Doom 3 render request (and how to select your own render!)

[Lonyo]

Well, in B3D's short interview with Carmack, he notes all the render paths possible (I can't believe I missed the obvious setting, damn)

"r_renderer [ arb | nv10 | nv20 | r200 | arb2 ]"
doomconfig.cfg file
seta r_render "xxx" (best by default), you can change the values to what you want.

I have a Radeon 9800
I have run tests with all 5 different renders, as set in the file, BUT, obviously the NV paths may not have worked.
ARB gave 29.6fps on the second run (10x7, med quality), it looked like crap, and was slower than all other renders.

IMPORTANT:
ARB2 gave 38.4 #1
42.5 #2
This matches with "best" setting in the file, so 9800 seems to use ARB2 as default.

R200 gave 36.6 then 39.8, slower than ARB2

NV10 and NV20 (as set in the config file)
NV10: 38.2 followed by 42.3
NV20: 38.3 followed by 42.4
Within 0.1 of each other, and 0.2 of ARB2, suggesting it defaults back to "best" when it finds an unsupported sisplay mode, or it defaults back to ARB2, I don't know.


Could any nVidia users have a play around? Switching between ARB2, NV10, NV20, Best and R200?
I foudn ARB2 to give the best quality, and the fastest speed, which is quite nice 🙂

 

[The Green Bean]

no chinese please

 

[Insomniak]

arb2 is probably best for you. r200 would be slightly faster, but arb2 provides better visual quality (more bells and whistles are on) and I doubt you'd notice the performance difference at all.

On my GeForce 4 Ti, things seemed to run about the same, but not look quite as good when I specified the NV20 render path, so I went back to best for the slightly crisper textures at the same framerate...just my personal experience.

 

[Lonyo]

Originally posted by: Insomniak
arb2 is probably best for you. r200 would be slightly faster, but arb2 provides better visual quality (more bells and whistles are on) and I doubt you'd notice the performance difference at all.

I found ARB2 to be the fastest, faster than R200 and it looked better (thought not hugely different). I found it really odd that ARB looked so crap, and played really badly too.
It might just be my setup though, because I installed the Humus tweak, which affects settings probably used for the R200 path (R200 cards), from what I remember reading of the thread, and may negatively impact R200 path performance. For others, R200 might be faster than ARB.

I don't want to complain about Carmack's work, I still consider him to be the industry leader in graphics engines. Though when I read the shader it striked me how many texture accesses it did compared to the relatively short shader, even for stuff that could just as well be done with math for a small cost in instructions. Using a dependent texture lookup for POW evaluation makes a lot of sense for R200 level hardware due to instruction set limits, but for R300 and up it's much better to just spend the three cycles it takes to evaluate POW with math instead of risking texture cache trashing with a dependent texture read, which may be much more costly, especially since the access pattern in this case will be far from linear. Also, using math improves the quality too, even though it may not be very noticable in this game.

From Humus' B3D thread.

 

[Insomniak]

Well, ARB2 was the render path that Carmack had intended for NV30/R300 and newer cards to run on anyway...so it makes sense that it's the most suited to your R350 there...