[H] HD 7970 Dual-X Review

[Lepton87]

To be fair I don't think it was unjustifiable cost cutting for a premium product. They probably designed that PCB when they still thought it would be a 300$ card.

 

[Zargon]

so its OK to let the 680 OC itself for reviews, but you cant OC the 7970?


no the dynamic OC is not that different then checking some boxes in CCC and moving a slider.

I was annoyed they didnt turn off the DOC for baselines in the GTC reviews.

 

[Nintendesert]

so its OK to let the 680 OC itself for reviews, but you cant OC the 7970?


no the dynamic OC is not that different then checking some boxes in CCC and moving a slider.

I was annoyed they didnt turn off the DOC for baselines in the GTC reviews.

This makes no sense. Especially years after Intel released CPUs with Turbo Boost and AMD has done the same.

You compare out of the box experience vs. out of the box experience. That's it. No settings tweaks, no filtering changes, no tessellation changes etc.

 

[SirPauly]

Based on HardOCP's findings an over-clock of 1260 doesn't smash or dominate a GTX 680 OC and the AIB's haven't offered their GTX 680 differentiators yet.

http://www.hardocp.com/article/2012/04/04/nvidia_kepler_geforce_gtx_680_overclocking_review/5

Even the area that many thought AMD would control and may even dominate; it's not there.

 

[Zargon]

This makes no sense. Especially years after Intel released CPUs with Turbo Boost and AMD has done the same.

You compare out of the box experience vs. out of the box experience. That's it. No settings tweaks, no filtering changes, no tessellation changes etc.

do you understand what a baseline is? and why you would want some benches ran on stock clocks with no boost?

 

[SirPauly]

Boost is stock.

 

[Ferzerp]

Based on HardOCP's findings an over-clock of 1260 doesn't smash or dominate a GTX 680 OC and the AIB's haven't offered their GTX 680 differentiators yet.

http://www.hardocp.com/article/2012/04/04/nvidia_kepler_geforce_gtx_680_overclocking_review/5

Even the area that many thought AMD would control and may even dominate; it's not there.

Pretty much. An OC'ed 680 and a highly OC'ed 7970 are pretty similar in performance. There is no "smashing" going on anywhere. The 680's have much less headroom (at least on the reference design), but start out quite a bit faster. They seem pretty consistently (anectdotally) able to reach low to mid 1200's of Mhz. From what I've read, 7970's seem to have a bit wider range and from high 1100's to very low 1300's. You need one on higher end of OC capability to match the performance of what most of the 680's seem to be getting (at least from reviews I've read, and people's reports on here of maximum clock).

 

[Zargon]

Boost is stock.

'stock' but different cards boost to different levels in different aps, and diff cards boost to diff levels in teh same game

making the benches worthless for baseline comparisons

 

[KompuKare]

This makes no sense. Especially years after Intel released CPUs with Turbo Boost and AMD has done the same.

You compare out of the box experience vs. out of the box experience. That's it. No settings tweaks, no filtering changes, no tessellation changes etc.

This old argument again: 680 Boost =! Intel Turbo Boost.

Sure Intel do bin chips but you can guarantee that if you buy say an i5-2400 it will do exactly as it's supposed to as long as you don't run it in 90°C case: 3.1GHz plus Turbo upto 3.4GHz and that's it.

GTX680, on the other hand, uses a far more complex method and goes from 1.006GHz up to anything from 1.058 to 1.2GHz because it takes into account total power draw etc. and not just temperature and core load like Intel's Turbo.

Since not all chips are binned the same, different GTX680s will Turbo differently even if installed in the same system with identical temperatures etc. YMMV just like normal overclocking, but not like Intel Turbo Boost.

Therefore the only level way for reviewers to counter that distortion would be for them to do their reviews with say 10 random 680 samples bought retail. Don't see that being viable though.

 

[SirPauly]

'stock' but different cards boost to different levels in different aps, and diff cards boost to diff levels in teh same game

making the benches worthless for baseline comparisons

Possibly, if one receives a great core or has a quality case with great cooling, water cooled.

It's about receiving more default performance for applications while still staying under the TDP threshold, while still being quiet and cool.

You bench default with default.

You bench over-clock with over-clock.

