Tesselation done properly..according to AMD

[Aristotelian]

What it is to you. Thankfully, others may differ.

Thankfully, others who aren't merely trying to push an agenda (such as defending a debating position merely for its own sake) will be looking at benchmarks before they make purchases. If you want to argue against my "what it is", you'll need to do more than attempt to dismiss an argument as a purely subjective claim.

 

[SirPauly]

So, if one differs than you it must be an agenda?

 

[Aristotelian]

No.

 

[SirPauly]

That's good to hear. It would be questionable to not look at benchmarks and for me current GPU limited games and synthetics are very important

 

[Aristotelian]

That's good to hear. It would be questionable to not look at benchmarks and for me current GPU limited games and synthetics are very important

It's not clear to me why synthetics are important when there is no necessary correlation between the synthetics and games.

I'm not sure why you use the word 'subjective' so often when it's clearly not what you mean. I'm referring to another thread where you refer to a German website's benchmarking and call it 'subjective'. If it's all subjective, your posts are meaningless in a shared context. It's as if you're saying "this is my opinion and you have your own and my opinion shouldn't matter to you nor do I want it to but I still want to state my opinion."

Perhaps we can continue the discussion of objective/subjective via PM if you're up for it, as it's off topic.

Like you (I checked your post history) I think more is better. I think that one needs to be able to perceive and experience that 'more' for it to be better. I don't think that experience is subjective (like some sort of epiphenomenal qualia). Maybe we're just going to have to agree to disagree, but I find that route rather disappointing.

 

[SirPauly]

Imho,

Synthetics help to gauge specific features and future forward abilities at times. A lot of gamers don't pay attention to them but for me, they're very important for future forward gauging considering future content hasn't been offered yet.

It's like anything else, when someone buys a video card -- how long do they intend to own it for? The less time, one may not care as much for future forward features because by the time there is content, they may have another more powerful GPU. But, not all gamers buy at the same time, own them for the same time, not made from the same cloth. Vast, so many different mind-sets.

For me, I intend to own my platform for 18-24 months and purchased my platform 3-4 months ago. I'm not going to upgrade 'till probably early to mid 2012.

 

[3DVagabond]

The problem with synthetic benchmarks is it's too easy for the manufacturers to optimize for them.

 

[busydude]

The problem with synthetic benchmarks is it's too easy for the manufacturers to optimize for them.

That is true... just look at [H] review of GTX 570.. where 6870 was faster than 470.

 

[taltamir]

The problem with synthetic benchmarks is it's too easy for the manufacturers to optimize for them.

its not just synthetic benchmarks... for example, both nvidia and AMD/ATI were caught cheating in popular games that are used heavily to determine which card is better... that is, the driver will detect you are running crysis, and you set it to 8x AA and they will run it 4x AA... the idea is that most people won't notice, but it will show a large performance boost over the competitor. Note that this specific example (8 vs 4x AA in crysis) is just one I made up now to illustrate the point. I remember crysis actually had something to do with water rendering cheat by nvidia, and I remember an AA cheat by ATI (don't remember which game) and a few others... its been a while since the last cheat was caught.

bottom line, is that they detect both canned benchmarks and games and optimize and sometimes even cheat on them.

 

[SirPauly]

That is why one has to look at all the data -- not just synthetics, not just canned benches, not just real world testing -- all of them. For me, GPU limited examples in real world examples carry a big stick or carries more weight, then synthetics, then canned. And appreciate sites that investigate a bit tougher than others.

 

[Rubycon]

The problem with synthetic benchmarks is it's too easy for the manufacturers to optimize for them.

Not to mention the winners of said benchmarks can and often do offer inferior gameplay experience overall.

 

[ronnn]

This is why I enjoy Kyle's subjective rants. He seems to be cynical about both companies equally, so the conclusion is always fun. Besides he does have a point.

 

[Aristotelian]

Not to mention the winners of said benchmarks can and often do offer inferior gameplay experience overall.

Precisely, hence my preference for gaming benchmarks. Nice seeing you in VC&G Rubycon - typically only see you in Cpus and Overclocking.

 

[SirPauly]

imho,

Also like Kyle's method but its based on a very small sampling and a bit too subjective but investigations may bring a different set of variables than canned benches, and very, very welcomed. However, at least the canned benches leave out subjectivity but easier to optimize. To ignore or dismiss synthetics is foolish to me; as it is to ignore real world examples. Actually it is foolish to ignore anything. When one picks and chooses data is closer to cherry picking than an over-all view.