r9-290 VS. GTX980 on a [email protected] 1440p Benchmarks

I got my hands on my friends GTX980 today and though I'd throw it against my 290 and see how fast this rocket on a tricycle could go.

This test was purely done to determine what GPU would run better on my more modest hardware and for the heck of it because I have free-time.

The contenders

EVGA GTX-980 ACX2.0 SuperClocked     PriceLink

2048cores | 4GB vram | CoreClock: 1266Mhz 

SAPPHIRE r9-290 TRI-X      PriceLink

2560cores | 4GB vram | CoreClock: 1000Mhz

TEST BENCH SPECS

[Disclaimer: This is in no way official or scientific, it is simply my rig that I had on hand, I do however think it is a good value and represents of lots of people's systems]

CPU: FX-6350 @4.5Ghz

GPU: Variable (stock clocks)

PSU: Corsair RM750

RAM: 8GB Patriot viper cas10 1866Mhz

MOBO: Gigabyte 990FX-ud3

Monitor: 1440p Xstar Korean

Drivers: [AMD] 14.12 'OMEGA'    //\\//    [GTX] 347.09

OS & OS Drive: Windows 8.1pro on a hyperX3K 120GB

Games Installed on... 6TB raid5 (made of 3x 3TB WD reds)  [this should not affect in-game, only load time]

Software: Fraps 90second benchmark all performed in the same part of the game

Before I get to the numbers I want to say that the 290 and 980 are different cards in different price tiers BOTH are great for the money. 95% of the time the more expensive card wins

===Gaming Benchmarks===    {min/avg/max}

=Metro 2033 [non-redux]=   Highest Settings

1080p

  • r9-290:   44 | 71 | 156 
  • gtx-980:  54 | 83 | 152

1440p

  • r9-290:   26 | 46 | 101
  • gtx-980:  34 | 54 | 104

=Metro Last Light [non-redux]=   Highest Settings (SSAA/off, Tesselation/high, Motionblur/low)

1080p

  • r9-290:    73 |  99  |116
  • gtx-980: 102 | 128 | 157

1440p

  • r9-290:   49 | 66 | 76
  • gtx-980:  65 | 84 | 100

=Crysis 2 Max. Ed.=       Highest Settings (motionblur/low)

1080p

  • r9-290:   68 | 90 | 134
  • gtx-980:  73 | 98 | 160

1440p

  • r9-290:   42 | 56 | 83
  • gtx-980:  44 | 58 | 95

=TombRaider 2013=     (benchmark)   Highest Settings  (Ultra, motionblur/none, TressFX/none)

1080p

  • r9-290:    88 | 113 | 142
  • gtx-980  126 | 158 | 192

1440p

  • r9-290:   58 | 72 | 88
  • gtx-980:  74 | 93 | 114

=Thief=  (benchmark)   Highest Settings


1080p

  • r9-290 (DX11):    30.7 | 50.1 | 82.1
  • r9-290 (Mantle):  59.7 | 72.2 | 94.1
  • gtx-980 (DX11):    42   | 65.4 | 100

1440p

  • r9-290 (DX11):      28  |   45  | 60.5
  • r9-290 (Mantle):  43.5 | 51.2 | 73.5
  • gtx-980 (DX11):   33.2 | 54.3 | 75.1

===Synthetics===      {FPS/Score/Min/Max}

=Unigine Heaven=      (DX11 | Quality High | Tess. Extreme)

1080p

  • r9-290:    61  | 1537 | 20.8 | 138.4
  • gtx-980: 81.4 | 2051 |   23  | 187.5
1440p
  • r9-290:    38.4 |  968  | 16.3 | 91.2
  • gtx-980:   51.1 | 1288 | 19.5 | 123.3

=Unigine Valley=      (DX11 |  Ultra  |  AA8x)

 1080p

  • r9-290:   56.3 | 2356 | 22.2 | 109.3
  • gtx-980:  62.8 | 2626 | 24.0 | 109.0
1440p
  • r9-290:    36  | 1505 | 17.5 | 68.4
  • gtx-980  40.3 | 1688 | 19.8 | 78.6

=-=-=Thoughts=-=-=

The 980 beats out the 290 by about 10-20% in every game excluding mantle

If you have the money for a 980, go for it. It's a beast of a card that eat everything but will come at a cost. However if you're a little tighter on the budget the r9-290 is the better bang for your buck.

I guess mantle really does help relieve cpu bottlenecks.

Pretty nice benchmarks. I think it would be interesting to try some more CPU bound games such as StarCraft II and Arma III to see how the 6350 holds up.

I need to own more games

Friend me on steam and look at my games list tell me if anything is more cpu bound.

Very nice. Would love a star citizen benchmark. Just a thought. Very much a cpu bound game but also needs a hefty GPU. Looking forward to see benchmarks when DX12 arrives. Mantle performance for everyone!

I'm playing SC right now and I'm not noticing any REAL improvements Hitches are still there and mins might be a little better. My cpu lacks IPC and the physics part of the engine is still single core (rumored more multicore stuff when 64bit comes sometime before February)

Nice Benchmarks tumbs up ☺

SC is indeed cpu bound, its probably the cpu which is holding you back. Still your Metro LL numbers are impressive.

NJM's OCD benefits the the community once again. Thumbs up and I look forward to the rest of the benchmarks.

Heaven & Valley added

i love your pics.

i like keeping in mind there is an 80% price difference between the two.

taken with a nikkor 35-105mm macro f3.5 & flash

that piece of poop has some of the worst contrast and saturation, thank god for post

only used it because it has macro and it's my widest (f number) midtelephoto lens atm

I think borderlands 2 and Pre-Sequel are also CPU bound. 

You should have just crossfied your saphire tri x :)

2x r9290's @ 1440p on

Unigine valley

FPS:    
60.8

Score:    
2545

Min FPS:    
22.5

Max FPS:    
102.1

will fire up tomb raider and see what I can get at 1440p

IF if had a second card

Nice. But 10% for like triple the money..meh

I have a second card, just give me yours and you can have fun with the 980 B)

Nice benchmarks, some good food for thought. Good stuff buddy :D 

Good benchmark dude :).

Btw, do you guys found any serious mantle CPU usage benchmark ? I don't know what to buy, fx 8320/50 or i5 4xxx. Are you experiencing  any CPU bottleneck when you're playing with mantle ?

If mantle can help the fx being better than the i5, so dx12 will, i think. In this case, buying a fx 8320 would be more future proof than any i5 or even maybe i7 for half the price.. 8 weak core should be better than 4 stronger core.

I'm on a tight budget and i can't change CPU/MOBO every years :(. If you can help me, you're welcome ^^.

Mantle is not that common of an API and buy based off of mantle is a promise that may not ever be fulfilled.

If you have the money I would go the i5, It's going to be good in all games. The FX parts rely on the game being able to use lots of cores efficiently. However if you have either of the cards in my benchmarks and you're playing at 1080p either cpu will be fine because you're already above playable at max settings.

Make a new thread for build help and the forum will come :)  If you're really on a tight budget the 8320 with a OC may be a good buy,

Thank you, i know that buying based on mantle's results is not that smart, but i'm trying to extrapolate to dx12, that is said to to the same. I'll make a new thread soon, thanks :).

I just wanted to know, how your CPU core usage is when you're playing with mantle activated ? Better than with dx11?