Originally Posted by Abundant Cores
Thats not what is shown in their slides...
The GTX 780 was running at 1006Mhz while the 7970 was running at 1100Mhz.. (with another @ 1224Mhz OC 780 which we will ignore)
So, yes the difference is 100Mhz to the 7970 (About 5% in real terms performance)
Metro LL: the GTX 780 was the same FPS as the 7970 (44.9 vs 45.1) because they had advanced PhysX on, so who knows with that....
Tomb Raider: the GTX 780 was 10% faster (42.9 vs 47.2) ok good....
Crysis 3: the GTX 780 was 22% faster (or is it, they don't say which in their graph is the one they overclocked and the slower 780 of the two is slower than the 7970) so *** [H]????????
FarCry 3: the GTX 780 was 10% faster (but they used Ultra Geomatry on the 780 and only very high on the 7970) so a somewhat useless comparison.....
Hitman: GTX 780 was 12% faster (oh, look at that, one more apples for apples comparison that makes sense) congratulations [H] tho 2 out of 5 needs improvement
thats all the games they tested.
So, a 10% overclocked GTX 780 was 10 to 22% faster than a 7970 GE with a 5% overclock, TPU's test (which i much prefer because they use the same setting throughout their testing and have a much larger game catalog) actually makes the GTX 780 look better in comparison than the [H] review.
[H] are the last people you should use for review comparisons because A, they have a very small Game catalog and B, mixup game settings so the review is not apples for apples. (in this case to Nvidia's detriment) [H] are completely useless for gouging comparisons, they aren't even consistent when it come to game settings in the same review....
TPU are consistent and they don't go for a game catalog favoring one GPU or the other, they don't overclock one this much and the other only this much to confuse the whole thing... instead they mixup a very even and extensive list and then just go to town with reference GPU's at stock unless its a branded test....
But all of that aside if those bottom slides are truly Apples for Apples then 18 to 33% does not bode well for the GTX 780 as that 33% is Metro LL which is extremely Nvidia bias. even My 7870 is throttled to 50% in some parts of that game with advanced PhysX off
Sorry but that is exactly what is shown in the bottom slides.
You seriously are going to question if the bottom slides are truly apples to apples? If you compare the bottom slides from [H] with those same games at TPU, the difference in fps between the two cards is just about the same as the reference vs reference models that TPU used. There is zero to debate here.
Yes, [H] does their reviews with a different format then others. Just because you prefer another site doesn't make their numbers any less meaningless no matter how much you want it to. Their reviews are actually very simple to understand.
Their first set of slides show what settings each game can be set at to get a playable frame rate with each individual card.
They do this so if you own one of those particular cards, they've done the legwork. That's why each card has different settings in the slides.
The second set of slides, the true apples to apples comparison, takes each card and compares them using the exact same settings.
For the record, here are the real figures on the apples to apples comparison instead of those completely wrong numbers you listed using different game settings to compare.
Metro LL 26.5% faster
Tomb Raider 31.0% faster
Crysis 3 19.9%
Far Cry 3 34%
Isn't it kind of odd that 3/5 games were part of AMD's gaming evolved program but you didn't mention that yet you singled out Metro LL because it was Nvidia title. Also odd that you didn't mention TPU includes Grid which uses global illumination that is not coded to be Nvidia friendly.
As for their game catalog, yeah it is smaller that TPU. But it has 5 of the biggest game released in the last 12 months and they are all picked because they are all GPU intensive. TPU's library includes a number of games that get over 100 fps and some even over 200 fps on a 780. What's the point of that other than artificially inflating each cards fps average?
Why do some of you AMD guys like to twist and manipulate numbers like this? Just like I caught sdlvx using 1 year old numbers on a home made graph in the other thread so he could falsely call Nvidia's memory controller broken, you are here trying to pass off numbers about 7970 vs 780 using games with different settings. I get it that you don't like Nvidia as you have stated but the OCN community deserves to have the truth not a manipulation based on your agenda.