GeForce GTX 780 Ti vs. Radeon R9 290X 4K Gaming @ [H]

FrgMstr

Just Plain Mean
Staff member
Joined
May 18, 1997
Messages
55,602
GeForce GTX 780 Ti vs. Radeon R9 290X 4K Gaming - It's time to take the GeForce GTX 780 Ti and Radeon R9 290X and see what kind of gameplay experience we can get on an Ultra HD 4K display. We saw AMD very much outpace NVIDIA at these 4K Ultra HD resolutions previously. We will find out which one dominates the super-high-resolution gaming scene.
 
Very nice review. Did the 290X throttle at all, or did it maintain 1GHz clock speed? What boost speed were you seeing on the 780 TI?
 
I'm quite impressed by the progress AMD has made. It's great to see that level of performance come down to the $550 price point. Things will definitely get interesting once custom cooler cards hit the market!

On a side note, any idea when those 331.70 beta drivers are going to be released? I can't find them on Nvidia's website.
 
Very nice review. Did the 290X throttle at all, or did it maintain 1GHz clock speed? What boost speed were you seeing on the 780 TI?

I don't think the 290X throttles in Uber mode. I even slightly overclocked mine to 1050 and left 3DMark running for about an hour and it never once throttled down from max core.
 
Very cool review. Nice to see some "out of the box" reviews with these 2 cards.

And for the record, I agree with y'all using Uber Mode. Makes sense to see what a card can do when it's wide open. Not testing at the cards fastest speed would be like testing the H100i only in "silent" mode.
 
Great write-up. Thanks guys.

Makes me even more excited that the R9 290 is $400. It's nearly as fast as an R9 290X and comes in $300 cheaper than the 780 Ti. The fact that you can Crossfire R9 290s for only $100 more than a single 780 Ti..woo. Great time to be a GPU whore. :)
 
Very nice review. Did the 290X throttle at all, or did it maintain 1GHz clock speed? What boost speed were you seeing on the 780 TI?

Only FC3, but even with that, it was performing better than the 780 Ti in that game, with a higher gameplay experience as well. Other games weren't throttling at Uber. I did not check the boost speed 780 Ti was obtaining.
 
I'm quite impressed by the progress AMD has made. It's great to see that level of performance come down to the $550 price point. Things will definitely get interesting once custom cooler cards hit the market!

On a side note, any idea when those 331.70 beta drivers are going to be released? I can't find them on Nvidia's website.

Hopefully this week, they were the drivers provided by NVIDIA to launch with and evaluate 780 Ti.
 
Yeah ManuelG over on guru3d forums stated that the 331.70 is a press release driver for the 780 ti launch and that Nvidia will be releasing a newer driver for the public later this month.
 
I'm really happy with my 290X, and have to agree with the gold award you guys gave it. The gameplay experience it provides for the price is truly incredible. I'm glad to see that it also keeps pace with nVidia's best.
 
Rockin' cards to run games usefully at 4k on it's own- let's hope Nvidia gets the drivers for SLI performance at 4k up to par by the time Brent gets there :).
 
So what about all the guys claiming that R9 290X dominates a GTX 780 Ti in 4K? Quite evenly matched cards. I would still take the GTX 780 Ti over the R9 290X due to vendor specific features that I like (3D Vision, PhysX, Adaptive Vsync, Shadowplay etc.).

Should have overclocked the cards as well. :(
 
I am so glad you guys took the time to clarify testing in Uber versus Quiet mode.

Based on what we just read, I can't wait to see 4k reviews of both video cards in CFX/SLI. Somehow I am not surprised that you couldn't maximize the graphical options in 4k.
 
I'd like to see some testing of SLI / CFX or maybe even three-way SLI/CFX if you can :)
 
So what about all the guys claiming that R9 290X dominates a GTX 780 Ti in 4K? Quite evenly matched cards. I would still take the GTX 780 Ti over the R9 290X due to vendor specific features that I like (3D Vision, PhysX, Adaptive Vsync, Shadowplay etc.).

Should have overclocked the cards as well. :(

Same gaming experience for 150$ less is not something one can easily dismiss unless you truly cant live without NVidia only features which to me are not worth the 150 premium but that's just me. Never could get used to playing 3d or even watching 3d movies and PhysX just takes away too much perf when enabled and isn't used by enough games to even matter. Adaptive Vsync is more interesting to me but that's still a ways off to matter this gen and shadowplay doesn't matter at all to me.
 
Last edited:
So what about all the guys claiming that R9 290X dominates a GTX 780 Ti in 4K? Quite evenly matched cards. I would still take the GTX 780 Ti over the R9 290X due to vendor specific features that I like (3D Vision, PhysX, Adaptive Vsync, Shadowplay etc.).

