Overclock.net banner

AMD R9-290 / NVIDIA GTX 780 Ti Bench-Off. First wave complete, with graphs! SpecViewPerf added.

4K views 32 replies 10 participants last post by  rdr09 
#1 ·
So, I've got a 780 Ti (Galaxy reference) coming this week and I already have an R9-290 (Powercolor reference). I plan to run some benches to pit the two against each other using my sig rig. Most reviews/benchmarks you see use the same games all the time, I'm looking at maybe doing some different ones. If anyone has any interest in potential results, let me know what kind of benches you'd like to see run on the two. I've got a fairly decent library of games and also 3DMark. Built-in benches on game engines are good but if anybody has any custom bench runs that they want to see used instead, let me know (and where to obtain said "run"). I may do some overclocked results too, but that's going to be limited as they're going to be using stock cooling.

By the way, benches are going to be at 2560x1440.




I'll update OP with the results / graphs after I've gotten time to run the benches. Please open spoiler tags to view results of each category.





(Despite frame rate, Dirt felt a little smoother on the R9-290 card.)


(Hitman had some jitters on the GTX 780Ti, but nothing serious. The R9-290 was very smooth.)


(GTX 780Ti seemed to be exhibiting some fairly noticeable microstutter during all of the Just Cause tests, also some random flashing colors/shapes, strange display anomalies. The 290 performed flawlessly here.)




(Bench crashed while testing GTX 780Ti a few times and also had some times where the bench wasn't loading correctly and was causing missing textures, bad frame rate, and obnoxious lights. Again, 290 performed flawlessly here.)




(Performance aside, SSAA modes on the R9-290 seemed to do a better job at eliminating jaggies, most noticeable on the 2x test.)


(Despite basically the same Avg score, the R9-290 was jittery and the GTX 780Ti was smoother (though some small textures/decals flashed/flickered). It should be noted that neither would be considered 'playable' with Ubersampling on at 1440p. I might re-test at a later date with Uber off.)


^GTX 780 Ti

^R9-290

UPDATE: Some other things to note:
-The GTX 780Ti has some pretty ridiculous coil whine at times, something the R9-290 actually doesn't exhibit (though I realize this can happen with any card, regardless of branding, my old 7970 did it but it improved with use).
-Both cards are audible while gaming and also start throttling. They have different sounding fans, but neither are of a tone that are real annoying. If you crank up fan speed, this all changes however; there is a reason I prefer to water-cool. Both under stock settings aren't that bad.
-Tested both cards (without benching) with both FarCry3 Blood Dragon and Deus Ex Human Revolution...
-Human Revolution ran about the same performance-wise on both cards (between 105-120 FPS with v-sync at 120hz)... However, there were a few jitters related to latency spikes (or maybe dropped frames) on the R9-290, although nothing too obnoxious. The GTX 780 Ti did play a bit smoother here. The 780 Ti did exhibit some flashing decals (such as bullet holes, etc).
-FarCry 3 Blood Dragon. Here the 780 Ti did have better performance and also smoother game-play. However, it's long been known that FarCry 3 titles weren't a strong suit for AMD and performed better on Nvidia. The performance on the 290 was still good and it was surely playable, but the 780 Ti gave a better experience. Although, the 780 Ti still exhibits moments of GPU usage dropping out of nowhere resulting in low FPS suddenly, which is easily noticeable while playing.
-During FarCry 3 Blood Dragon I finally got to hear the GTX 780 Ti's fan ramp up. It was rather loud during certain scenes of play, not unbearable, but certainly quite audible... Sort of a honkier (lower mid-frequency) than the 290's, which is a bit higher pitched. Louder than I've heard the R9-290's fan get. (Keep in mind, this is both with tested while at stock settings.)
-AMD's picture controls are, in my opinion, better. I'm able to achieve near perfect results while calibrating with the help of Lagom.nl. Nvidia control panel leaves a little to be desired here, I cannot quite get the gamma, contrast, or brightness near as perfect as with AMD when using Lagom.nl. Even at stock, AMD's colors are a bit more consistent. This is nice since it requires less messing with each individual color's brightness/contrast etc. I do calibrate by doing each color individually and not with the "all colors" sliders. Another thing that prevents getting Nvidia perfect is the fact that their sliders "jump" from let's say 50% to 53% or down to 47% from 50%... You cannot select 48%, 49%, etc. Sometimes the value that I feel would work ended up falling in these "between values" that Nvidia does not let you select. Achieving a totally neutral grey is near impossible due to this, and certainly annoys me.
 
