NVIDIA GeForce GTX 690 Reviews

[onethreehill]

http://www.anandtech.com/show/5805/n...are-ultra-fast
http://www.guru3d.com/article/geforce-gtx-690-review/
http://www.hardwareheaven.com/review...roduction.html
http://www.hardwarecanucks.com/forum...90-review.html
http://hexus.net/tech/reviews/graphi...graphics-card/
http://hothardware.com/Reviews/NVIDI...04-GPU-Review/
http://www.legitreviews.com/article/1921/1/
http://www.pcper.com/reviews/Graphic...pler-Greatness
http://www.techpowerup.com/reviews/N...Force_GTX_690/
http://www.xbitlabs.com/articles/gra...e-gtx-690.html
http://www.hardocp.com/article/2012/...eo_card_review
http://www.bjorn3d.com/articles/NVID...view/2241.html
http://www.computerbase.de/artikel/g...force-gtx-690/
http://www.overclockersclub.com/reviews/nvidia_gtx_690/
http://www.neoseeker.com/Articles/Ha...vidia_GTX_690/
http://www.tomshardware.com/reviews/...mark,3193.html
http://www.hardware.fr/articles/864-...1000-test.html
http://techreport.com/articles.x/22890

[SKYMTL]

http://www.hardwarecanucks.com/forum...90-review.html

Includes PCI-E Scaling, Surround and 3D Vision Surround.

[One_Hertz]

Excellent card. Unfortunately it is nowhere to be found.

[koc]

What a beast

NV-GTX-690-90.jpg

Well done Nvidia

[labs23]

...And priced beastly. XD

Few driver release and this should be able to catch up with the SLI'd 680.

[Chickenfeed]

Although these are halo niche products to the extreme, this makes a HD7990 somewhat irrelavent unless AMD undercut Nvidia by a significant margin and given we won't be seeing 2 full 7970s on board, the only place they have to be competitive is on price like the good old days ( I remember paying only 600 for my 4870x2 ) Even with highly binned dies, best case I'd expect 7950 CF performance and even then likely louder, hotter and more power hungry.

Very impressive engineering achievement

[SKYMTL]

...And priced beastly. XD

Few driver release and this should be able to catch up with the SLI'd 680.

Well......not necessarily since every driver release for the GTX 690 will likely also benefit the GTX 680 SLI solution.

[Motiv]

well that is one hell of a card. My only concern would be future-proofing with the RAM it has.

[One_Hertz]

Wow basically no microstutter:
http://www.computerbase.de/artikel/g...t_mikroruckler

[kaktus1907]

Wow basically no microstutter:
http://www.computerbase.de/artikel/g...t_mikroruckler

The results of the GeForce GTX 690 have surprised us. Because the frame is distributed from 42 to 29 milliseconds which is again clearly better than in the normal SLI system. Even at 30 frames per second, we can see only a minimal difference to the single GPU configuration, which are so completely playable! In other titles, repeated this result, it is therefore justified not only with the game.

[SKYMTL]

I wonder which program they use to measure frame by frame variations. I'm quite sure FRAPS' polling rate is 250 milliseconds.

[iboomalot]

Excellent card. Unfortunately it is nowhere to be found.

the 680s are sold out too and it was release 6 weeks ago.

[PatRaceTin]

Good

i will save my buck until GTX685

[One_Hertz]

I wonder which program they use to measure frame by frame variations. I'm quite sure FRAPS' polling rate is 250 milliseconds.

It's definitely not 250ms... FRAPS frame time graphs show microstutter very clearly (at least on ATI setups).

[SubZero.it]

Wow basically no microstutter:
http://www.computerbase.de/artikel/g...t_mikroruckler

That's wonderful!

[LordEC911]

Although these are halo niche products to the extreme, this makes a HD7990 somewhat irrelavent unless AMD undercut Nvidia by a significant margin and given we won't be seeing 2 full 7970s on board, the only place they have to be competitive is on price like the good old days ( I remember paying only 600 for my 4870x2 ) Even with highly binned dies, best case I'd expect 7950 CF performance and even then likely louder, hotter and more power hungry.

I wouldn't go around betting on your prediction...

[cx-ray]

Ryan Shrout with Tom Petersen from Nvidia Live Event.
http://pcper.com/live

[Chickenfeed]

I wouldn't go around betting on your prediction...

