Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
Graphics rendering is about as "inherantly" (sic) subjective as any other field in science.
The weather is inherently subjective (ie. it's warm; it's wet; it's windy), yet we still try and quantify this information and apply the scientific method, because the information is much more useful to us that way.
You bring meaning to timedemos by running them with every single possible setting.I see what you're saying, but there are many many measurements for weather conditions (temperature, humidity, wind chill, precipitation). This information is useful to us because most of us, when going to weather.com, can look and see "76 F, Sunny, Winds 5mph," and come to the subjective conclusion of a beautiful day. This quantified information is well defined in the context of real life, and leads to what's going on in reality quite directly.
When looking at video card reviews, saying "This card averaged 89fps in a timedemo" tells you much less about subjective things like the smoothness of the gameplay, the quality of the rendering of different areas, texture blending and anti-aliasing quality, etc. There are no good ways to quantify image quality or playability, in the same way you can quantify weather conditions. Even between the red team and the green team, 4xAA can mean completely different things, because each card/driver pair implements AA in a totally different way.
Weather is described in terms of these metrics and standards, just as video games are described in terms of the quality of the experience.
Actually, you've got that backwards. Other websites' results would have a chance during a defense whereas [H]'s wouldn't based purely on the fact that the results are only applicable to Kyle's experience, and nobody elses.
You bring meaning to timedemos by running them with every single possible setting.
Card A:
Timedemo 1: 89
Tiemdemo 2: 85
Tiemdemo 3: 35
Tiemdemo 4: 75
...
...
Tiemdemo n: nn
Just because you don't play the timedemos doesn't mean that you cannot test every single possible situation your card would face by using timedemos.
Most sites that benchmark test multiple games, and use in-game timedemos that were recorded by their staff of actual gameplay.
Having reliable information that's quantifiable allows you to compare it to other sources. Then, and ONLY then can you make an informed decision. Otherwise it's just taking someones word for it.
Game over. Every single non scientific argument has been completely and utterly destroyed.
@ SmokeRngs: As I said earlier, if your computer is croaking on AI and choking your video card that way, your review sites clearly did not effectively inform you well enough to build a computer that doesn't suck, and failed completely and utterly at their mission statement of helping you buy good parts.
Game over. Every single non scientific argument has been completely and utterly destroyed.
@Niceone
Who do you guys think design these cards? Scientists and Engineers.
How do you think they quantify the performance of the cards beforehand for games that have yet to be released? How do you think companies can claim 'our new generation performs xx% better than the last!' before it has even had silicon tapped out?
With numbers and benchmarks.
How do I know? I did several years of VLSI(Very Large Scale Integrated Circuit Design).
I sat there with a simulator that showed me my gated clocks. I crunched, by hand, how fast my devices would be on certain hard paths, and then determined its nominal clockspeed.
@ SmokeRngs: As I said earlier, if your computer is croaking on AI and choking your video card that way, your review sites clearly did not effectively inform you well enough to build a computer that doesn't suck, and failed completely and utterly at their mission statement of helping you buy good parts.
Game over. Every single non scientific argument has been completely and utterly destroyed.
You're not seeming to grasp the concept. It doesn't matter what the fucking companies' engineers claim if the results in REAL TESTING aren't there!
We aren't saying that the cards are not benchmarked or produced without science in mind (on the contrary, we all know they are)! We are saying the average gamer doesn't care how they are benchmarked, no matter if you want to call it 'scientific' or not, if it cannot accurately represent a real gaming experience!!!
And that accurate testing is what I'd like to see focused on, not what some engineer has to say about the % increase!
"REAL TESTING" what is that? As far as I'm concerned, there's only one scientific method in this world. Every school on the planet teaches it. Want to know why? Because it can be effectively applied to any, and all types of information that require quantification.
The only argument worth repeating that I have heard so far is that you remove load from your computer under a timedemo. But even that is weak and paltry. You can easily reproduce load on a cpu with any given number of tools.
I've already approached the 'lets not hiccup' argument in my first post, and I fall into that one too. However, [H]s benchmarks do NOT represent this category. They allow the framerate to dip below 30fps frequently, as can be seen by the line graphs. So you can't even use the settings posted here on [H] and your argument falls to pieces once again.
I'm sorry but I can't locate the part where you actually answer my question. Could you please try to say wheres the real science behind any review site? You started with asking who design these cards..and how do they get their performance numbers that they are telling. Well those numbers that manufacturer tells us is rounded up by engineers and marketing department. Usually those numbers are purely theoretical and..suprise suprise..doesn't really apply in real life. Ironic isn't it?@Niceone
Who do you guys think design these cards? Scientists and Engineers.
How do you think they quantify the performance of the cards beforehand for games that have yet to be released? How do you think companies can claim 'our new generation performs xx% better than the last!' before it has even had silicon tapped out?
With numbers and benchmarks.
How do I know? I did several years of VLSI(Very Large Scale Integrated Circuit Design).
I sat there with a simulator that showed me my gated clocks. I crunched, by hand, how fast my devices would be on certain hard paths, and then determined its nominal clockspeed.
Self claiming victory is pretty big indicator that one is either low on valid arguments or just simply arrogant on this one. Either way it's not effective argument. Your style is ironically pretty far from being scientifical.Game over. Every single non scientific argument has been completely and utterly destroyed.
Thing is that [H] is not only one who gets similar differencies between actual game play and time demos so the question right now is why time demos seem to favor AMD.There's only one scientific method in this world?
I don't really know how to respond to this statement, it just baffles me.
Instead, I won't even address that, but go back to the point: The timedemos DO NOT accurately represent playing the game. The results are there. You can go look yourself. I've tested much of this myself, and while obviously my numbers don't exactly match up with Kyle's, it is evident the numbers in the timedemos, or claimed benchmarks for many games, DO NOT correlate to what kind of performance I actually see (this is what 'real world' testing means...you can call it anything, but it's merely playing the actual game!).