See less See more
16
#3 ·
Unfortunately, it's not an unlock-able one. I was hoping it would be. I'm just curious in benching them since I'll have both of them. I will say that with a very minor overclock done in AMD Overdrive (I usually use Afterburner), the 290 was producing a 3DMark Firestrike Extreme score of 4926.
 
#4 ·
Quote:
Originally Posted by TamaDrumz76 View Post

Unfortunately, it's not an unlock-able one. I was hoping it would be. I'm just curious in benching them since I'll have both of them. I will say that with a very minor overclock done in AMD Overdrive (I usually use Afterburner), the 290 was producing a 3DMark Firestrike Extreme score of 4926.
could you do clock for clock and max oc comparisons? i know the 780 Ti can oc higher. what's the max oc of your 290?
 
#5 ·
#6 ·
Quote:
Originally Posted by rdr09 View Post

could you do clock for clock and max oc comparisons? i know the 780 Ti can oc higher. what's the max oc of your 290?
I haven't spent a lot of time overclocking the 290 yet because of the stock cooler. I'll dive in a bit further when I get to doing some tests with the two cards.
Quote:
Originally Posted by AlphaC View Post

Specviewperf 12. Compare to http://www.spec.org/gwpg/gpc.data/vp12/summary.html
creo-01
or specwpc , compare to http://www.spec.org/gwpg/wpc.data/specwpc10_summary.html
Catia-04
Showcase-01
Snx-02
Sw-03
Medical-01

(it has a GTX 780 entry http://www.spec.org/gwpg/wpc.data/techgage/P9X79-E_GTX780/resultHTML.html)

DOWNLOAD HERE: http://www.spec.org/gwpg/downloadindex.html
Thanks for the download link, I'll look into doing these as well.
 
#8 ·
Quote:
Originally Posted by PontiacGTX View Post

BF4 with mantle, I would like for a comparison
I would do if Mantle support was released yet. It was pushed back to some time in January, right?
 
#10 ·
Yes, please.
thumb.gif


I know the 780TI will come out ahead but I really want to see by how much.
 
#11 ·
The 780 Ti is supposed to be delivered today (if winter weather doesn't delay the UPS driver). If all goes to plan, maybe I can get some benching done this weekend. Despite their $300 price difference, I really do not think they are substantially far away from each-other performance-wise. We will see.
 
#12 ·
From where I'm sitting, I'm only really interested in numbers if ....

-120 / 144 Hz Monitors
-identical cooling
-both overclocked to max

Though one test I'd love to see someone do is ..... instead of clock for clock ..... decibel for decibel as fan noise really irks me ....my rad fans generally never go over 865 when running Furmark .... I have to over ride fan control scheme to test Delta T under max fan speed conditions.
 
#13 ·
Quote:
Originally Posted by JackNaylorPE View Post

From where I'm sitting, I'm only really interested in numbers if ....

-120 / 144 Hz Monitors
-identical cooling
-both overclocked to max

Though one test I'd love to see someone do is ..... instead of clock for clock ..... decibel for decibel as fan noise really irks me ....my rad fans generally never go over 865 when running Furmark .... I have to over ride fan control scheme to test Delta T under max fan speed conditions.
Well, my monitor is running at 120hz. However, I do not think either of these single cards are going to push 120 frames on any recent games with eye candy up all the way. With aging games/engines, I'm sure there are some.

I agree with you about the other points... Ideally, one of these two are going to stay in my main system... I'm benching partially to see which one will stay (and will be rewarded with a waterblock/backplate). I know the 780 Ti will outperform the 290 in the majority, but I am interested to see by how much with games other than the usual review-site benchmarks. I will also be giving my own verdict on overall smoothness of play.
 
#15 ·
Ok, It took me a bit longer to get around to starting this than I had expected, but I completed a first round of testing at stock clocks with a bunch of different games (some older, but still). OP updated.
 
#16 ·
Quote:
Originally Posted by TamaDrumz76 View Post

Ok, It took me a bit longer to get around to starting this than I had expected, but I completed a first round of testing at stock clocks with a bunch of different games (some older, but still). OP updated.
Awesome job with the user review! Thanks for spending the timei literally never trust any of the large review sites.

