7970 vs 680 thread.

[ALXAndy]

I cant be bothered to go look at some reviews but most reviews use 4xaa and i have noticed that the gtx680 seems to suffer more when aa is used comparing to the 7970. I would need to look more into it but from the few reviews i looked at the gtx680 definately took a bigger hit going from no aa to 4xaa. Maybe 8xaa takes an even bigger hit when comparing to the 7970. It needs more looking into and i am not saying this is fact.

But it was OK when the GTX 480 took a clear lead over the 5870 when AA was applied?

TBH mate that was the only thing the 480 was better at. Lots of AA.

So if it was important then it's important now.

The GTX 680 is a fantastic card. IMO (leaving price aside) what they have done with a mid ranged part is nothing short of phenomenal. It really is quite amazing. However, cracks are going to appear in time. It is, at the end of the day, a cheaper card to make than the 7970. This was pointed out in a review and it does make sense. The 7970 uses heftier power delivery and components and higher quality ones as it does need them.

I did say it before, but I read a very interesting article about how Southern Islands was, to all intent and purpose, Fermi. It gets too hot and the clocks should be able to go much higher given that it is a high end attempt. AMD are struggling with it in the same way Nvidia struggled with Fermi. No doubt they will revise it though.

 

[ALXAndy]

I think the term clock for clock was misinterpreted.

Surely any comparison should be between:-

Default clock and default clock?

Max OC and Max OC?

Clock for clock was raised because people were saying that not all 7970s could reach X mhz. So as soon as the overclocked benchmarks came in the suggestion that all 7970s are not the same (and so on) started to come about.

So a clock for clock set of benchmarks were posted, as well as max OC vs max OC (OC3D).

Hahahaha, Vram, HAHAHAHAHAHAHA ......

You were lied to and you believe porkies.

Again ... HAHAHAHAHAHAHAHAHA, Vram .... HAHAHAHAHAHAHAHAHAHA!!!!!

(<---- 1 Gb Vram, running Skyrim with the unofficial 2K HD texture pack from Nexus plus max settings, 4x AA and 16x AF, recommended Vram 1.2 - 2 Gb, runs 100% smooth, lag and stutter free while my GTX 560 tis sit on maximum Vram load).

Again ... Vram, HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA, oh HAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHAHA ......

More five year old scribblings.

 

[ALXAndy]

Overclockers, not overclocking since GTX680 release.

Well yes it would seem so.

It's funny really. For so many years we have celebrated the death of Hitler and idealism, but deep down so many people would still have things their way all the time if given the option.

Megalomania reigns supreme it seems.

Either that or we live in a world full of narcissists.

 

[ALXAndy]

i cant understand all the fuss over the 680's clock boost feature,people calling it cheating because it "overclocks" itself.acording to wikipedia overclocking is "the process of making a computer or component operate faster than specified by the manufacturer by modifying system parameters."at no point does the 680 do this it just raises or lowers the clockspeeds whist staying within the manufacturers specs.the whole argument that it is cheating is bogus considering t never OVERclocks anything

I think you have missed the point.

No one is complaining that it dynamically overclocks itself. No one.

What we have brought up is that it dynamically overclocks itself

Which means that compared to a card that doesn't the results are obviously going to be off.

It's not rocket science to understand. Basically people are saying yes, it's perfectly OK and fine to compare a card that overclocks itself to one that doesn't.

And it's when you overclock the 7970 that the tables turn.

 

[ALXAndy]

Nobody here will disagree that the 680 is selling faster than the 7970 did.

That's not what we are talking about Greg. So it's kind of irrelevant.

Apple Iphones sell like hot cakes. It doesn't mean they are better than their Android counterparts.

 

[ALXAndy]

Oh great, now Andy is back ... yay for more entertainment I suppose.



What a hypocrite, oh yes you most definitely are complaining about this and referring to it as cheating constantly.

I've never met any one so awfully poor at discussion as you. Ever.

And I have been using internet forums since 1996.

You make no valid points, and all of the ones you attempt to make are just rubbish.

As for entertainment? well it's kind of one sided and unfair really. You pop up talking a load of rubbish and acting like a five year old and then the game just becomes too easy.

Any credibility you may have had was lost many pages ago.

 

[ALXAndy]

Do You know what guys !

Unless someone actually owns both a 7970 and 680 they should not post as they can only argue over 3rd party data....simple, yes ?

