260 vs 560 vs 6950 2GB benchmarks - 1680x1050

Bankie

2[H]4U
Joined
Jul 27, 2004
Messages
2,454
I recently upgraded from a GTX260 to a GTX560 and ran some benchmarks. After a few days I started getting artifacting even while on the desktop. It could have been a bad card or the 400/500 series motherboard compatibility issues. Since I bought the card at a Micro Center that's an hour and a half away I figured I'd return it for a 6950 and avoid another trip just in case it is motherboard incompatibility. I splurged and spent the extra $25 for a 2GB card. After installing the 6950 I ran some benchmarks for comparison purposes.

EVGA P55 SLI
i7 870 (stock clock for now) + Scythe Mugen 2
4GB G.Skill Ripjaw CAS7
EVGA GTX260 overclocked OR MSI Twin Frozer II 560 OR Sapphire 6950 2GB
80GB Intel G2 SSD (Windows + Games)
1TB Samsung F3 (Games)
500GB WD (Media)
Win 7 64 bit

Drivers: Nvidia - 266.66 - AMD - Catalyst 11-2

Note: With the 560 I didn't really experiment with PHYSx/CUDA options disabled in the games that offered support. I did run benchmarks without them but didn't write all of them down as I didn't expect to disable them. I've marked when it was from memory so feel free to be believe/disbelieve those numbers.

Note2: 6950U refers to the 6950 with the shaders unlocked. The BIOS I'm using is an edited 6950 BIOS and does not have the increased clocks or voltages.


Just Cause 2 - Concrete Jungle - All options maxed - Bokeh filter/GPU Water simulation - 4xAA

260 - 19.04 FPS
560 - 33.04 FPS

560 - 46.3 FPS - No Bokeh / GPU Water
6950 - 40.9 FPS - No Bokeh / GPU Water
6950U - 41.4 FPS - No Bokeh / GPU Water

I prefer this game with CUDA effects on but they don't add a lot to the game. The 6950 can't do the extra effects but it runs silky smooth at 8xAA. I never actually played the game without the effects on the 560 so I can't really say if the extra 6 FPS at 4xAA added to the experience. I also can't say how smooth it is at 8xAA without the effects as I didn't even try 8xAA. I'll call it a tie just because the 6950 runs so nice at 8xAA.

Mafia 2 - All options maxed - AAA on - Physx High


260 - 22 FPS
560 - 32.7 FPS

560 - 70 FPS - PHYSx Off - I didn't write the number down, I just remember it was in the low 70s.
6950 - 62.7 FPS - PHYSx Off
6950U - 63.7 FPS - PHSYx Off

I prefer Mafia 2 with PHSYx enabled. The extra particle and cloth effects are very noticeable in game. It does hitch a bit on the 560 with PHYSx at High but it's very smooth at Medium. I'd much rather play with the 560 as I find the extra effects to be very cool.

Dirt 2 - All options maxed - 8xAA


260 - 46.9 FPS Min. / 70.7 FPS Avg.
560 - 79.4 FPS Min. / 101.5 FPS Avg.
6950 - 62.6 FPS Min. / 76.2 FPS Avg.
6950U - 59.7 FPS Min. / 72.4 FPS Avg. <-------------- I have no idea why but it's repeatable on my system.

The 560 is MUCH faster in the benchmark but gameplay feels pretty much identical between all three cards.

GTA4 - All options max except for Shadows: High - View Distance: 39 - Object Distance: 50


260 - 49.17 FPS
560 - 64.7 FPS
6950 - 62.52 FPS
6950U - 64.45 FPS

At these settings I find that the 6950 has a small and barely annoying "hitch" to it that the 560 doesn't. It's hard to describe so I'll just say that it almost appears to drop a frame or two every so often. It's still perfectly playable. One area the 6950 does shine is that you can max the Distance settings with little effect on the frame rate. Very High Shadows are still out of reach as they add a slow/laggy quality to the gameplay. The hitch hurts this game with me more than the extra quality settings help so I'd rather use the 560.