Should have overclocked the cards as well. :(

Well if we wanna get real technical here, with those $150 you could easily buy a waterblock and OC way past Ti levels. But in all honesty, I think CFX and SLI is gonna be where it gets good, since you will actually be able to play with good FPS me thinks.
 
I have a question:
When 780s released, among with the referance models, almost simultaneοusly were released aftermarket models as well with great cooling systems.
On the case of 290X, we are still waiting for decent aftermarket cooling systems. The only aftermarket models have been released so far, are from gigabyte, sapphire, xfx, and at all of them the cooling systems is identical to the referance. (* or at least they look identical)
So the question is what is the reason for this delay?
 
True. I don't expect to play at 4K without SLi or CFX. I barely manage 120 or 144 fps vsync with my 780 Sli at 1080p.
 
So what about all the guys claiming that R9 290X dominates a GTX 780 Ti in 4K? Quite evenly matched cards. I would still take the GTX 780 Ti over the R9 290X due to vendor specific features that I like (3D Vision, PhysX, Adaptive Vsync, Shadowplay etc.).

Should have overclocked the cards as well. :(

Getting there, much to cover :)

I also have to say, I have a personal interest in ShadowPlay for myself. However, ShadowPlay under Windows7 I have found out is limited to only 4GB video files, about 15Min of recording gameplay footage. You will have to run Win8 to have unlimited file sizes and gameplay recording time. I'm on Win7 on my main gaming machine, so to make ShadowPlay work for me as well as Fraps does I'd have to upgrade to Win8, something I'm not too happy about. I may do it though, because I want to experiment with ShadowPlay for replacing Fraps for recording game videos. I like the fact I can run at my native 2560x1600 and no matter what it will scale the video to 1080p, with fraps I have to run the games at 1080p if I want to record video and output at 1080p. I also like the fact ShadowPlay can use the GPU to transcode the video on the fly, and the format is perfect, saving my poor CPU what would take an hour to transcode.
 
Getting there, much to cover :)

I also have to say, I have a personal interest in ShadowPlay for myself. However, ShadowPlay under Windows7 I have found out is limited to only 4GB video files, about 15Min of recording gameplay footage. You will have to run Win8 to have unlimited file sizes and gameplay recording time. I'm on Win7 on my main gaming machine, so to make ShadowPlay work for me as well as Fraps does I'd have to upgrade to Win8, something I'm not too happy about.

... WTF?!?! That is a @#* up move by nvidia on shadowplay.


I was looking forward to getting a nvidia card primarily because of shadowplay but now it seems that I still might have to buy a video recording program so now that factors into cost for me.
 
... WTF?!?! That is a @#* up move by nvidia on shadowplay.


I was looking forward to getting a nvidia card primarily because of shadowplay but now it seems that I still might have to buy a video recording program so now that factors into cost for me.

Indeed, ripped straight from the reviewers guide for ShadowPlay:

Q: How long can I record for?

A: On Windows 8, ShadowPlay can record up to 20 minutes of video in Shadow Mode. There is no recording limit in Manual Mode.

On Windows 7 or Windows Vista, ShadowPlay can record up to 10 minutes of video in Shadow Mode and up to 4GB (around 15 minutes) of video in Manual Mode

and in another part explaining the benefits

- Records up to the last 20 minutes of gameplay in Shadow Mode (10 minutes in Win 7)

- Records unlimited length video in Manual Mode (Up to 3.8GB in Win 7)

So to get the most out of it, Windows 8 is required.
 
As usual, nice article. Going to skip the first round of monitors I think now that I see I can't play most games at max detail. Save me some coin next year. Curious as to what adding another card will do to the stats.
 
So what about all the guys claiming that R9 290X dominates a GTX 780 Ti in 4K? Quite evenly matched cards. I would still take the GTX 780 Ti over the R9 290X due to vendor specific features that I like (3D Vision, PhysX, Adaptive Vsync, Shadowplay etc.).

Should have overclocked the cards as well. :(

Agreed. The low fan speed throttling issue with the 290X cannot be ignored on my part, although some people will happily live with the loud fan speed in uber mode. And yes, it is loud, it's the same noise as was produced by the 7970 (and I owned the 7970). It's all subjective but i'd opt for the 780ti for the same reasons as quoted, plus you get quiet performance without the excessive throttling that the 290X has issues with when using low fan levels. I'm not cool with how AMD has seemingly regressed to Fermi levels of heat and noise, which is quite odd because I remember AMD launched an ad campaign poking fun of how hot and loud the Fermi was. Oh how things have changed.