They already have. So has some one who has a 7950. And he said he was quite disturbed that the 7950 was faster in AVP.

 

[ALXAndy]

Somewhere in that thread that hurts my eyes, he says that nvidia are not going to have a lot of buyers or something.

That is my relevancy to that post.

Ah OK fair enough. They'll sell simply based on the early reviews.

I mean you have review sites telling people what great value for money they are.

 

[ALXAndy]

Add Andy as well then please.

At least the majority of reviews state the GTX 680 is faster, unlike Andy's claims that it's not.

I started the thread. And I started it to get the comparisons up.

And, they've gone up. We've also had some good discussion going on too.

If you don't like the thread and want to continue to think that the 680 is faster there are plenty of threads on here ATM that will cater to your needs.

 

[ALXAndy]

He simply isn't interested in answering such questions. Logic doesn't come into it.

I've answered it many times yet you obviously are not paying attention.

Whether an overclock is dynamic or put there by the user it is still an overclock.

It is still pushing the part you are overclocking past its default frequency.

The 680 does that itself, the 7970 doesn't. Thus, comparing them is like comparing apples to cat food.

When they are both overclocked the results are more fair yes? so if they were both overclocked and the 680 was still faster then it would be the faster card overall.

Is that logical enough for you?

 

[ALXAndy]

Its default frequency is variable - so no.

It isn't variable. It's 1006mhz or whatever. It then overclocks to 1050 and beyond when it needs to.

 

[ALXAndy]

A neutral buyer that plays BF3 will obviously see this as yet another good feature to have, especially given the performance of the card.

So why not look at more than just plain fps, which out of the box puts the GTX680 in the lead in the majority of games.

Sure if you mainly play the games that perform better on the 7970, then just get that, but if you're going to be play more why not get the card that'll perform better overall, cost a little less and contain features such as PhysX, TXAA and Adaptive Vsync. All of which will only improve more.

Not to mentio that there are some M-ITX buyers out there that ths GTX680 would be ideal for given it's TDP and use of 2-6pin power connectors.

BF3 doesn't use Physx.

Yet, in another thread he is telling people that buying a 680 is a wise choice because it does.

 

[ALXAndy]

The very definition of overclocking is increasing frequency beyond the manufacturers specification[\b]. Here the manufacturer's specification is for a variable clockspeed. As I said, it isn't exactly complex

This has been done for years in mobile devices, and due to the power constraints of high transistor density GPUs, we're seeing it in full size parts. This trend will only grow over time.

You do realise that Nvidia refer to the technology as dynamic overclocking yes?

http://www.techpowerup.com/reviews/NVIDIA/GeForce_GTX_680/30.html

 

[ALXAndy]

Thread summarised: 680 owners are happy; 7970 owners are happy. Everyone's happy so I don't understand the need to argue about which card is better. I think the fact it's still going on pretty much tells you it's close!

It's not about arguing what card is better. Most will make their choice and feel that they have the better card.

It's about performance and what each card can do.

 

[ALXAndy]

What you mean is: Techpowerup refer to it as "a kind of dynamic overclocking".

You can be as pedantic as you wish. Really.

The bottom line is the card increases its frequency. You can word that however you like (boost, turbo boost, Knight Rider Turbo boost Michael) but it doesn't change it.

The card is overclocking itself, and then reverts back to a base clock. So, it is logically dynamically overclocking itself as it is increasing the base frequency.

All of that aside because it doesn't really matter now - when both cards are manually overclocked beyond their factory settings they are about level.

 

[ALXAndy]

i get it , Andy is ****ed cos he paid **** loads for a 7970 when he could have got a 680 and saved some cash for a better card "out of the box" cos he doesnt overclock

Hi. We had all of that about ten pages ago.

As what I believe to be "an enthusiast" I am capable of not letting personal feelings get in the way and finding myself more interested in trying to decipher the truth, and how these cards perform when pitted equally.

IE - at their maximum levels.

Even *IF* the 7970 came out on top that wouldnt' matter. Nor if the 680 was genuinely faster overall.

It's not a new thing for people to get rather excited when a new product comes out and get it a bit wrong when writing a review. Human beings will have a preference, and the Kepler launch was very exciting.

However, it's a bit off when in your excitement you forget to compare the cards properly and fairly and then jump to a conclusion based on your opinions and not in depth details.