Why the hell not? Its simple deductive reasoning. 2 680s out perform 2 7970s more than not while using less power. Even if they managed to get 2 7970s at full clocks on a multi gpu board it wouldn't have an undisputed performance lead. Given the power draw thats not going to happen, not unless they decide to out some rediculous thermo nuclear 3 8pin monstrosity... AMD simply don't have the wiggle room on the power envelope that Nvidia has with GK104. Is it so crazy to stipulate that if AMD do release a dual Tahiti board that it will A) under perform the GTX 690 and B) cost less due to A?

Whats so unreasonable in that train of thought?

[boxleitnerb]

Please tell me something:

Why is it that not one review outside Germany has tested the frame metering of the 690? It was mentioned as a feature by Nvidia and yet not one word in any of the reviews listed above. The only frametime measurements were from Computerbase, PCGH and Hardwareluxx - all German tech sites. Same with SGSSAA and TrSSAA/AAA (okay, HardOCP does the latter, but they are the only ones afaik). It seems that most reviewers are content to test generic stuff only. Some test 3DVision and multimonitoring, but beyond that - nothing.

The state of the review community worries me. I get that it takes much time and work to write a proper review, but why not test one game less and instead write a page about how the rendered frames are presented. Has it not reached the people reviewing CF/SLI that fps alone are a rather worthless metric with AFR? Seriously, this pisses me off. At least HardOCP makes an effort and conveys their experiences in words (measurements would be better, of course). The rest? Nada, zilch, nothing.

[AKM]

Please tell me something:

Why is it that not one review outside Germany has tested the frame metering of the 690? It was mentioned as a feature by Nvidia and yet not one word in any of the reviews listed above. The only frametime measurements were from Computerbase, PCGH and Hardwareluxx - all German tech sites. Same with SGSSAA and TrSSAA/AAA (okay, HardOCP does the latter, but they are the only ones afaik). It seems that most reviewers are content to test generic stuff only. Some test 3DVision and multimonitoring, but beyond that - nothing.

The state of the review community worries me. I get that it takes much time and work to write a proper review, but why not test one game less and instead write a page about how the rendered frames are presented. Has it not reached the people reviewing CF/SLI that fps alone are a rather worthless metric with AFR? Seriously, this pisses me off. At least HardOCP makes an effort and conveys their experiences in words (measurements would be better, of course). The rest? Nada, zilch, nothing.

Don't count out techreport and their new testing methodology, their 690 review isn't out yet.

[FlanK3r]

Review from OBR
nvidia GTX690 - dual monstrum

nvidia is not my favourite brand, but very well nVidia, good card!

[SKYMTL]

Why is it that not one review outside Germany has tested the frame metering of the 690? It was mentioned as a feature by Nvidia and yet not one word in any of the reviews listed above. The only frametime measurements were from Computerbase, PCGH and Hardwareluxx - all German tech sites. Same with SGSSAA and TrSSAA/AAA (okay, HardOCP does the latter, but they are the only ones afaik). It seems that most reviewers are content to test generic stuff only. Some test 3DVision and multimonitoring, but beyond that - nothing.

Several reviewers and I discussed this at an event recently. For the life of us, no one in the group could replicate the German sites' results on a consistent basis. We've all tried. If I can't replicate results from one run to another, you bet I won't include the results since they may be completely wrong.

[boxleitnerb]

Several reviewers and I discussed this at an event recently. For the life of us, no one in the group could replicate the German sites' results on a consistent basis. We've all tried. If I can't replicate results from one run to another, you bet I won't include the results since they may be completely wrong.

You mean the microstuttering? So if you test microstuttering, you don't get consistent results if you look at one particular scene with particular movement (savegame, no ingame benchmarks)? Why do your results have to be similar to other reviews? Afaik the frame metering doesn't work equally well everywhere. Forward movement != sideways movement, different engines and scenes will exhibit different characteristics.

Even if you don't want to present your results, at least mention the problems AFR is prone to and share your gameplay experience in terms of smoothness. A qualitative statement is better than nothing. If people look at the numbers of SLI/CF, they think "wow, 50fps in 5760x1080!", spend $1000 bucks and are disappointed because they get jitters, sometimes right up to a point where the game is unplayable.

[SKYMTL]

I'm not saying that they have to be consistent with other reviews. What I am saying is that the results are not repeatable. Sometimes microstutter is clearly there. Other times, in the same scene, frametime monitoring doesn't seem to pick it up and when it does, it is never, ever as clearly defined as in the review linked above.

[Rollo]

Excellent card. Unfortunately it is nowhere to be found.

There's an owners thread at Overclocker's.net I ran across this afternoon. Apparently EVGA.com was the only place selling them today stand alone today. Also saw "general availability" on 5/7. FWIW.