Building a PC for Xenoblade

[akiratheoni]

Thanks for the replies guys. What's the best but cheapest video card that could run Xenoblade? Currently have the GTX770 overclocked on my wishlist right now but I think I might want to go for something cheaper since it's the only game I'll be playing on this PC. Is there something in the $200 range or so (pretty flexible but $420 for the GTX770 is stretching it a bit) that would be able to play Xenoblade at high/highest settings?

[admin89]

If you plan to play Xenoblade at highest settings (4xIR + 9xSSAA) , even a GTX 770 won't stand a chance
Nothing less from either GTX 780 or GTX Titan to do that job
If you only play it at 4xIR , either a GTX 660 or GTX 650 TI Boost is fine

[cluthz]

GTX660 can do 4xIR and some AA (not 9x SSAA but the normal AA is just fine) I could try with SSAA if you want to see how it runs and where it does bottleneck.
I played Xenoblade with GTX285 in Mac OSX (OpenGL) with 4xIR and 2xMSAA, it was fine. The GTX660 is a bit faster in dolphin (it's much faster in PC-games tho, but dolphin is a bit more bandwidth happy).

You will not be unhappy with the performance with a GTX660 in dolphin.
(I have a Asus DCII OC, running boost to 1183MHz and 6400MHz effective RAM clock)

[cluthz]

As I promised, for xenoblade with latest 3.5-1408.

TLR: Xenoblade can run at Maxed settings in DX11 (4xIR and 8x AA (quality 32)), but not DX9 (even 4XSSAA will make FPS drop below 100% from time to time) with a GTX660.

With DX11 4xIR and 4x Samples (quality level 16) the GTX660 uses from 50-70% load depending on where you are. This is with no framelimit on and the game was running from 40-60 FPS depending on area. (This is way faster than 100%, so at 30 FPS the game will tax the GPU even less.
DX11 with max AA (8x quality level 32) it uses close to 100% GPU power and we're GPU bound, FPS "drops" to around 35-40, still much faster than 100%

With DX9 4xIR and 4xSSAA, the GPU load is close to 100% and FPS does go from 20-30 FPS, so DX9 with 4xIR and 4XSSAA is not advisable on the GTX660
This is with Win7 64 bit, rest of my specs are in signature.

With OpenGL and 4xIR and 4x Samples AA around 60-70% load. FPS a bit lower than DX11, (35-45 FPS) probably CPU bottleneck. 16xQ CSAA still manages full speed with OpenGL. (OpenGL tests was done on Win7 too, with 3.5-1335, as 1408 did crash when I use OpenGL).

The test is done running around Colony 6 and Colony 9, there are areas that will be more demanding.

[NaturalViolence]

Well SSAA is more demanding so yeah that makes perfect sense.

[Comae]

Seriously nowadays in 1080p the AA become near useless, you can just put a simple x4 and its cheap and perfect.

[ExtremeDude2]

I think you need to look closer :<

[Comae]

I have very good eyes, when you go further 1080p you dont need to push the AA too far, its a loss of perfs for almost nothing. AA is much more for those who play in lower resolutions.

[Shonumi]

It depends on the game, your display, the type of AA you use and your preferences. Some games look fine to me without AA (but look better with AA anyway). Xenoblade Chronicles comes to mind. Other games I can't stand without AA since the aliasing is quite noticeable even when I increase the Internal Resolution to 4x. Tales of Symphonia comes to mind.

MSAA is less demanding than SSAA (though it technically yields a lower IQ) but if you use OpenGL or D3D11, the performance hits should be minimal for midrange GPUs (like my GTX 550 Ti) in most games even at 3x or 4x IR. GPU intensive games (Super Mario Galaxy 1&2) are a different story. At an absolute minimum people can force FXAA through their drivers, or possibly even other AA methods.

[NaturalViolence]

Comae Wrote:I have very good eyes

Not as good as mine apparently. Or more likely your standards are just lower. Even with 16x SSAA I can still see aliasing at 1920 x 1200. Granted it's not enough to bother me.

As far as only needing 4x I would say that applies to any resolution. Anything past 4x gives you diminishing returns because with 4 samples you have enough to cover both the horizontal and vertical axis sampling. As someone who recently went from a 1366 x 768 display to a 1920 x 1200 display I can attest to the fact that on both displays there is hardly any difference between 4 and 8 samples per pixel no matter what type of AA you're using. SSAA still produces a noticeably better image in both cases though.

For me aliasing is just as noticeable on a 720p native display as a 1080p display. It's just smaller. Instead of a couple of large sawtooths on edges you'll get lots of smaller sawtooths, which is actually more annoying to me. However it depends on whether you're talking about using a lower resolution display or running the game at a lower resolution on a high resolution display. The upscaling involved with the latter tends to make aliasing a lot more noticeable which may indeed require more AA. But I've found that in the case of the former the difference isn't usually very noticeable if at all. Especially since most low resolution monitors are usually smaller and maintain similar DPI/PPI. DPI and viewing distance are far more important to the visibility of aliasing than total resolution.