[ht4u.net] 55 DirectX 11 graphics cards in the test

[MrK6]

Briefly looking throw the graphs, why is the 7970 GE only at 1000MHz? Shouldn't it be 1050MHz? Anywho, looking at the graphs is rather a let down, as it shows how much graphics have stagnated. People who scored a good-clocking 7970 at release are approaching two years with no tangible upgrades.

 

[3DVagabond]

Erm, of course the faster cards will cost more.. :whiste:

Product stack vs product stack is perfectly valid.

So, because the 7970ge and the Titan are the 2 fastest cards from the two companies it makes comparing them logical? Even though the Titan is more than 2x the money. :\

 

[MrK6]

So, because the 7970ge and the Titan are the 2 fastest cards from the two companies it makes comparing them logical? Even though the Titan is more than 2x the money. :\

I think anything in the market is worth comparing, which is why we take price/performance into account. I think the results show what poor values the GTX 780 and certainly the Titan are compared to the GTX 770 and HD 7970 GE.

 

[omeds]

Performance and value are two different metrics. You decide which is more imortant to you, or combination of.

His comment was correct, Nvidia is faster in said game. Conversely, AMD offer better value in said game.

 

[Final8ty]

Performance and value are two different metrics. You decide which is more imortant to you, or combination of.

His comment was correct, Nvidia is faster in said game. Conversely, AMD offer better value in said game.

Yes they are but they are not mutually exclusive and the majority of reviews and consumers rightly so dont treat then as such or the review would only use the fastest card from each brand otherwise.
So in a given price bracket someone has every right to say that one or the other is faster, you then compare the over all from each price bracket to label which brand is over all fastest in said game, otherwise its not which brand in fastest over all and only which card is fastest in said game.
You could easily end up with NV having the fastest top end card but being slower everywhere else down the range, so you cant generalise NV being faster in a game when it only applies to 1 or 2 cards, it will be the NV Titan, the NV 780 and not NV being faster.

 

[3DVagabond]

Performance and value are two different metrics. You decide which is more imortant to you, or combination of.

His comment was correct, Nvidia is faster in said game. Conversely, AMD offer better value in said game.

The concept is flawed. We weren't comparing companies, we were comparing cards. Not all nVidia cards are faster and, conversely, not all AMD cards offer better value.

 

[Carfax83]

Briefly looking throw the graphs, why is the 7970 GE only at 1000MHz? Shouldn't it be 1050MHz? Anywho, looking at the graphs is rather a let down, as it shows how much graphics have stagnated. People who scored a good-clocking 7970 at release are approaching two years with no tangible upgrades.

They apparently disabled the boost feature for both the AMD and NVidia cards....which I think is a big mistake and very misleading.

 

[omeds]

The concept is flawed. We weren't comparing companies, we were comparing cards. Not all nVidia cards are faster and, conversely, not all AMD cards offer better value.

No it's not flawed. If you get the fastest GPU each company offers, Nvidia is faster in said game.

Of course the faster card will be priced higher, moving it to another price bracket.

 

[3DVagabond]

No it's not flawed. If you get the fastest GPU each company offers, Nvidia is faster in said game.

Of course the faster card will be priced higher, moving it to another price bracket.

Maybe if you repeat it one more time?

 

[SirPauly]

Performance and value are two different metrics. You decide which is more important to you, or combination of.

Indeed!

For me, place more emphasis on performance, gaming experience value; others may be performance over-clocking value; others may desire the absolute best performer. Differing choices for differing needs!

The key with the performance index investigation was based on strictly performance metrics, which was the point of it.

 

[railven]

Briefly looking throw the graphs, why is the 7970 GE only at 1000MHz? Shouldn't it be 1050MHz? Anywho, looking at the graphs is rather a let down, as it shows how much graphics have stagnated. People who scored a good-clocking 7970 at release are approaching two years with no tangible upgrades.

With the OC capabilities of a good 7970 chip, it sure has delivered for the long run.

I was hitting 1125/1575 day one on my Sapphy, and when I went for max clocks I was able to hit 1260/1750 @1.225v and ~70C which was just mind blowing for me.

But as you said in your other post, not many games require me to even do 1125/1575. Here is to hoping a new generation of consoles lifts the bar.

 

[omeds]

Maybe if you repeat it one more time?

You will understand?

Indeed!

For me, place more emphasis on performance, gaming experience value; others may be performance over-clocking value; others may desire the absolute best performer. Differing choices for differing needs!

The key with the performance index investigation was based on strictly performance metrics, which was the point of it.

Yeah exactly m8.

 

[3DVagabond]

You will understand?

I understood you the first time. I just don't think it's any kind of a meaningful comparison. Also don't see the point in me stating my position again, just to have you repeat yours, etc...

 

[omeds]

I understood you the first time. I just don't think it's any kind of a meaningful comparison. Also don't see the point in me stating my position again, just to have you repeat yours, etc...

You dont think comparing flagship vs flagship is meaningful? Ok.

 

[3DVagabond]

You dont think comparing flagship vs flagship is meaningful? Ok.

No, I don't. Besides, it's not flagship against flagship. While it could be argued that Titan is nVidia's flagship, I'd support that position, the 7970ge is not AMD's flagship, the 7990 is. Being a dual GPU card, I don't think a straight comparison with it against the Titan can be done without taking that into account, I also feel the same when comparing $1000 cards to $400 cards. Just as dual GPU is a qualification, so is price. If we are going to just compare who is faster period, without qualifications, then in most charts AMD is faster because the 7990 is generally at the top of the list.