I think the 290X is a great chip with insanely good high resolution performance, but for me personally the shit reference design kills it for me. Which is too bad because I really liked prior ATI/AMD cards.
 
Great article, guys. The $/performance ratio on AMD this round is exciting. Hopefully in the next 18 months we can get some good 4k displays on the cheaper side.
 
Getting there, much to cover :)

I also have to say, I have a personal interest in ShadowPlay for myself. However, ShadowPlay under Windows7 I have found out is limited to only 4GB video files, about 15Min of recording gameplay footage. You will have to run Win8 to have unlimited file sizes and gameplay recording time. I'm on Win7 on my main gaming machine, so to make ShadowPlay work for me as well as Fraps does I'd have to upgrade to Win8, something I'm not too happy about. I may do it though, because I want to experiment with ShadowPlay for replacing Fraps for recording game videos. I like the fact I can run at my native 2560x1600 and no matter what it will scale the video to 1080p, with fraps I have to run the games at 1080p if I want to record video and output at 1080p. I also like the fact ShadowPlay can use the GPU to transcode the video on the fly, and the format is perfect, saving my poor CPU what would take an hour to transcode.

Using Open Broadcaster Software, I can transcode game footage on my 7950's while streaming as it supports OpenCL. OBS can also record game footage to a hard drive from what I understand, but I haven't felt the need to do so. If you don't mind slightly lowered quality, it can even do the transcoding on the fly using the GPU on your Intel processor. So you can stick with Win 7 64 and get what you need done.

It's completely free to boot. :)
 
@Brent_Justice. Somewhat off topic, but how do the 4K monitors look when you select a lower resolution? I was going to grab one of the newer displayport models next year, but the review got me to questioning my sanity if I'm going to have to lower my settings that much to game at an acceptable frame rate. Does it look really crappy when you run 1600p, 1440p, 1080p? I remember CRT crispness so I'm being careful about avoiding LCD washed out colors when running lower resolutions as it irks me to no end.
 
Awesome! This have to be "the" article right now in the interweb which clears all the smokes surrounding these flagships. These benchmarks not only show that both cards are equally matched, they also clearly demonstrate that even for these dreadnoughts, 4K is still somewhat 'terra-incognita', at least in this solo act! I'm sure a CFX/SLI head-to-head is in order and would definitely increase the level of playability. Will be waiting for that!

The only thing amiss in this test is an overclocked R9 290!!
 
I thank you for the great review.

This is something I may or may not ever do, due to the prices of these monitors, but to me it looks like a dead even match.

To have great IQ I think it's only logical at this point to assume you need SLi or Crossfire.

The $150 difference make sense when you look at the heat/noise.....for my taste, I am going to have to watercool my 290xs to reduce heat (and overclock the card) and have no noise.
The 780s are practically silent at any fan speed. Therein lies the difference and negates the price difference.
 
Correct me if I am wrong but isn't "uber mode" basically an overclocking switch?

If so, why not compare it to an overclocked 780 Ti? I mean with the EVGA precision utility, overclocking the Nvidia card is done with a mouse click.
 
I'm really excited to see some temp specs on the non-reference 290x cards. I don't quite trust the long time use of a card that is going to sit around 100°c constantly. I tend to keep cards around for many years before retiring them, as you can see in my sig. :p
 
Last edited:
Great Review! So seeing these results at 4k, I think my resolution (2560x1600) my be the best right now for a single monitor setup. With the 4k you have to make too many sacrifices. Tomb Raider for example - I debated about playing at 1600 or lowering to 1080p and enabling TressFx. With my 680, that was the choice. I think one of these cards (or even a 290) will get me a better experience.

It seems pretty obvious, that if you are gonna go 4k, you will need to go SLI or Crossfire. I would bet that a 290 crossfire setup would rock at this resolution.

I am so happy to see AMD coming in at these price points. There is now good upgrade option for me without spending the better part of $1K!
 
Correct me if I am wrong but isn't "uber mode" basically an overclocking switch?

If so, why not compare it to an overclocked 780 Ti? I mean with the EVGA precision utility, overclocking the Nvidia card is done with a mouse click.

Because Uber mode is an official AMD setting. Just like GPU Boost is to nvidia.
 
Correct me if I am wrong but isn't "uber mode" basically an overclocking switch?

If so, why not compare it to an overclocked 780 Ti? I mean with the EVGA precision utility, overclocking the Nvidia card is done with a mouse click.

