GTX 660 Ti Review Comment Thread

[eyefinity]

Originally Posted by SKYMTL

Please show me proof that NVIDIA paid millions to EA. And how would anyone know who has closer contacts? Without validation on BOTH sets of cards, gaming on either AMD or NVIDIA just wouldn't work.

Please tell me you're joking. Everybody knows that Nvidia bribed Crytek.

As for close contact with Bethesda, they are on record as saying that themselves. I find it hard to believe that you don't know both of these.

You're dredging up an article that's almost two years old? Well then....at that time I was also benching AvP, a game that very much favored AMD: NVIDIA GeForce GTX 580 Review

Are you seriously trying to deflect the Lost Planet 2 point with that? There are no comparisons to be made with the joke that is LP2 and the completely valid (first dx11 btw) benchmark that is AvP.

Again, you proved my point. While NVIDIA had very little / no hand in the optimizations in the game, it was NVIDIA that brought about the DX11 patch that brought advanced graphics to CryEngine. It wasn't AMD. And they did that without cutting AMD off at the knees on the performance front with proprietary shader models.

Yes, all those massively overtessellated jersey barriers and hidden underwater tesselletion too. You can see they put a lot of effort into what they paid Crytek for.

How could anyone make Showdown look good on NVIDIA?

You do realise that plenty of sites benchmark it without global illumination ( i believe this is the major whine point from Nvidia?) and find that the Radeon's are still performing better on average? But of course, seeing as you've already determined that it's completely unfair and broken (and would replace an Nvidia winner in Dirt 3), you don't want to do that do you?

Maybe you don't understand what goes into selecting a game to add to our charts. First and foremost, it has to be stable...so you can eliminate Sleeping Dogs and Hitman right off the bat. Next up, we need to validate that benchmark sequences are repeatable without undue influence being applied by the CPU. Hence why Anno 2070 never made it. Alan Wake didn't make the cut either since it doesn't stress the GPU all that much to begin with.

I know full well what goes in to *your* game selection process, and where the orders come from.

Now how could I do that? I don't immediately cast aside a title simply because it favors one over the other unless there are obvious optimizations or proprietary features that cut the competition's performance. I literally go through a half dozen games a month in an effort to ensure we have the most comprehensive testing suite around. Few make the cut but that's not my problem.

Yes and while you're at it, you make sure you find which part of games run better on Nvidia, just like your mate Smith at Anandtech.

The battlefield 3 spread of results is up to 35% faster on Nvidia down to 2% faster on AMD. Now I'm quite happy to accept that Nvidia is faster in this game on average, but when I see Anandtech and yourselves being at the top-end of those...well isn't that just a real coincidence?

Any chance you and Smith are benching the exact same part of the game? You know, the part where Nvidia does a lot better than average? Or did you both find two different parts of the game where Nvidia is a lot further ahead?

Want to explain why Toms gets these results?



Sure Nvidia is ahead, something more like what is reasonable - 20% if you take the 670 vs the 7950. Yours and Anandtech's? Almost 40%. Again...wow what a coincidence.

You sir, are delusional. TweakTown has been nothing but critical of NVIDIA in every one of their reviews. They are constantly taking NVIDIA to task over not supporting them. But they are biased in NVIDIA's favor? Are you kidding me!?

LOL you can't be serious. Look at his benchmark suite. LOOK AT IT and tell me he's not massively biased in favour of Nvidia.

What's your point? You completely glossed over the fact that I was ragging on AMD's lack of transparency. NOT ONCE did I say the HD 7950 was a poor purchase.

No it's just that every other Nvidia card is a better one, according to you. The XS guys know it as well.

You're drinking too much of Charlie's KoolAid. Every manufacturer knows better than to threaten or bribe sites.

Charlie is a complete fud, but at least he's honest. I can handle twats, what I can't handle is plain dishonesty. That's why you are getting my ire, and not him.

[Soullessone21]

Originally Posted by SKYMTL

I'm actually going to say the Galaxy GTX 660 Ti GC.