 

[_Rick_]

They apparently disabled the boost feature for both the AMD and NVidia cards....which I think is a big mistake and very misleading.

Well, they did it, because they observed that leaving boost enabled was misleading and hence a big mistake. I agree, for the most part.

The problem with boost is, that for benchmarking it will always give better results, than for actual gaming, because benchmarks usually only run a few minutes at best. A card takes a certain time to get temperature soaked, and runs faster, until that point is reached.

Benchmarking with the goal of obtaining results that have real world relevance requires that boosts are disabled, or each benchmark is run 10 minutes, before a measurement is made. The latter is too time-consuming for most people to do.

Actually, disabling boost should help the better card, as boost mostly boosts average FPS by increasing FPS when they are already high, but has trouble coping, when FPS are dropping during more complex scenes.

 

[omeds]

No, I don't. Besides, it's not flagship against flagship. While it could be argued that Titan is nVidia's flagship, I'd support that position, the 7970ge is not AMD's flagship, the 7990 is. Being a dual GPU card, I don't think a straight comparison with it against the Titan can be done without taking that into account, I also feel the same when comparing $1000 cards to $400 cards. Just as dual GPU is a qualification, so is price. If we are going to just compare who is faster period, without qualifications, then in most charts AMD is faster because the 7990 is generally at the top of the list.

What a load of bollocks. Tahiti vs GK110 is fastest GPU vs fastest GPU. 690 vs 7990 is dual GPU vs dual GPU.

Fact is, you will try and twist it any way you can to get the outcome you desire.

 

[3DVagabond]

Performance and value are two different metrics. You decide which is more imortant to you, or combination of.

His comment was correct, Nvidia is faster in said game. Conversely, AMD offer better value in said game.

Notice there was no mention of GPU in your original post to me? Just nVidia is faster.

What a load of bollocks. Tahiti vs GK110 is fastest GPU vs fastest GPU. 690 vs 7990 is dual GPU vs dual GPU.

Fact is, you will try and twist it any way you can to get the outcome you desire.

You're just qualifying it to meet your desired outcome. Just nVidia vs. AMD doesn't work so you modified your argument to single GPU. Either there are qualifications, could be price, could be single GPU, or there aren't any and it's nVidia vs. AMD. You can't have it both ways.

 

[sontin]

Well, they did it, because they observed that leaving boost enabled was misleading and hence a big mistake. I agree, for the most part.

The problem with boost is, that for benchmarking it will always give better results, than for actual gaming, because benchmarks usually only run a few minutes at best. A card takes a certain time to get temperature soaked, and runs faster, until that point is reached.

Benchmarking with the goal of obtaining results that have real world relevance requires that boosts are disabled, or each benchmark is run 10 minutes, before a measurement is made. The latter is too time-consuming for most people to do.

Actually, disabling boost should help the better card, as boost mostly boosts average FPS by increasing FPS when they are already high, but has trouble coping, when FPS are dropping during more complex scenes.

You can adjust the temperature target with Boost 2.0. So ignoring this feature and forcing a boost step which i guess no buyer will ever use is not showing the true performance of the cards.

The problem with this method is that you will get much better numbers with a custom cooler. These cards are running always under 80°C and will use the max clock all the time...

 

[omeds]

Notice there was no mention of GPU in your original post to me? Just nVidia is faster.



You're just qualifying it to meet your desired outcome. Just nVidia vs. AMD doesn't work so you modified your argument to single GPU. Either there are qualifications, could be price, could be single GPU, or there aren't any and it's nVidia vs. AMD. You can't have it both ways.

No points for 2nd.

Don't try and twist it. Again you are looking for ways to get to your desired outcome.

I'm not going to continue this with you, Nvidia have the fastest GPU in said game. AMD offer the best value in said game. That's it.

 

[3DVagabond]

No points for 2nd.

Don't try and twist it. Again you are looking for ways to get to your desired outcome.

I'm not going to continue this with you, Nvidia have the fastest GPU in said game. AMD offer the best value in said game. That's it.

So you are just going to ignore what you originally posted to me? And I'm the one who's twisting things?

 

[omeds]

Yes, you are the one twisting things to fit your desired outcome.

Ignore what? Nvidia offer the fastest GPU for said game. That's it. I'm sorry if it doesn't agree with your preconceived notions. Conversely, AMD offer the best value, by far.

 

[3DVagabond]

My desired outcome is what, exactly?

 

[The Alias]

^^ oh gosh quit your bickering . The fast is : Amd is better than nvidia in each price bracket they compete in, but Nvidia offers gpus in price brackets amd doesn't . End . Of . Story . In the end I think the statement should be "Amd is usually faster in this game compared to Nvidia ."

 

[_Rick_]

The problem with this method is that you will get much better numbers with a custom cooler. These cards are running always under 80°C and will use the max clock all the time...

Which is why for every card they test, they determine "average" boost. If chip makers chose to TDP/temp hard-limit their cards, and deliver sub-par cooling solutions, well that's their fault, isn't it?

But just using boost mode, is essentially allowing the manufacturers to cheat on the benchmarks, and whoever is better at cheating, wins.

Additionally, testing "out of box" expected performance is HT4U's intended goal, when doing these large cross-market tests. Of course you can then set higher TDP limits and temperature limits, increase cooling, and spend another 20% over the card's value, to get 5-8% higher average clocks, but then you'll probably be taking a look at the detailed test of each card, so you can do the math, whether that's worth it for you or not.