm
by QuestOfDreams
Come on guys, I could use a little more feedback. This is important since one of them will be in the next full installer.
(On my Radeon 9600 the trianglefans are 5-10 fps faster than the trianglelists, on the geforce4 ti4200 the trianglelists are 2-3 fps faster.)
I have not updated to 075 yet. I'll do that soon. Should that make difference with the driver?
The driver can be used with 072A and 075. There shouldn't be a difference.

Posted: Fri Jan 06, 2006 2:42 pm
by AndyCR
sorry, after work today i'll test it on my radeon 9600pro desktop and intel extreme graphics 2 for mobile laptop.

Posted: Fri Jan 06, 2006 5:15 pm
by steven8
Same results with 075. It makes sense that we somewhat parallel on the FPS, since they are both Invidia Cards. Is there any way to bring the video set-up back into play? The user could choose his or her particular Graphics card and then RF would would map to the better driver for it? That is, if we can get more results to show one driver is better for ATI and one for Invidia. Just thinking aloud. You have solved the problem, as far as I am concerned. 2 or 3 fps either way is a minute problem compared to not being able to run at all.

You're a genius QOD!! Thanks again. :D

Posted: Tue Jan 10, 2006 4:36 am
by Pete
fooling around trying to sort out devs problem...
ran into some odd results testing on different versions. i was trying it on a laptop running inteligrated 82815 to see if i could get it to work.

for all tests where needed (not nout's demo) i edited the d3d24.ini (not sure if these are all the right names. hopefully you get what i mean). left the rf.ini as is.
bpp=16
zbufferd=16
bbuffercount=1
extra textures=0
fsaa=0
pretty sure i left everything else as is.

1. launched nout's conversation demo that came out just prior to 072. worked fine. i installed your new dx7s (trifan and trilist) into it. both worked fine. if you remember this rf still used the video setup.

2. i installed a fresh 072. worked fine.

3. installed trifan and trilist into the fresh 072 install. both wouldn't initialize because of no z buffer on device. why'd they work with nout's demo but not this? and why did 072's default dx7 work?

[edit: i also tried installing 075 over 072. zbuffer failure with the default 075 dx7. it wasn't a clean 072 though. actually it was quite messy. but thought i should mention it as whatever it is seems to have started there.]

seems like as of 072 there was a fall back that worked. so i guess my question is: is whatever caused this due to new features or did it just break?

Posted: Tue Jan 10, 2006 4:19 pm
by QuestOfDreams
2. i installed a fresh 072. worked fine.
Was it version 072 or 072A?
in 072A stencil buffer shadows were added which may have changed the requirements for the zbuffer
there weren't any changes in the driver going from 072A to 075...

Posted: Tue Jan 10, 2006 8:26 pm
by Pete
now that you mention it, bet it was 072. i'll double check later but that sounds like what's going on.

[edit: confirmed... 072. 072A and up won't work on an 82815. zbuffer initialize failure]