R9 290 vs GTX 680 Benchmark & Comparison

Hello again all,

As you probably know, the latest battles in the graphics card department are raging, and I felt like doing some benchmark comparisons for anyone interested.

Basically, I'm rehashing my previous effort with the 6990 vs 680 found here.

Once again, I ran two FRAPS benchmarks for 60 seconds each. I picked an area that I typically see the LOWEST performance in (same place as the last comparison). This way, I know the cards are taxed, and the results are indicative of the worst case scenario.

Benchmark 1: Runway start at Senaki-Kolkhi with the TAD on the left MFCD(default display), and the TGP (White hot) on the right. I made a waypoint in a nearby town, and slewed the TGP to it. This way the TGP display has trees (lots of trees), some buildings, and smoke.

Benchmark 2: Right after benchmark 1 hits 60 seconds, I throttle up and begin the second test with take-off. Once in the air I hit auto-pilot at a 5 degree climb. TGP is still staring at the waypoint.

For consistency, I flew the benchmarks then ran the actual tests on the recorded track. This ensures the exact same experience across every test.

Each benchmark was run on low, medium and high settings. Specific settings, and results are below.

So there were a couple surprises...

First, on low settings the GTX 680 floored the R9 290. If you have either of these cards, you're not gonna play on LOW settings...but I thought it was interesting anyway.

Second, the R9 290 performs about the same regardless of the AA level. I ran 16x AA and got the same results as 8x and 4x. I didn't think to run a 16x test before I took the 680 out so I don't have numbers for that, but the 290 can max out AA with no real hit to FPS.

Final Thoughts: After reading a lot of reviews, I decided to get an R9 290 because it performs very closely to the 290X, and is ONLY $400. I sold my GTX 680 for $300, so my upgrade was pretty cost effective. In my case, the worst case performance of the 290 is about 20-30% better than the 680. For a net cost of $100, I'm happy with that.

HOWEVER...

If you have a 680, I would stick with it another generation, unless you play other games that benefit more from an upgrade, or you are able to make a deal like I did. Outright, I do not feel the increase is worth the cost of a new card. Once EDGE comes out, the option to get another GTX 680 for $300 or less will surely stomp the single 290 option.

As mentioned, this is representative of "worst case". For the average mission, my experience on the GTX 680 was usually ~30fps. The R9 290 is giving me around 40fps. With mirrors on and MFCDs set to 1024 every frame, I'm getting about 25fps.Your mileage may vary. I'm currently running a 1920x1200 main display with MFCDs exported to 8'' lilliputs. This does reduce FPS vs displaying MFCDs in the cockpit. Another consideration is your resolution. As the resolution goes up, the 290 is known to pull away from the 680 (or 770).

I hope this has been informative, feel free to leave any questions or comments below. Thanks!

Edit: Drivers used for testing were Nvidia's 320.49 and AMD's 13.11 Beta 9.2.

Good work, It would be nice to have 1 Standardized Track that everyone can use for Bench marking Purposes.

Here's a link to download the track, in case anyone wants to replicate for their own benchmark/comparing purposes. You'll see the nosewheel steering light flash 3x as the first benchmark begins. You'll also see it come back on at the end of bench 1, and flash 3x again as bench 2 begins. Part 1 fps are pretty steady, part 2 varies as you gain altitude. Recommend running a 60 sec fraps bench to get reliable figures.

http://www.mediafire.com/?h27148511gmxtdq

Good work, It would be nice to have 1 Standardized Track that everyone can use for Bench marking Purposes.

Just one note:

DCS is still DX9-based app, but today's new GFX are not optimised primary for DX9-performance. Then it is quite possible that in *some* scenarios DCS might play better on 1-gen older high-end GFX, than on the very new graphic card. But I suppose EDGE will switch to DX10/11 and *then* you might profit from new GFX more...

R9-290X Drivers havent been DX9 Optimized yet.

I updated the reserved 2nd post to include a link to the benchmark track. Thanks for the idea, SkateZilla.

I'll run this when I get back home!

R9-290X Drivers havent been DX9 Optimized yet.

...and I doubt they will ever be (at least not so effectively as for DX10/11). It is simply not worth the effort. New (today's) GFX-cards are rarely benchmarked (and used) in DX9. And if some compromises had to be done when doing chip-design, I think they'd sacrifice some of DX9-performance for higher DX10/11 score. It is not rare that new GFX has equal (or slightly lower) performance in DX9, than the previous-gen GFX...

Cool, look forward to hearing your results TZeer.

Another thing to think about, isn't the GTX770 the same thing as the GTX680 but with slightly faster memory?

http://www.dailytech.com/NVIDIAs+GeForce+GTX+680+Gets+Tuned+Rebranded+as+399+GTX+770/article31660.htm

Another thing to think about, isn't the GTX770 the same thing as the GTX680 but with slightly faster memory?