-----------------------------------------------------------------------

Other games: I played other games but did not have the desire to install Fraps for benching. STALKER Complete seems like it may be a bit smoother on the 6950 than on the 560 but there is some strange shadow flashing that I don't remember noticing when using the 560 . Borderlands is hard to say since it crashes at startup with the AMD card. Plants vs Zombies runs great on both cards! ;)

Other thoughts: I've been in the green camp for a while because of bad past experiences with AMD drivers. Unfortunately, I've had a not so good experience with the AMD drivers/card this time around also. The first issue was that the driver would not install at first, I had to Google the issue in a 800x600 browser window before I found instructions that informed me that I needed to install the drivers from a command prompt, which did work. Another issue is that I run dual monitors, a DVI and an older VGA monitor, and I had to do another search to find out that the bottom DVI port will not send a signal to a DVI adaptor -> VGA monitor. Overall the issues weren't huge but doing research in a low-res browser window is not much fun.

Subjective hardware quality: The fully enclosed case on the Radeon has the look and feel of a high-quality product. The MSI 560 has a nice cooling setup but it feels a bit flimsy.

Fan noise: The MSI 560 is absolutely silent and was quieter than my 5x 120mm case fans at all times. The 6950 is very quiet at both idle and load but it is a tiny bit louder than the rest of my system though not annoyingly so.

Conclusion: For the resolution I run at and the games I play, I'd rather have the 560 for the extra features and the sometimes higher frame rates. The 6950 is a tiny step down for most of my usage but there's something to be said for having a working card. :D
I'll be keeping the 6950 and may update again once I feel like doing the shader unlock.


Note: Yes, I know my grammar isn't the best. Hopefully it's still readable.
 
Last edited:
Interesting review. Not surprised to see the 560 come out slightly stronger as these are all nvidia-biased titles. Still, as you rightly point out, DiRT 2 is playable on both because the frame rate is a minimum of 60+ on both cards at all times. This is why I exclude low frame rate benches from my comparisons.
 
I've never heard that Dirt2 and GTA4 are Nvidia biased. IIRC, Dirt2 used to be a showcase title for AMD.

I should have run some Metro 2033 benchmarks but I was originally just interested in seeing how much faster the 560 was than the 260 in the games I actually play. :p
 
GTA4 being nvidia biased is a foregone conclusion. Cards are literally competing with their 3 year old counterparts from the other side. You can almost only run the game on a geforce.

DiRT2 is an engine that Fermis are specifically optimised for so you see geforces perform well above average in it too. Any title with PhysX in it should also be obvious.
 
GTA4 being nvidia biased is a foregone conclusion. Cards are literally competing with their 3 year old counterparts from the other side. You can almost only run the game on a geforce.

DiRT2 is an engine that Fermis are specifically optimised for so you see geforces perform well above average in it too. Any title with PhysX in it should also be obvious.

Well any title with PhysX should be obvious but HardOCP frequently shows that the Radeons are just as fast if not faster than their NVIDIA counterparts in Mafia2. So, I'm not quite sure it can be completely labeled as NVIDIA biased. Other reviews show similar results for JC2; at higher resolutions AMD/NVIDIA are close enough that I'm not sure it's that biased.

[H] also shows the 260 as being just barely faster than the 4870 in GTA4.

I'm not convinced that the 6950 would be noticeably better at this resolution in any other games. I think the main thing is that the 560 is just a bit faster at this resolution and that my results would likely start to favor the 6950 if I had a higher resolution display available.
 
It wouldn't be, the HD6950 and GTX560 are pretty evenly matched at 1920x1080. On average though, the 6950 is not slower than the 560, just not appreciably faster.
 
Added benchmarks done with the shader unlock mod. Not much changes; a couple FPS here and there. For some reason Dirt 2 actually lost a few FPS but it's still smooth.

I'll post some benchmarks with 6970 clocks later.
 
Back
Top