Why should nVidia disable this? It's a great feature. More default performance -- no extra power requirements, nice acoustics and cool.

 

[Zargon]

Possibly, if one receives a great core or has a quality case with great cooling, water cooled.

It's about receiving more default performance for applications while still staying under the TDP threshold, while still being quiet and cool.

You bench default with default.

You bench over-clock with over-clock.

Why should nVidia disable this? It's a great feature. More default performance -- no extra power requirements, nice acoustics and cool.

I didnt say it shouldnt be a feature, I said it should be turned off for benchmarks to establish a baseline.

boost = overclock no matter how you slice it, and I agree default v default, OC v OC.

OC'd 7970 v boost enabled 680 isnt exaclty apples to apples, but neither is boosted 680 v stock 7970

I have no doubt that the 680 still wins, I am nto looking to make the 7970 a better card, just get reliable benchmarks


and intel turbo boost and nvidia boost are not the same, since the boost isnt 100% guarenteed nor is it the same on every card, like it is with intel cpus

 

[SirPauly]

It is nVidia's baseline.

 

[Ferzerp]

We haven't gotten more stats from people in this thread, but so far, we only have a variance of a whopping 13Mhz from my card and the one other person who replied's base max boost. That's slightly over 1%. I think you're trying to make a problem where none exists.

 

[Keysplayr]

Have you even seen that PCB?
Are you kidding? Got one right here.

Its power delivery?
You can see power delivery? Ok superman.

The whole PCB just screams cost cutting.
In what way? You just saying it doesn't make it true you know.

It has to be heavily modded just to OC it on LN2 by gluing additional power delivery, you can do that with stock 7970.
This is comical, you do understand that you just said under LN2, right? Please keep this as real as you can muster.

It's just barely adequate for air cooling.
adequate for what? People can o/c pretty high from what I've seen, on the stock cooler and it does it with much less watts consumed than it's competitor 7970. So maybe the 7970 NEEDS a heavier duty power system than the 680 does. This does not equate to the 680 having a lower quality or insufficient power system or PCB. You are making ALL of this up out of your imagination. No idea what you have to gain from doing this.

Maybe for you it's fine just as the PCB of GTX590 but other people have higher standards.
I don't know what a GTX590 has to do with a GTX680. Not at all. Strawman.

It's a fact that its PCB is poor it is not up for a discussion.
Only in your own mind. You make things up, you believe them. That's not my problem or anyone elses. Sorry.

What's up for a discussion is if it really matters. For air cooling I don't think it does but it still leaves a bad taste in my mouth considering it's a 500$ card.
Positively moon-eyed to know how it could leave any taste in your mouth at all. You've had no experience with a 680 let alone any bad ones. Unless I'm mistaken and you have one. There isn't any excuse for your claims here. No reviewer has had any problems or complaints about the quality of the 680. And if they did "allude" to anything it was most likely conjecture at best.

Every single one of your claims is completely unfounded and a pure fabrication on your part. I am very disappointed that this tactic be used to try to reduce the perceived value of the GTX680. This GPU is very good all around. So is the 7970, just not "as" good in most ways.

You now have my full attention, sir.

EDITED : Typos, grammar.

 

[SirPauly]

Not really man, I never read any long posts.

I read it and some very good points!

 

[Lepton87]

Every single one of your claims is completely unfounded and a pure fabrication on your part. I am very disappointed that this tactic be used to try to reduce the perceived value of the GTX680. This GPU is very good all around. So is the 7970, just not "as" good in most ways.

You now have my full attention, sir.

EDITED : Typos, grammar.

If you think 3-Phase PWM and locking voltage control is all cool and dandy in a 500$ card then there's nothing to discuss. You can't control GTX680 voltage for a reason, they want to avoid GTX590 fiasco. I can't imagine the amount of backlash nv fanboys would give to AMD's card for something like that, double standards.

UPDATE: apparently GTX680 is 4+1 not 3+1 but it doesn't change the fact that taking away software voltage control in such a card is just wrong.

 

[Zargon]

We haven't gotten more stats from people in this thread, but so far, we only have a variance of a whopping 13Mhz from my card and the one other person who replied's base max boost. That's slightly over 1%. I think you're trying to make a problem where none exists.