Slightly suspicious what the 780 ti is doing with the huge fps spikes which would skew average fps.

Either way can't be happier with xfire'd 290s on water, you have validated my purchase more then ever. Can't believe how much I saved getting these over a Ti. LOL.
 
#17 ·
That's a great setup, it should do you well. Cross firing would be fun but I'd probably need a more powerful power supply.
 
#18 ·
#19 ·
Well it looks like they changed the mirror for the SPECviewperf download... The other one kept crashing and eventually went offline and now it's working (it points to a different url now).
 
#20 ·
Quote:
Originally Posted by GenoOCAU View Post

Either way can't be happier with xfire'd 290s on water, you have validated my purchase more then ever. Can't believe how much I saved getting these over a Ti. LOL.
I couldn't justify either ") .... the 780 was an ez overclocker with 25% on the core pretty easy .... I'm anxious to see what kinda OCs he gets on these.... so far everything I have seen on both doesn't excite me.
 
#22 ·
OK, I ran SpecViewPerf test for the 780 Ti last night, it took a painstakingly long time... I'm not too impressed by the results, but alas, it's not a workstation card. They are being added to OP. I also reorganized OP with use of spoiler tags to keep it tidy. 290 version of SpecViewPerf test coming soon as well as the other SpecView bench.

Some other things I just felt like adding...

I cannot crank up the settings in source-engine based games on the 780Ti, it crashes every-single-time upon trying to launch game (first was immediately after applying the highest in-game settings). Deleting the .cfg file does not fix it... Not sure what to do. No problems when using the 290.

Deus Ex Human Revolution and also the Missing Link DLC will not launch when using the 780Ti. I've tried many different things including completely uninstalling and re-installing, nothing will make it work... Game launches fine with 290.

Gas Guzzlers Extreme is experiencing rather choppy game-play with latency spikes on the 780Ti... It runs beautifully with the 290... Sadly, this is an Nvidia sponsored title.

I keep running into a GPU usage issue with the 780Ti also... Many times I am playing a game and the GPU usage will not go over 50% and performance is terrible, as is expected... CPU usage was nil as well, so it was not bottle-necked. In Borderlands 2, alt-tabbing out of the game and back in a few times fixed it and was getting 99% again... Is it just me or am I running into a bunch of driver issues with the 780Ti? Sadly, it's reminding me of the horrors of last time I owned a GTX product. I do like the card and am tempted to keep using it if I can straighten out these issues, as I do love down-sampling from 4K in certain games, especially those of which have no good method of AA. Though I will admit, the R9-290 has been much more plug and play. I haven't encountered a single issue with any game I've launched when using it; color me impressed there.

I will continue to report any findings regarding issues and concerns.

EDIT: I also added a few comparison photos of the two cards to OP.
 
  • Rep+
Reactions: AlphaC
#24 ·
Quote:
Originally Posted by AlphaC View Post

Thank you for running specviewperf 12!
smile.gif
Anytime! I'll be running the SpecWPC on the 780 Ti next (maybe tonight)... then I need to pop the 290 back in and test that with those 2 benches.

It just occurred to me what the other Spec bench is, so perhaps I'll just stick with the ViewPerf for the means of this comparison. I hadn't realized the WPC is for the whole pc... heh. I guess I'm really just comparing the GPUs, so...
 
#25 ·
Quote:
Originally Posted by AlphaC View Post

Thank you for running specviewperf 12!
smile.gif
R9-290 SpecViewPerf results have been uploaded. It went by MUCH quicker with the 290. The tests that barely moved (like slide-show levels of low frame-rates) on the 780 Ti were fluent and smooth on the 290. I'm rather surprised by the results. Really, there were only a couple that were pretty slow on the 290 and even then, not as slow as the 780 Ti.

Also, unrelated to quote, but some other gaming related findings added to OP.
 
  • Rep+
Reactions: AlphaC
#26 ·
For some odd reason, applying 16x Anisotropic filtering with the GTX 780 Ti crashes some games... That's just silly right there. For example, any Source engine games do this (other games too, but it has happened with every Source game I tried). The only way to get a source game to boot back up is to enter -autoconfig (or something along those lines) in the launch parameters. This must be some kind of driver bug... Has anyone with Nvidia experienced such shenanigans?
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top