Honestly never tried Galaxy they are not under any of my approved vendors.... very strange

[SKYMTL]

Originally Posted by Soullessone21

Honestly never tried Galaxy they are not under any of my approved vendors.... very strange

Interesting. Try Palit. They may be under that brand as a sub vendor.

[SKYMTL]

Eyefinity: It looks like I will have to try another tack here since you somehow believe in a vast conspiracy against AMD where none exists.

I cannot properly respond to your allegations since you still have not clearly shown how we were biased towards NVIDIA in the GTX 660 Ti review.

The chart below (posted in the review) clearly reflects what we said in the conclusion:



Reading through the conclusion, you should clearly see that we are actually going out of our way to recommend the HD 7950 in a number of situations.

Without direct reference to the GTX 660 Ti review, I will not respond to your posts, nor should the members of this site.

[eyefinity]

Ok let me clear something up.

I'm not saying you have invented the numbers. I fully believe that in your benchmarks, each card performed like the number say.

What I am saying is, it is possible and in fact easy for a reviewer (who isn't using an inbuilt benchmark) to make certain results better than others.

This is your job. You admit to running game after game, month after month. You might not admit it, but you have clear leanings towards Nvidia and anyone can see that.

Now, armed with your mouse and keyboard, and taking benchmarks...do you think it's just possible that you are...lets say subconsciously...making the scene perform a little bit worse on the Radeons? We both know how easy it is to crater fps, or maybe even just make it run a little worse on average.

That, is what I'm saying is happening here and at Anandtech. I think you know the best way to make Nvidia cards look good, and you use it to the full. Even if you are being 100% completely fair in your benchmarking process (keyboard and mouse with fraps), you still have the ability to choose sections of games where a certain card performs better than another. I suggest that some of your results (massive BF3 win for example), is proof of that.

I guarantee you that given one week, I could find the best results for AMD in your benchmark suite of games, and it would give some very "surprising" results as well. How do we know that you haven't done this in favour of Nvidia? How do we know? We don't know, we can only see what appears to be...consistently favourable Nvidia results, and wonder.

[NSdeadly1]

Wow, just wow! I just read through this entire thread and now have a headache lol That said, great review Sky. If i wasn't leaning towards a higher res monitor I'd be running a pair of 66ti's in sli. Performance/price at 1080p is quite good and those Galaxy GC cards look sexy as heck! As a side note, I have to say I usually take any review i read with a grain of salt just for the simple reason that not everyone benches components the same. I trust Sky's reviews though because he never steered any of us wrong back in the day when he was on the NCIX forums and we were bombarding him with questions about power supplies and such. Keep up the good work HWC!

[AkG]

Originally Posted by eyefinity

Ok let me clear something up.

I'm not saying you have invented the numbers. I fully believe that in your benchmarks, each card performed like the number say.

What I am saying is, it is possible and in fact easy for a reviewer (who isn't using an inbuilt benchmark) to make certain results better than others.

This is your job. You admit to running game after game, month after month. You might not admit it, but you have clear leanings towards Nvidia and anyone can see that.

Now, armed with your mouse and keyboard, and taking benchmarks...do you think it's just possible that you are...lets say subconsciously...making the scene perform a little bit worse on the Radeons? We both know how easy it is to crater fps, or maybe even just make it run a little worse on average.

That, is what I'm saying is happening here and at Anandtech. I think you know the best way to make Nvidia cards look good, and you use it to the full. Even if you are being 100% completely fair in your benchmarking process (keyboard and mouse with fraps), you still have the ability to choose sections of games where a certain card performs better than another. I suggest that some of your results (massive BF3 win for example), is proof of that.

I guarantee you that given one week, I could find the best results for AMD in your benchmark suite of games, and it would give some very "surprising" results as well. How do we know that you haven't done this in favour of Nvidia? How do we know? We don't know, we can only see what appears to be...consistently favourable Nvidia results, and wonder.

Now this post is more like we see here at HWC. I honestly wish you had started with this one.

My only question is....what would be the end game of doing what you suggest? Why would Skymtl - who runs AMD in his own rigs I might add- make company A look better than company b? There would be no gain from all that extra work and a lot of potential downside. Maybe as a fellow reviewer I am simply too close, so please, what would 'we' have to gain by skewing results and actively looking for ways to skew the results? Why not do what we do...and test them and let the chips fall where they may?

