Borderlands 2 GPU & CPU Performance Test

I suppose hoping to find out how a GTX 295 performs was a bit too much to hope for

I have an i5 2500k @ 4.8ghz and an Evga 680 SC at stock settings. I play this game on the highest settings (physX at high as well) and I am seeing framerates in the high 70s when I turn off Vsync. I don't think PhysX is taking as huge of a hit as this article states. However, great review Steve. I was having a debate just an hour prior to you posting this as to whether or not Borderlands 2 is harder to run than Borderlands 1.

Just a small note on the PhysX tests, the AMD cards using the High PhysX setting is way too optimistic. Obviously PhysX must be performed on the CPU with AMD cards, and it appears that the reviewer did not use a proper benchmark location. Shooting bullets on the ground or fighting Bullymongs is not good testing for PhysX.

A good test section for PhysX is the BoomBewm boss fight. There is a massive difference in the number of rigid bodies and physics calculations here compared to regular gameplay. Even with an i7 2600k watercooled to 5.0Ghz, the framerate gets crushed into mid 20fps with a HD7970. A GTX680 on the other hand will maintain that 60fps.

Hello Steve,

I saw your article over at Anandtech and had questions about your Hybrid PhysX Results. As someone who's been working on getting improved results with this setup, the fact that you didn't use a second GPU to offload makes me wonder if you properly tested PhysX for your review.

Here are three screen shots of me testing with PhysX Hybrid
System: Intel i5 2500k @ 4.4ghz/GPU1: Radeon HD 7970 @ 1125/1525/GPU2: GeForce 9800 GTX+
Settings: Max/No FXAA/PhysX High
Pic 1: Notice the blood (a PhysX effect) and my FPS with zero GPU2 load (GenL Mod doesn't work, this was me testing with his)
http://I.imgur.com/kLtNP.jpg
Pic 2: Same as above, except the blood cleared, FPS doubled:
http://I.imgur.com/gcXDz.jpg
Pic 3: This is using a different Hybrid PhysX Mod (found at ngohq.com) and this time you see GPU2 has load, and my FPS has increased with blood on the screen:
http://I.imgur.com/1ReMV.jpg

I've since switched my aging 9800 GTX+ to a GTX 460 and only now am I getting 60+ FPS with PhysX effects. I don't see how you got your results WITHOUT a second GPU for offloading. Did you test blood on your end?

Thanks for your time.

No i3 again?! or maybe G series intel 2 core CPU's
i3 is matching with FX x4 and phenom x4 with price

Techspot was surprised the 3770k was outclassed by the 3960X? It's happened in several games including Tribes and Skyrim.
And this is another game that shows you get what you pay for with those Radeons... PhysX is a very obvious advantage and the gaming experience with a GTX is once again the best available on the planet. Period.

Could you please include at least one of so popular SB or IVB - i3, Pentium cpus. :S

We aim to please. From GameGPU.


[Source]

Good article. I've found that this game on maximum detail + FXAA drags down to 55fps on GTX680-SLI 19x12 with a 2700K@4.7GHz. Latest drivers too so any tips to improve this I may have missed?

Do you also know that that the line PhysXLevel=2 for "WillowEngine.cfg" is located ... Documents\My Games\Borderlands 2\WillowGame\Config\LauncherConfig

It generates this on saving the changes in the game menu, simply change that value on exiting the game normally.

What do you mean 19x12?

1920x1200 as LNCPapa rightly said... gives me more vertical height that normal 16:9 displays lack. Vital for a mainly FPS gamer.

I was also wondering how many others were playing with all ingame sliders to the max and what your average framerate was (use Afterburner, Fraps, Xfire, etc).

I still have the feeling this game needs patching to sort performance on this port despite the recent mini-patch on release. I used to use 2K textures for Borderlands and I dread trying that for B2 which still uses 1024. I remember the same laggy feeling with BF3 on Ultra but my GTX680-SLI manages that with ease at 125fps average. Maybe I have it all wrong and these 'Ultra' options are for future GPUs, not current ones. Trouble with that theory is by Q2 next year this game will have be played to death and we'll have grown bored of playing it umpteen times - despite being as amazing as it is

Regarding AMD FX performance:

What happens if you disable every other "core" (the interger half of it) and rerun the game test ? Since FX-4170 is doing quite well in comparison with the FX-8150 this means the game mostlikely isn't using more than 4 threads at any given time so running on 4 real cores (as opposed to 2 cores + 2 gimped integer clusters) should produce even better results.

While you are at it, why not post the Task Manager screenshot showing CPU usage in game like some other review sites do ? This was you can easily tell how many CPU-heavy threads the program is using ...

LOL @ 3x7970's for gaming @ 19x12 and bitcoin mining.

Wow, maybe I should have read the article before posting. The Radeons perform quite well with PhsyX on after the chop mod. Still, its Nvidia's technology.

Can someone notify the author or site that their methodology is wrong and are spreading misinformation. There have been a few comments already addressing their flawed method.

This article should be yanked or marked with a giant disclaimer. Not all PhysX effects will work properly with just a Radeon, had the author bothered to check he'd have realized this.

Shoddy article. NeoGAF/Anandtech already caught on to the misleading conclusions, don't be fooled.

I did some PhysX testing with my i3 in case anyone is interested...

I still don't understand how AMD GPUs can use PhysX just by editing an .ini, when PhysX requires CUDA cores which AMD GPUs physically don't have...

Must be a "Back to the Future" class power usage...

...since HD 7970 time in the marketplace is 8.5 months.
Retail launch 9 January, 2012