So let's tackle this one thing at a time, shall we?
Quote:
|
I have to say your Eyefinity results for BF3 in HD 7970 in the recent GTX 670 review are really suspicious. The gap between HD 7970 and GTX 680 has increased significantly from your previous reviews.
|
Can you show me how it is suspicious? You quoted a bunch of websites, none of which uses the same benchmark sequence as I do. I see LegitReviews, which uses Rock and a Hard Place, but judging from their lone screenshot, they benchmark at the beginning of the level, which isn't anywhere near as demanding as our section. This isn't anything against their practices but it highlights how you can't randomly pick a review and assume it makes your case for you. We're
completely transparent about the sections and length of our benchmark runs.
As you will see in an upcoming article, in the course of a few driver revisions from NVIDIA, they have significantly improved their performance in BF3. AMD on the other hand has experienced performance LOSSES in some cases.
What you also fail to realize is that we were only of the ONLY sites that reported NVIDIA's kneecapping in Shogun 2 due to an idiotic driver screw up by Creative Assembly. If it wasn't for that patch, NVIDIA's cards would have been EVEN FURTHER ahead of AMD's once it came time to divvy the marks.
Quote:
|
I have spoken to a user in ******* forum who has tested both HD 7970 and GTX 680 in multi monitor. His opinion is very clear . the HD 7970 is faster.
|
I'll put this as diplomatically as possible: so what? Some random user says one thing. Someone else could post another. I'll take my own results over those of a single forum user any day of the week.
Quote:
|
The HD 7970 is faster than the GTX 680 and GTX 670 significantly at Eyefinity resolutions
|
Really? Seriously? Wow....I never knew that...... :
If that doesn't debunk your accusations, let's quote from the article:
....they now have to contend with a card that retails for the same price as a HD 7950 but runs dead even with a HD 7970 in everything except multi monitor resolutions. Quote:
|
FYI operation swordbreaker is the most demanding level in BF3.
|
B.U.L.L.S.H.I.T.
If a site says that, they are simply too damn lazy to play past the first level of the game. They use this "most demanding level" crap as a way to justify their lack of in-game knowledge.
Quote:
Originally Posted by raghu78 You can take a look at that website. The demanding games they chose like Alan Wake, Witcher 2 Enhanced edition, Anno 2070, Metro 2033, Crysis 2 portray the HD 7970 in different light. Other than Metro 2033 the rest of the games released in the last 12 months and are quite popular. |
Wait a sec here....you want me to take a site that tests a $399 GPU at a SINGLE RESOLUTION 1080P seriously? Are you kidding me? Yeah, some of those tests aren't CPU limited at ALL....
Quote:
|
AMD needs to catch up and thats what its doing with GCN and HD 7970.
|
GCN is AMD's answer to Fermi. They are now quite evidently a generation behind.
Quote:
Originally Posted by raghu78 Do you realise the HD 7950 and HD 7970 can compete extremely well with GTX 670 and GTX 680 when all the cards are overclocked to their best  |
Really? I have seen a few GTX 680 cards that can hit 1.5GHz Boost clock with a minor voltage increase. I can't remember the last time I was able to get a HD 7970 + some extra voltage to hit 1.2GHz with any kind of stability. At those speeds, the GeForce cards will simply mop the floor with AMD's best. Essentially, you are comparing apples and oranges like many do when they don't understand how GPU Boost works in relation to PowerTune.
And no, a GPU-Z screenshot or a 3DMark11 run doesn't equate stability.
Quote:
Do you know that there are lots of people who use the GPU for serious GPGPU work or for running GPGPU enabled day to day apps like Adobe Photoshop, video editing software like vReveal etc. Look at Apple's machines. They are very popular because of their video and photo editing software from Apple. Do you want to know how many people take photos and shoot videos and work with them on their PC. I guess the answer to that is very clear from youtube. |
I fail to see your point. A GTX 670 will perform just as well as a HD 7970 in GPGPU apps. The same can be said of a HD 7970 and a GTX 680. The reason for this is due to the fact that these high end GPUs only need a tiny fraction of their processing power for GPGPU apps. It usually falls to how well a given company can optimize their drivers for a certain application.
A good example of this is MediaEspresso. NVIDIA and AMD perform similarly to one another from one GPU to the next (minus entry level stuff, but we're not talking about that here) and yet, here comes Intel with their INTEGRATED GPUs and blows both companies out of the water.
Quote:
Originally Posted by raghu78 |
"Much closer"?
******* shows a 20% win for the GTX 670 at 2560 x 1600
WE show a 30% win for the GTX 670 at 2560 x 1600
10% isn't all that much considering the massive difference from one mission to another.
Quote:
|
And here I see a GTX 670 defeating a HD 7970 in Eyefinity easily. It baffles me.
|
Can you show me a site that tested the GTX 670 and HD 7970 in multi monitor resolutions with the latest drivers from each and used our exact benchmark sequence? No? Didn't think so. As such, you can't make assumption based upon apples versus oranges comparisons. For all we know, AMD's performance may have issues with some elements of the section I test with.
Quote:
|
Is benchmarking truly a science or art ? What is the point of these benchmarks if there can be so much differences between one site and the rest ?
|
As we acquire more knowledge, things do not become more comprehensible, but more mysterious.
- Albert Schweitzer
Honestly man, while I recommend EVERYONE reads MULTIPLE sites, it sounds like you are driving yourself crazy over something us "professionals" can't even answer. From my perspective, we do everything we can to ensure accuracy (4 runthroughs, etc) but even then......
Benchmarking is one hell of an inexact science
-SYKMTL