Uber mode increases the fan speed curve to reduce throttling of the core. Exactly the same as GPU Boost but with a hardware control instead of software.
 
So what about all the guys claiming that R9 290X dominates a GTX 780 Ti in 4K? Quite evenly matched cards. I would still take the GTX 780 Ti over the R9 290X due to vendor specific features that I like (3D Vision, PhysX, Adaptive Vsync, Shadowplay etc.).

Should have overclocked the cards as well. :(

Kind of like the guys claiming the 780 ti dominates a 290x? ;)

Good article anyway, interesting to see NV catch up closer in 4k with the 780 ti.
 
Correct me if I am wrong but isn't "uber mode" basically an overclocking switch?

If so, why not compare it to an overclocked 780 Ti? I mean with the EVGA precision utility, overclocking the Nvidia card is done with a mouse click.



Read the article:


R9 290X Quiet Mode versus Uber Mode Testing
We need to make a statement about how we are going to continue to test Radeon R9 290X in reviews. This topic is born out of user feedback questioning what mode we are going to run R9 290X cards in, and why we made that choice in past reviews, and this. This is also a current hot topic among review websites as to what to mode to test the R9 290X in reviews by default given the two performance modes available.

When using the reference Radeon R9 290X video card in all reviews, we will test in "Uber Mode" by default. There are several reasons as to why. Firstly, we can do this because Uber Mode is an official performance mode supported by AMD and built into every retail R9 290X. It is not factory overclocking, rather, it is a performance mode fully supported and warranty by AMD. Second, it is easily obtained by users flipping a switch on the card, that anyone and everyone who owns an R9 290X has the option to enable. Third, we do not want to "artificially" hold back the performance of a video card. If we do so, what are we showing really? We want the potential performance to come through.

We have chosen to test in Uber Mode because that is what is relevant to enthusiasts and gamers who buy this high-end $549 video card. If you are spending that much money on a video card, you don't want to hold it back and not get your money's worth.

If we were to test in "Quiet Mode" by default, all we would be doing is setting our performance review standard by sound rather than the potential performance of the GPU. We are all about testing video cards for gaming performance and the gaming experience delivered. We don't set the bar by sound. If we did that, we'd have to ensure that every video card we reviewed ran at the same decibels while testing its performance. That is just ridiculous. HardOCP will remain giving enthusiasts the performance testing they desire.
 
Kind of like the guys claiming the 780 ti dominates a 290x? ;)

Good article anyway, interesting to see NV catch up closer in 4k with the 780 ti.

Yeah the 290X is great if you don't mind 55% fan speed times two, which is basically insanely loud. Come on. I've used the AMD reference cooler over several generations of cards. Anyone telling me with a straight face that 55%+ is "quiet", well, i'd say they're nuts.

That basically shows me that AMD pushed their chip to unreasonable levels just to barely catch up to NV; they can't do it with reasonable acoustics, though, and once you normalize fan speeds the 290X performance falls off a cliff. I suppose nvidia could get into the noise wars with AMD if they wanted to, but I like their approach of better efficiency / better acoustics versus overclocking the shit out of your chip and making it insanely loud in the process. Fuck that. The performance of the 290X is great if you don't mind those caveats, though. I just wish AMD hadn't made a borderline shit reference design, then the 290X would be a much more desirable card. I don't think it is desirable AS IS, despite the insanely good 4k performance.
 
...but 290/X's are capable of 15-20% overclocks, maybe more with voltage control. This wouldn't be possible at all if they really were "pushed to unreasonable levels".
 
Yeah the 290X is great if you don't mind 55% fan speed times two, which is basically insanely loud. Come on. I've used the AMD reference cooler over several generations of cards. Anyone telling me with a straight face that 55%+ is "quiet", well, i'd say they're nuts.

That basically shows me that AMD pushed their chip to unreasonable levels just to barely catch up to NV; they can't do it with reasonable acoustics, though, and once you normalize fan speeds the 290X performance falls off a cliff. I suppose nvidia could get into the noise wars with AMD if they wanted to, but I like their approach of better efficiency / better acoustics versus overclocking the shit out of your chip and making it insanely loud in the process. Fuck that. The performance of the 290X is great if you don't mind those caveats, though. I just wish AMD hadn't made a borderline shit reference design, then the 290X would be a much more desirable card. I don't think it is desirable AS IS, despite the insanely good 4k performance.

No one said it is fucking quiet, people are saying that it isn't a fucking jet engine like you keep trying to say it is. Every fucking thread where the 290x is mentioned you go on your inane fucking rant about how the 290x either supposedly throttles or bursts ear drums, give it a fucking break already.
 
Back
Top