FRONTPAGE NVIDIA GEFORCE GTX 780 Graphics Card Review

They didn't just get yelled at. Unsubstantiated rumor had it that some partners didn't get to put out first-run Titans because of it.

What if they made easy access solder pads where you would just solder a 0ohm resistor (or wire) to link two pads, and that would enable more volts? Like what ASUS did that on their GTX580 Matrix.

What if they just had a bios switch?

That would probably still be 'soft' control in NVIDIA's eyes. It seems to me that they basically want you to have to use a soldering iron, making warranty claims easy to differentiate.

Xeon_KI said:

I pointed out a single review that is inconsistent from a list of other popular review sites.
I'm not cherry picking 1 review or one game bench.

Click to expand...

you pointed out the site that did multiple resolution test.
if anything, that was probably the most comprehensive of them all.
but since they didn't list 7970GE > 680 like the rest, you cherry picked it out as wrong.

yea, that's exactly what you did.

you see, this is what I dislike about people on the forum in general.
even at this point you claim you are not cherry picking results.
You and I, the difference is I accept other sites saying 7970 > 680, their test, their result. Good for them. I personally find them pretty equal. So when techspot says 680 > 7970, I told myself, "hm. consistent, avgs out fine"

but not only you, even some more season posters are quick to call them inaccurate, or 'not updating enough', based on 'absolutely nothing'.. if that is not AMD fanboyism, I don't know what is.

I am saying this with good intention, not trying to be condolscenting or anything. I am trying to point out, you have a flaw in logic, and you might already been subjected to marketing bias, that you should be aware of, hope you can read my post that way.

Nvidia has been doing GREAT things, look at GTX 780 !! amazing piece of work!!!
I think people should begin to recognize that the coming generation of GPU for now, crowns on Nvidia.

Who knows, let's hope AMD catches up and stop playing denial.

MattNo5ss said:

For the GTX780 SC ACX review, I re-benched my GTX 680 using the newest drivers available at the time, 314.22, and I used the data from EarthDog's most recent review for the HD7970. The following are the clocks the GPUs were running at during testing and the drivers used:

EVGA GTX 680 (Reference)
Driver: 314.22
Core: 1084 MHz
VRAM: 1502 MHz
Reference GTX680 cost: ~$450

MSI HD 7970 TFIII OC BE
Driver: 13.4
Core: 1050 MHz
VRAM: 1375 MHz
Price: $400

Click to expand...

I am surprised that the 680 didn't score higher on BF3 and Arkham City.

Darknecron said:

I am surprised that the 680 didn't score higher on BF3 and Arkham City.

Click to expand...

me too.

but results are results. let's avg it to the pool.
or maybe the latest catalyst drivers improved the other side too.

trueblack said:

you pointed out the site that did multiple resolution test.
if anything, that was probably the most comprehensive of them all.
but since they didn't list 7970GE > 680 like the rest, you cherry picked it out as wrong.

yea, that's exactly what you did.

you see, this is what I dislike about people on the forum in general.
even at this point you claim you are not cherry picking results.
You and I, the difference is I accept other sites saying 7970 > 680, their test, their result. Good for them. I personally find them pretty equal. So when techspot says 680 > 7970, I told myself, "hm. consistent, avgs out fine"

but not only you, even some more season posters are quick to call them inaccurate, or 'not updating enough', based on 'absolutely nothing'.. if that is not AMD fanboyism, I don't know what is.

I am saying this with good intention, not trying to be condolscenting or anything. I am trying to point out, you have a flaw in logic, and you might already been subjected to marketing bias, that you should be aware of, hope you can read my post that way.

Nvidia has been doing GREAT things, look at GTX 780 !! amazing piece of work!!!
I think people should begin to recognize that the coming generation of GPU for now, crowns on Nvidia.

Who knows, let's hope AMD catches up and stop playing denial.

Click to expand...

What I actually did was read your post where you state nVidia kills AMD cards in Bioshock and I thought, hmmmmm... that is not what I remember seeing the day before.
So I decided to actually fact check it across multiple reviews and was faced with an anomaly.
I thought to myself, "I wonder why these result are so different than the others."
After a moment or so, remembering the Anand Bench is not often updated due to the amount of datapoints and came to a possible answer.

Also both you bluezero5 started out stating the 680 was the clear gamers choice under the premise it being better at game benches, while the 7970 is better at synthetic benches.
Using multiple benches, from multiple sites I proved that this is false.
Only than did you both change your story and talk about SLI, better 3D support, etc...

Maybe if you would just stick to facts or try not to pass opinion off as facts, people wouldn't call you out and you wont have to change your story.

This conversation bores me at this point.

OK, I just read the whole darn thread cause you decided to call out on me.

I will leave you two to your quarrels, but since you called me out on one thing, I will explain.

