Originally Posted by Arizonian
I'm new to bench competition. I have one question.
IF this series AMD 7970 claims they now run tessellation as good as Nvidia, why do we have to run two benchmarks with Tess on & off to enter? Isn't the point of benching to have the highest possible graphic settings offered?
When I game I look to have settings turned up as far as I can go. Wouldn't that pertain to benching?
On a side note, there should be a 'How to overclock the GTX 680
' like the tips provided for the 7970. To be fair, for a noob bencher (like myself) to join in the fray as well.
The Tessellation for the Heaven 3.0 benchmarks is always on, its just normal vs. extreme. Normal is what you are likely to see in a game, Extreme is just for benchmarking. Both are included here because they both have merits.
The reason why AMD users may be called upon to bench with it off is if it appears its been disabled and show potential cheating. You can hide a little tessellation adjustment, but you can't hide disabling it completely, the scores go off the charts. Is this what you meant by on/off?
The good reason to include Normal is to have another non-biased benchmark that shows off the cards in terms of FPS. It also will illustrate the effectiveness of a good CPU overclock. In essence its a GPU benchmark that can be CPU bound. We will see how well we can make the GPUs dance based off the overall system performance.
The second Tess Extreme benchmark is to specifically test the Tessellation with GPU performance of the cards. The benchmark results aren't meant to override each other as one being better, but compliment each other. These benchmarks will display more useful data to discerning eyes.
The least discerning eyes will just see a clear victor, while people like myself will see what kind of bottlenecks appear, which card is better at raw tessellation, which video card is better at a blend of settings etc., we include benchmarks of Eyefinity and 2560x1600 but only a small percentage of gamers actually use that resolution. While it is certainly probable that a card can "win" in benchmarks at those resolutions, we often see some cards before better than the competitors at lower resolutions or higher resolutions than what the focus is on reviews.
Same goes for in game settings, while it may totally destroy a competitor at highest settings, it might lose at slightly lower settings. Things like this show us bottlenecks in architecture, or better yet, issues with drivers (i.e. results that should be higher.) We can mitigate one card dominating all benchmarks by making more benchmarks, and making them dynamic. You could say we are doing the ultimate Overclock review of the cards this way.
I would love to write the GTX 680 overclocking guide but sadly, do not have one :( I will definitely include it if someone provides a good detailed one for noobs.
Edited by RagingCain - 3/30/12 at 8:35am