Quick Question about W2 configuration

[waltc.480]

Quick Question about W2 configuration

Running W2 with my newish 2GB 7850 I find that Ubersampling is a snap (whereas with my 1GB 5770 it was a slide show), and the setting in the .ini I'm using is Unbersampling = 1. I've read on certain sites that there is a Ubersampling = 2 setting which supposedly is a bit better graphically, but when I use that setting I'm not sure of what's happening.../> Is there such a setting that anyone knows about? (I had thought the Ubersampling switch was a simple on/off 1/0.)

Also, with the latest Catalysts I use "enhance application settings" for FSAA, which I have set to on in the .ini. I've read that the game doesn't allow direct overrides by IHV drivers for FSAA settings, but I'm wondering if this setting is a bit of a compromise that might be compatible with the game. It seems to look *better*--but then, that may just be a mind trick I'm playing on myself, too.

Last, I've found that by using the .ini to set the rendering options I can turn FSAA on while turning Sharpen off, which seems to alleviate most of the screen-door dithering effect I see in the shadowing around flotsam. Wondering if there is another approach there that anyone is aware of? (This is not a big issue that upsets me, actually.)

Thanks!

 

[tommy5761]

I always thought as you did that Uber was either on or off and no middle ground . Maybe send a reply to tech support .

 

[astheoceansblue]

When you say "ubersamplng is a snap", what FPS are you drawing and what's the rest of your machine like?

 

[GuyNwah]

Here's the difference:

UberSampling=1 is 16x AF. You can't set this in the options dialog.
UberSampling=2 is Ubersampling as it's commonly understood.

Ubersampling incurs a frame rate penalty of 2x (more on really inadequate cards).

16x AF may incur only a small frame rate penalty.

 

[alextyc1]

Difference between ubersampling 0 and 1 is noticeable?

 

[tommy5761]

Difference between ubersampling 0 and 1 is noticeable?

Definitely noticeable . As in the framerate will be approximately half with it enabled .

 

[alextyc1]

Definitely noticeable . As in the framerate will be approximately half with it enabled .

The framerate drops in half when its set on 2? What about 1?

 

[tommy5761]

The framerate drops in half when its set on 2? What about 1?

Don`t know about 2 . Strictly set to 0 or 1

 

[waltc.480]

When you say "ubersamplng is a snap", what FPS are you drawing and what's the rest of your machine like?

I can only put it like this: all things being equal, with the 1GB 5770 I had to turn ubersampling off because =1 generated a < 10 fps "slide-show" when running the game (all other in-game settings maxed out.) With my 2GB 7850 (clocked to 1.05GHz/6GHz effective ram speed--card will do this ROOB with AMD utilities and no overheating--it's my stock congif, accordingly) the game runs so fluidly I often forget that ubersampling is enabled.

I run all in-game settings at their highest/Ultra settings, except for DOF, which I have turned off in every option instance just because I don't care for the effect.

Of interest to you might be the fact that I'm running an AMD FX-6300 @4.2GHz with stock fans & voltage (AMD seems to be getting great yields with Vishera), and 8GBs of 9-9-9-21-2T system ram at its stock 1.6GHz speed and its stock 1.5v. Win8x64. But as I say, it seems to be the gpu that makes all the difference here.../>

 

[waltc.480]

Here's the difference:

UberSampling=1 is 16x AF. You can't set this in the options dialog.
UberSampling=2 is Ubersampling as it's commonly understood.

Ubersampling incurs a frame rate penalty of 2x (more on really inadequate cards).

16x AF may incur only a small frame rate penalty.

If your interpretation is correct, then W2 will be the first and only game in which my former gpu, a 1GB 5770, had any performance problems running 16xAF.../> Just setting it (abbreviating) u=1 dropped the frame rate in half with the 5770, down into the single digits, I'm sure. Looks like something more than just 16xAF is going on even with u=1. But I'm going to check on u=2 again with my current gpu and be a bit more attentive. Last time I tried u=2 I wasn't sure anything was happening so I set it back to 1. Thanks for the info, though...! Will check back with results...

 

[waltc.480]

Don`t know about 2 . Strictly set to 0 or 1

Parallels what I've always heard...but will check out the 2 setting and get back.

 

[GuyNwah]

Funny that UberSampling=1 was troublesome for you with a 5770; I was running it with a 5670 without trouble. But that was at reduced resolution (1440x900). UberSampling=2 killed my frame rate, by about a factor of 2.5.

 

[waltc.480]

Funny that UberSampling=1 was troublesome for you with a 5770; I was running it with a 5670 without trouble. But that was at reduced resolution (1440x900). UberSampling=2 killed my frame rate, by about a factor of 2.5.

Yea, that *is* strange...! Possibly senility is already mucking things up in my case, but that is the way I recall it--and I did use u=1 with much earlier versions of the game--which may or may not have had an impact on the situation. Ok, I've been testing with my current system (as already outlined above), and honestly I can discern scant difference between the 1 or 2 setting @ 1920x1200 60Hz, vsync off, in either performance or image quality. The frame rate for both settings is completely fluid, with the possibility that the 2 setting may in fact be more demanding--although simply judging it via my perception it is hard to state that categorically. After several runs in the game with both settings, I really find I actually prefer the 1 setting visually as of this moment.

Initially with the 1GB 5770 I was running the game @ 1920x1200, all options maxed to very high or ultra, but with Ubersampling=0, and I was averaging ~25 fps (key word is *averaging*, not consistent) so long as u=0. Soon as I'd set it to 1, though, that's when the frame-rate nosedived into Anyway, that's almost all I can think of...

Also--and this may have nothing whatever to do with the subject because I'm not intimately involved with the game engine, it's possible that differences in texture-reserved gpu ram make a difference, too, with respect to the ubersampling setting--with the 1GB 5770 I reserved 400mb-600mb, but with the 2G 7850, I've reserved 900mb (TextureMemoryBudget=900.) The 1GB card was left with scant frame-buffer room (600-400mb) whereas the 2GB card is left with 1.1GB of frame-buffer after allotting 900mb exclusively to texture caching. Just guessing! But that's sort of the fun of this--we can do inductive reasoning and see what we come up with!

Thanks for the info--very informative. My conclusion is that u=2 works but is of slightly less image-quality enhancement as the u=1 setting, which is a completely subjective reckoning, to be sure.