My points are simply:

1, "i think" gtx 680 > 7970 in games, that's my opinion, and mine alone. I happen to have both in 1 system. really.
2, I believe nothing on the internet should be trusted, so I made a comment about 'testing it yourself'. I hope you came to your conclusion yourself.
3, The rest was just causal talk. (I brought up SLI/3D cause you asked me what else makes GTX680 a win in gaming! you asked !!! I didn't change any topic!!!! )

Xeon_KI said:

What I actually did was read your post where you state nVidia kills AMD cards in Bioshock and I thought, hmmmmm... that is not what I remember seeing the day before.

Using multiple benches, from multiple sites I proved that this is false.

Click to expand...

hmmmmmm. I strongly disagree here.. that's calling confirmation bias. You see a result that you didn't like, look for cross reference to find something that justifies you.
In the field of science, making such statement will cause you your job, cause that's what we call 'data manipulation'.
Finding something to FIT your theory.


for example, you can also easily find another site that supports gtx680 > 7970 ghz for your example of Bioshock too, here:
http://www.bit-tech.net/hardware/2013/04/23/bioshock-infinite-performance/4
http://www.guru3d.com/articles_pages/bioshock_infinite_graphics_performance_review_benchmark,6.html
Both of these studies seems to be inline.. with the other site you showed me in the beginning.
and all I googled was gtx680 7970 bioshock, and used the first 2 results.
So at this particular point, I am slightly uncertain why you question the first site's credit, calling it 'wrong'....
as clearly some people tested otherwise, and it is more than 1 for sure...


So which site should we trust? NONE. trust only yourself. (that's ALL I AM SAYING!! )
I am not here to make any decisions at all, I don't even play games that much, most these titles I never even heard of.

I am also not agreeing with the other parties.. I am just saying, what you said right there, (which I believe is referring to my statement about the internet.) is a taboo in science.
I find it weird why Bobnova and Earthdog decides to 'thank you' for saying that... not very... eloquent of them.
they of all people.. should know better about trusting some sites and not trusting some.. weird.

I find this not constructive to the forum, and will make this my last comment on this thread.
If you have further inquires directed to me. Just PM me, let's not bother others.

bluezero5 said:

OK, I just read the whole darn thread cause you decided to call out on me.

I will leave you two to your quarrels, but since you called me out on one thing, I will explain.
My points are simply:
1, "i think" gtx 680 > 7970 in games, that's my opinion, and mine alone. I happen to have both in 1 system. really.
2, I believe nothing on the internet should be trusted, so I made a comment about 'testing it yourself'. I hope you came to your conclusion yourself.
3, The rest was just causal talk.



hmmmmmm. I strongly disagree here.. that's calling confirmation bias. You see a result that you didn't like, look for cross reference to find something that justifies you.
In the field of science, making such statement will cause you your job, cause that's what we call 'data manipulation'.
Finding something to FIT your theory.

I am not agreeing with the other parties.. I am just saying, what you said right there, is a taboo in science.
I find it weird why Bobnova and Earthdog decides to 'thank you' for saying that... not very... eloquent of them.
they of all people.. should know better about trusting some sites and not trusting some.. weird.

I find this not constructive to the forum, and will make this my last comment on this thread.
If you have further inquires directed to me. Just PM me, let's not bother others.

Click to expand...

I understand the confirmation bias part, but I think it is just as important to have multiple data points that can show a counter-argument to a statistic instead of trusting one source again going the opposite way. I don't believe Xeon did anything in a negative manner, I would have done that same to point that I found conflicting numbers on a different site. The problem is the sheer number of variables that can cause slight differences in results (one boost here versus not there, ambient temperature, different motherboard, processor, speeds, drivers, better silicon on one chip than the other, etc). IMO if two are within 1-2% of each then it is within margin of error.

Granted, the majority of this thread could be its own thread (maybe 'benchmarking/review bias/notes or something)

Long in the tooth, the HD 7970 is still doing well for itself, but as you can see, in every metric NVIDIA has displayed here, you get a large increase over the HD 7970 GHz Edition’s performance.

Click to expand...

Indeed, no bias here.

Just can never let nVidia have a win =P

I think the time has passed were you can really tell a winner from benchmarks they are to close to call and much to close and difficult to be consistent with testing. What all this arguing shows me is that the GTX 780 is not worth the cost.

With all this variable performance crap it's impossible to tell who is ahead and by how much and since the reviews have been more difficult, this is a growing problem where people don't trust reviews anymore and to test it your self is a bunch of crap, Who has the money or setup to waste that much money.

bluezero5 said:

for example, you can also easily find another site that supports gtx680 > 7970 ghz for your example of Bioshock too, here:
http://www.bit-tech.net/hardware/2013/04/23/bioshock-infinite-performance/4
http://www.guru3d.com/articles_pages/bioshock_infinite_graphics_performance_review_benchmark,6.html
Both of these studies seems to be inline.. with the other site you showed me in the beginning.
and all I googled was gtx680 7970 bioshock, and used the first 2 results.
So at this particular point, I am slightly uncertain why you question the first site's credit, calling it 'wrong'....
as clearly some people tested otherwise, and it is more than 1 for sure...