So you can see it however you wish. If you think I am annoyed because I bought a 7970, or need to "justify it" or whatever other scenario then that's up to you mate.

If I was really that bothered I would have put it up for sale and ordered a 680. Maybe if the 680 truly was the better card and it was completely clear I would have done? who knows !

However, before I would have even thought about it I would have made sure I had all the facts. Because it would be an expensive mistake to make if there was no difference in it.

 

[ALXAndy]

I suggest this thread be closed and do another one with 99% less BS, OP should request it.

It should be OK

Hopefully some more information will come to light soon.

Still wondering why Vortez have not reviewed the 680

It's a shame, because they followed up their 7970 review with a series of very interesting articles where they overclocked both the 7970 and the 580.

Oddly there are no posts on the forum to ask why either, so I hope nothing untoward is going on.

ALXandy,it not not dynamically OVERclocking itself thou,the clue is in the name OVER clocking i.e to clock OVER the manufacturers parameters at no point in time does it raise the clocks OVER there own thresholds

I think we have it covered now hibbet. If anything I will simply reword it from overclocking to increasing and or changing frequencies to help frame rates.

There's no point in being pedantic about it now. It is what it is

Whatever you call it, and however you look at it the "function" was purely designed to help with game frame rates.

In the same way that adaptive Vsync was designed and put to use.

And, I reiterate, I think it's very clever. However, I also think that it's unfair to compare a card that does not adjust its clock frequencies.

You never know, AMD may be able to do this on a driver level providing there isn't a clever little doodad bolted onto the 680 that keeps things stable.

I will also mention again that yes, "out of the box" the 680 is indeed the faster card. But, I am not interested in out of the box, I want to see the full capabilities of BOTH cards.

 

[ALXAndy]

Hmm. The plot thickens (well, if you're a suspicious git like me )

I just hope this isn't because Vortez did those articles to put the 580/7970 arguments to bed.

I mean they're not a huge outfit, but I usually find what they have to say incredibly interesting.

quote didn't show but I read it.

Jeff. For many a year "out of the box" performance has been deemed completely and utterly irrelevant mate.

What is important is what a product can do when you overclock it and find its true limits, then test it.

For example ; Opteron 140. It was a server part so should have never been worth anything in desktop terms. It was slow out of the box.

However, slap it in a desktop board and overclock it? It was a complete animal and thus, went down in history.

See also - Celeron 300A. It cost around £80 or so. The top end CPU at the time was the Pentium 2 400. The P2 400 used a 100mhz base clock, the Celeron used a 66mhz one.

All you had to do was set the base frequency on the motherboard with a jumper to 100mhz, and providing luck was on your side you ended up with a part that cost less than a third of the high end part and was equally as fast.

Ever since people could overclock they have. And it has gained significant importance. Large coolers, GPU coolers, even software to aid in it.

So to all of a sudden completely want to dismiss that and compare something unfairy? well, I will leave you to work that out.

 

[ALXAndy]

With regards to the dynamic boosting that the 680 does, is the suggestion that this is overclocking because it doesn't always run at this and so boost itself when required?

Are there not similarities between this and the way that when I go from doing stuff in 2D to doing 3D stuff (i.e. games) the clocks on my GPU increase to handle the extra work?

So are we saying that the 7970 doesn't do this and runs the same clocks when idling at the desktop as when playing BF3? Surely that's bad for power usage?

That's another angle to look at it. But, as with most of the others it's completely unimportant.

What is important is maximum performance.

If the AMD card was adjusting its top side clocks when in games there would be much more of an uproar over it.

But yes. All of these new super wonderful features aside (including zerocore) are not important.

 

[ALXAndy]

TBH,the only worry I had about the technology, is if it could be used to warp benchmark results,ie, boost the GPU extremely high for benchmarks,but in real extended gameplay situations reduce the boost significantly. Both companies have a history of optimising for benchmarks,so you could understand my concerns.

However,I did find this article form one of the biggest French review websites where they disabled the boost function:

http://www.hardware.fr/articles/857-23/performances-gpu-boost-overclocking.html

It seems the boost gives an additional 4% to 5% performance at 1920X1080,so its not huge.

Interesting.

It is interesting to note that without Boost GPU, the GeForce GTX 680 would have been content to match the Radeon HD 7970. Overclocking the GeForce GTX 680, and especially her memory, the greatest benefit to situations in which it was behind.

So basically it reakons that clock for clock there isn't anything in it.