It is nVidia's baseline.

you then do not understand what a baseline is

for it to be a baseline it needs to be utterly repeatable. on different cards, in the same game with the same settings. this prevents that. teh boost #'s in the [H] review were all over the place

 

[Keysplayr]

If you think 3-Phase PWM and locking voltage control is all cool and dandy in a 500$ card then there's nothing to discuss. You can't control GTX680 voltage for a reason, they want to avoid GTX590 fiasco.

Again, a total fabrication on your part regardless of how many power phases there are. 3 or 300. The 680 is performing very well at stock and overclocked and while you "can" deny it, that doesn't make it untrue. Anything else you'd like to imagine up for us today?

 

[Lepton87]

Again, a total fabrication on your part regardless of how many power phases there are. 3 or 300. The 680 is performing very well at stock and overclocked and while you "can" deny it, that doesn't make it untrue. Anything else you'd like to imagine up for us today?

So you think it is not at all disappointing that you have to resort to hardware modes to increase voltage? I want more in a high-end card, voltage control is a must for me.

 

[chimaxi83]

So you think it is not at all disappointing that you have to resort to hardware modes to increase voltage? I want more in a high-end card, voltage control is a must for me.

The card is 4 phase for GPU, 1 phase for VRAM.

I agree completely though, to take voltage control away from users on a $500 card is inexcusable.

 

[SirPauly]

you then do not understand what a baseline is

for it to be a baseline it needs to be utterly repeatable. on different cards, in the same game with the same settings. this prevents that. teh boost #'s in the [H] review were all over the place

Innovation moves forward with dynamic clocks and volts.

 

[Jaydip]

So many claims that 680 is not designed for enthusiasts.Do you guys follow a handbook of gpu design or what?X phase vrm's doesn't attribute to higher engineering standards.Do u believe nvidia engineers can't add some extra vrm's for their gpu?I remeber some stupid motherborad design where they have 24phase vrm's for sandy bridge ,it is not engineering it is called stupidity.Engineering makes the technology a commodity so most of the folks can afford them.Remember MarsII from ASUS?Nvidia could have surely designed a card like that but how many people would have bought a 1200$ card?Nvidia and AMD engineers are way more intelligent than their AIB counterparts so using solid chokes,Japanese capacitors,military class components are really marketing gimmick rather than engineering breakthrough.No doubt in the future we can see Armani designed 680 or 7970 which will cost 4999$ and enhance the e-peen by 100% when installed()

 

[Keysplayr]

So you think it is not at all disappointing that you have to resort to hardware modes to increase voltage? I want more in a high-end card, voltage control is a must for me.

What I think is, that I am tired of reading made up fabrications about poor quality PCBs and components.
And, voltage is adjusted automagically on the 680 when you increase the base clock. So, while you don't have "direct" control of the voltage, you can influence it via base clock or GPU offset. Still permits a very respectable o/c for the 680.

 

[notty22]

Every single one of your claims is completely unfounded and a pure fabrication on your part. I am very disappointed that this tactic be used to try to reduce the perceived value of the GTX680. This GPU is very good all around. So is the 7970, just not "as" good in most ways.

You now have my full attention, sir.

EDITED: Typos, grammar.

I agree. This idea that you are getting something sturdier with no evidence, is a strange angle to debate. A faster card, that does it pulling less power, that only needs 2 6 pin connectors, should be viewed as a positive.
I'm pretty sure the gtx 480 with it's header heat pipes and ability to pull 300-400 watts was never thought impressive.
Or are we talking about AMD's card being longer ? I swear in poll's, length isn't important !
I'm impressed with the cards design /gpu design that allows gddr5 memory to be clocked stock at 1500mhz, highest ever.
I swear , I used to read that was important. Even without relevance to overall performance. But of course now, Nvidia is running the memory faster, so it's not.
And the memory o/c's further

It's like feeling victory that you have a higher red line/rpm shift point in a slower car.
It's a bad car analogy, but there it is.
I've already noticed the z77 Intel boards look barer in the voltage area's, and that's because they can be. They pull less power. Period.

 

[SirPauly]

I am pretty sure there may be AIB GTX 680 differentiation with some, real focus on over-clocking head room for the consumers or gamers that may be underwhelmed by the vanilla GTX 680.