Click to expand...

Bit tech review, a month old
Guru review, 2 months old.
This isn't rocket science here(no pun intended), a good researcher uses the best/most up to date info and from as many places as possible to form an opinion.
I'm not going to buy every card and bench them across multiple games on multiple platforms, running multiple OSes.
That is why I love to view and browse these types of sites... I have the curiosity of knowing, but I don't have the time or the means.

I use the best information I have access to, to form my opinions. I'm still waiting to see better data.

Culbrelai said:

Indeed, no bias here.

Just can never let nVidia have a win =P

Click to expand...

In the sentence you just quoted, I said the GTX 780 out-performs the HD 7970 GHz in every metric. If you actually think that's bias, you truly need your head examined.

EDIT - From my conclusion:

What has the GTX 780 got to offer? Stellar performance? Check. Quiet, effective cooling? Yep, it has that too. It’s even power-efficient for its performance level. Yes folks, NVIDIA has taken the top single GPU spot and now offers a more affordable option with very near the same performance. AMD has their work cut out for them.

Click to expand...

Seriously dude, you need to check your fanboy attitude, it's really tiresome.

wingman99 said:

I think the time has passed were you can really tell a winner from benchmarks they are to close to call and much to close and difficult to be consistent with testing. What all this arguing shows me is that the GTX 780 is not worth the cost.

With all this variable performance crap it's impossible to tell who is ahead and by how much and since the reviews have been more difficult, this is a growing problem where people don't trust reviews anymore and to test it your self is a bunch of crap, Who has the money or setup to waste that much money.

Click to expand...

I dont think its so much who has the money or setup as, if you are interested in buying a card, you have the money and the setup already, no?

The problem there is most, just dont know how to properly benchmark, test, and compare. Its not rocket science by any stretch, but knowing how to properly test and compare isnt something people just 'know'.

EarthDog said:

I dont think its so much who has the money or setup as, if you are interested in buying a card, you have the money and the setup already, no?

The problem there is most, just dont know how to properly benchmark, test, and compare. Its not rocket science by any stretch, but knowing how to properly test and compare isnt something people just 'know'.

Click to expand...

How do you properly test without all the video cards to compare?

There is to many variables to compare you data with someone else, like where you did you did your benchmark in the game and number of players and the same path you take benchmarking.

wingman99 said:

How do you properly test without all the video cards to compare?

There is to many variables to compare you data with someone else like where you did you did your benchmark in the game and number of players and the same path you take benchmarking.

Click to expand...

Most (if not all) games that this site benchmarks with have a canned benchmark built in.

So, yes, they are the same.

wingman99 said:

How do you properly test without all the video cards to compare?

There is to many variables to compare you data with someone else, like where you did you did your benchmark in the game and number of players and the same path you take benchmarking.

Click to expand...

Perhaps take a read about or testing methods as that is covered in that thread...

To summarize, I was on the same page as you as far 'run throughs' and the inevitable variability that it can cause. The only benchmark we have that is not canned is BF3. We tested BF3 and its variability in the section we use (very beginning of SP, the train scene, again SINGLE player not multiplayer as you are correct there) and the % difference was REMARKABLY similar. In that scene, there isn't a lot of 'world' to do different things in, you are inside of a train and the same guys shoot the same way at you for the most part. IIRC, the difference between runs are within a % point when we tested this. It blew my mind to be honest, but the facts are the facts. Unless you deviate completely from the brief run through, like sit looking at a chair for faster FPS or flames/smoke for lower FPS for an inordinate amount of time, its surprisingly accurate.

As far as comparing between the viewers. We all specifically buy setups that match. 3770K, Z77, ram speed, etc. Yes there are likely some minor differences there, however, we do not have the resources to set up a lab and test on a single benching station. It is what it is, however, the differences are A LOT less than you are imagining from both of your points.

Manual, actual gameplay runs are useful since they present real gameplay numbers, but, because of possible inconsistencies mentioned above by wingman, they are not the best for comparing cards, especially cards that are close in performance to one another.

Canned benchmarks are best for straight card-to-card comparisons since everything is consistent, and actual gameplay numbers are good at giving a general idea of what performance one can expect to get while playing a game with a similar setup. In my opinion, they complement each other.

As EarthDog said, our BF3 manual run is the first scene in single player in a train. The train is a straight line, there's only one way to go and you have to kill the enemies in your path. Pretty much as consistent as a manual run can be.