 

http://www.dailytech.com/NVIDIAs+GeForce+GTX+680+Gets+Tuned+Rebranded+as+399+GTX+770/article31660.htm

Pretty much

770=680

760=670

Etc Etc

R9-280X = 7970 GHz Ed.

R8-270X = 7870 GHz Ed.

R7-260X = 7790 Refresh

...and I doubt they will ever be (at least not so effectively as for DX10/11). It is simply not worth the effort. New (today's) GFX-cards are rarely benchmarked (and used) in DX9. And if some compromises had to be done when doing chip-design, I think they'd sacrifice some of DX9-performance for higher DX10/11 score. It is not rare that new GFX has equal (or slightly lower) performance in DX9, than the previous-gen GFX...

There's still a large market for DX9 Titles.

They'll start tweaking it after they finish the DX9 Frame Pacing Drivers.

Interesting...

I have on order the GTX 780 so how would the R9-290x compare to it...??:music_whistling:

And will get another GTX 780( SLi both vid-cards ) after once funds re-established..after my new system is finished..

Pretty much

 

770=680

760=670

Etc Etc

 

R9-280X = 7970 GHz Ed.

R8-270X = 7870 GHz Ed.

R7-260X = 7790 Refresh

 

 

 

 

 

There's still a large market for DX9 Titles.

 

They'll start tweaking it after they finish the DX9 Frame Pacing Drivers.

Interesting...

 

I have on order the GTX 780 so how would the R9-290x compare to it...??:music_whistling:

 

And will get another GTX 780( SLi both vid-cards ) after once funds re-established..after my new system is finished..

R9-290X Review w/ Benches Vs 780

http://www.tomshardware.com/reviews/radeon-r9-290x-hawaii-review,3650.html

GTX 780Ti = GK110B (2880 CUA Cores, 15 SMX Blocks, Disabled Double Precision Floating Point)

GTX Titan = GK110 (2688 CUDA Cores, 14 SMX Blocks)

GTX 780 = GK110 LE (2304 CUDA Cores, 12 SMX Blocks, Disabled Double Precision Floating Point)

Check out some reviews of the gtx 780ti and r9 290x...most sites show performance being nearly equal. Same thing with gtx 780 and r9 290. Hard ocp has good reviews on each card and tech report even has scatterplots that give a nice visual.

Thanks for the benchmark. I find it usefull.

Just a question: I see you have an 'unusual' resolution of 3120x1200. Does it play much on the results?

Long version: It does a little bit. I've got SoftTH setting the virtual resolution and the extra 1200px width is only used to export MCDS, so it's not a significant loss. I did do a quick test in 1920x1200 to see how it felt but I didn't do specific settings or run a bench.

Short version: I'll try to rerun the benches in that resolution tonight or tomorrow and post an update for ya.

3120x1200 vs 1920x1200

Alrighty, had a chance to do that second comparison. Here's the results:

Not exporting the MFCDs actually made a sizeable difference on Low settings. On high, there's 2-3 fps lost. 15-18fps is the absolute lowest I've seen, but I'll reiterate that this is intended to represent the worst-case scenario. In normal circumstances I'm seeing 40-45fps, with lows of about 30 when inbound to land in a busy area (trees!). Again, hope this is informative in some fashion!

Thanks for the info, I have just bought an R9 290 along with a new MB CPU RAM PSU and SSD.

One thing I have read in reviews of the 290 is the cards tendency to be quite loud.

Do you have any issues with fan noise levels under load? if so have you done anything to rectify?

Kind regards

Dave.

Regarding the fan noise. I have a GTX295 and it used to be quite loud under load so I've disassembled it and replaced the crap they call a thermal grease with a good one. That's about it. Hope it helps. Regarding the video cards, thanks for testing but mine still is working on high in dcs quite fine. Old Gen n stuff :) but I agree, I'd go with an AMD card, not just cause its cost effective, but also cause Star Citizen will be optimized for them. Good investment for the future upcomin years incase you get the top one.

Some 290s can be unlocked to 290Xs

Some 290s can be unlocked to 290Xs

Anything in particular you can share?

Anything in particular you can share?

So Far the highest success rate is the PowerColor Reference R9-290s, Being Flashed with a Specific Asus R9-290X Bios, to unlock the Shaders and TMUs.

http://www.overclock.net/t/1443242/the-r9-290-290x-unlock-thread/260

So Far the highest success rate is the PowerColor Reference R9-290s, Being Flashed with a Specific Asus R9-290X Bios, to unlock the Shaders and TMUs.

 

http://www.overclock.net/t/1443242/the-r9-290-290x-unlock-thread/260

Thanks for info! Some reading to do as I plan to purchase soon.

Anyone else with a AMD R9 290? ?

Thanks :)

DCS will be DX11 by early next year when EDGE is integrated into DCS.