[SKYMTL]

Ok. Lets stay on this path but I am only going to afford you a small amount of rope. These are my final answers on this bias topic.

Originally Posted by eyefinity

What I am saying is, it is possible and in fact easy for a reviewer (who isn't using an inbuilt benchmark) to make certain results better than others.

It is certainly possible. I won't discount that. Unfortunately, there is absolutely no way to prove wrongdoing or -on the flip side of that coin-complete accuracy.

We do what we can to maximize oversight. We do that by posting our benchmark sequences publicly in the methodology thread. Anyone is free to run those exact same sequences with the same cards and prove our numbers either correct of absolutely false. We even hand out our save game files should members ask for them.

I think even you can admit that's a heck of a lot more transparency than 90% of the other sites out there.

Now, armed with your mouse and keyboard, and taking benchmarks...do you think it's just possible that you are...lets say subconsciously...making the scene perform a little bit worse on the Radeons? We both know how easy it is to crater fps, or maybe even just make it run a little worse on average.

How, by looking at the ground on NVIDIA cards and waving around the mouse erratically when using AMD? Again, the burden is on anyone to prove the numbers correct themselves....which is why we provide all things possible so you or anyone else can.

That, is what I'm saying is happening here and at Anandtech. I think you know the best way to make Nvidia cards look good, and you use it to the full. Even if you are being 100% completely fair in your benchmarking process (keyboard and mouse with fraps), you still have the ability to choose sections of games where a certain card performs better than another. I suggest that some of your results (massive BF3 win for example), is proof of that.

Do you really think I have the time to go through every game and every area of every game with a fine toothed comb to determine which is best for each card? That's literally impossible. There is no vast conspiracy but (and this is important) it has been proven before that BOTH AMD and NVIDIA have optimized their drivers for certain built-in benchmarks....which is why we don't use them.

Past that, I cannot address your concerns since the burden is on you to prove your allegations. So far, I don't see you doing so.

I guarantee you that given one week, I could find the best results for AMD in your benchmark suite of games, and it would give some very "surprising" results as well. How do we know that you haven't done this in favour of Nvidia? How do we know? We don't know, we can only see what appears to be...consistently favourable Nvidia results, and wonder.

In that same vein, there is no way to know that I DIDN'T do exactly what you are accusing me of. Everyone uses different benchmark sequences which is why every one of us has been saying that you have to take a wide variety of results and judge for yourself the purchase to make. If you have done enough research and think the HD 7870 is equal to the GTX 660 Ti, so be it. Our results say otherwise but nonetheless, enjoy your purchase.

[Galcobar]

Originally Posted by SKYMTL

Neither of us are wrong. However, several OC cards have expanded power limits hard wired into their custom BIOSes. There is no way to compensate for that in software as there is no way (without studying the BIOS file that is) to know how much -if any- it has been modified.

We choose to use the reference BIOS since it is the safest way to ensure we are replicating stock conditions.

I understand your point on power limits since they seem to be percentage-based for Nvidia, and the value of certainty in applying the stock BIOS (by the way, with no reference cards did you obtain it or use the reference values to create it?).

Maybe I'm a little thick today -- it is Friday -- but you're saying the base clock cannot be affected by software and Ryan says the base clock can be affected by software.

How can both be correct?

[eyefinity]

You say I can't prove it, that's true. However I can build evidence that when fully compiled would be enough to make people wonder if bias isn't a factor.

Very, very few other sites have the 670 beating the 7950 by 37.5% in BF3 for example.
Very, very few other sites got into such a tizzy over the 7950 boost, basically accusing AMD of deceit.
You might think that if the 7950 boost was such an issue to a person, the same person would have an issue with Nvidia seeding overclocked cards to everybody for reviews? That seems reasonable to me at least.
And for an absolute fact, that same person would be utterly embarrassed to even mention AMD's GCN shenanigans in the same breath as the sort of shit Nvidia has pulled over the years - and still is. Unless they had a heavy bias for Nvidia.

That's me done, btw.