You say my argument falls to pieces, however, I don't even know what yours is!
Self claiming victory is pretty big indicator that one is either low on valid arguments or just simply arrogant on this one. Either way it's not effective argument. Your style is ironically pretty far from being scientifical
Attempting to force results to be even is the WRONG way to approach scientifically testing a hypothesis. Proper scientific method is: Keep a control group, have identical test circumstances, and see how the results DIFFER. Not tweak the environment and circumstances until results are almost identical. Any real scientist after actual information will fight to the death over this process - which [H] is very actively tossing out the window.
I couldn't care any less about what [H] thinks the 'best playable settings' are for a game. In fact this is one of the most useless 'points' I have seen a review site attempt to make. Not everyone carries the same weight on what makes something 'playable.'
Forgot to mention that it was just my opinion .beat me to it.
It's funny how he seem to think that it's strong argument to self claim victory
Thing is that [H] is not only one who gets similar differencies between actual game play and time demos so the question right now is why time demos seem to favor AMD.
It's funny how he seem to think that it's strong argument to self claim victory
Forgot to mention that it was just my opinion .
It's just that it seems that he doesn't trust that people will believe his arguments and he has to claim victory to get some comfort.
I believe you are the one that has it backwards. Other websites following [H]'s criteria, method and examples would show the same findings. There would be some margin for error, but that has to be expected in any type of test. It is reproducible.
The scientific community would laugh other sites review processes out of the building due to using faulty testing tools from the get go. When your "tools" don't match up to the results in the "real world" you seem to have a bit of a problem there.
You want to know why other sites hate this type of "review"? It's labor intensive and costs a lot of money. Plus, it doesn't have as many graphs which makes it "look" like it has less content. Most sites don't want to put the time and effort into something if they can get by with something cheaper even if it means their results don't agree with what really happens. They just want to keep running the same useless timedemos over and over because people have been conditioned to think that is what real gameplay is. We aren't in the infant state of 3D accelerators anymore where the differences between cards was only with the framerate. All the different graphical and IQ settings the different cards have make a huge difference. That is what is being tested here, not framerates. 20 graphs of the average FPS of different resolutions and levels of IQ in a timedemo don't tell me a damn thing.
Get over the fact that timedemos don't follow the scientific method. They are flawed as tools and don't show a damn thing about real world gameplay since they don't even take into account things such as AI or some physics. It's a recorded demo, nothing more.
That is one of my critiques of [H]...playability varies between users. Apples to apples is better IMHO.
THIS!
The article is a wake up call for PC sites out there, just relying on canned demos provided by the game just doesn't cut it. If I went to AT and saw the Crysis results there I would think I could game at 1920 resolution with HIGH settings. The fact is you can't. The review is therefore defunct as a review. It does not help one to buy the correct GPU to fit their monitor.
I hope that sites like AT have the balls to fess up that they are wrong and start putting the hard yards in to get complete reviews, even if they included both timedemos AND real time play then we can make up our own minds!
Actually it would be better if there were both included. Problem with AtA is that it does tell difference only in one setting.That is one of my critiques of [H]...playability varies between users. Apples to apples is better IMHO.
You're right.. I mean he was just talking about science and scientific methods and I was pointing out that he doesn't seem to know about them..countering his arguments that was on topic. Perhaps there would be better ways to express this.Let's not start the personal attacks, I don't want this thread closed yet! It can actually be productive!
And we all know it's your opinion, no need to mention it
You're right.. I mean he was just talking about science and scientific methods and I was pointing out that he doesn't seem to know about them..countering his arguments that was on topic. Perhaps there would be better ways to express this.
I don't see why it must be an either or thing.
Theoretical throughputs are just as important in figuring out a card as the current playability in a handful of games chosen by a reviewer.
The biggest problem with [H]'s test is when they are internally invalid.
When after talking about how one setting isn't the same on two different cards the reviewer then proceeds to test in that questionable setting.
Also who evaluates which of the setting maximized are more important?
Most people I know who play(ed) Oblivion question the choices in most of the reviews, especially things like shadows on grass versus actually more grass.
It's subjective, and just like any canned review is not the only source of good info. The best thing to do is to read MORE reviews and get a global picture, which goes contrary to what this article seems to state which is "[H]'s method is the only way, ignore all the other heathens..."
I'll give [H] editorial staff more credibility when they don't contradict themselves so much, especially when it's obvious those contradictions are only in one direction.
Considering all the brouhaha over the power consumption of the R600 versus it's potential only matching or barely exceeding that of the GTX, I'm left wondering why the change in stance when the HD3870X2 consumes more power and also performs below the GTX? This reaction of "At full load it understandably eats up a large amount of power, but still not tremendously compared to the single 8800 GTX." seems contrary to sentiments less than a year ago.
Why the change of heart, a major change in the price of electricity, availability of PSUs, what? Or is this because the authors finally realized that this is unfortunately the case where for the top of the hill people don't care about a few watts, they'll SLi/Xfire together behemoths if that's what's required to get smooth Crysis gameplay on a 30" LCD?
The idea that this is any more relatable to people who don't have the same rig is a little optimistic to say the least.
To think there is one right way of doing this misses the point, and this article is more about defending one's own choices rather than looking to move the entire methodology/model forward.
However coming from a place that like to comment on the validity of other sites openly, it's not surprising the back/forth on this issue.
These video card reviews are just a part of the picture when making an informed decision. Attacking Anandtech may not have been the wisest move, as their methods aim to be scientific (and sometimes they fail) and by definition, Kyle's is a subjective review, and they form an opinion by popping in the card and playing games.