Wich card is the king of the hill (nv40 or R420)

[Fred da Roza]

I wouldn't expect 50-50 parity, but a 12% lead given the error margins and uncertainty of the market right now might just constitute an irrelevent or statistically insignificant data point.

Although IMO the chips are clearly more evenly match this round this is ridiculous. One could say 81% believe the R420 is equal or better; a 64% lead.

 

[AlphaWolf]

Roughly 40% of the poll here thinks the cards are equal or the NV40 is better (for the record, I voted "about equal" and I think many fair minded people who admit things are uncertain did the same, vs others who are loyal/zealous on particular vendors), what do you think the margin of error is on this poll? If you believe in your own poll, and with B3D being the most "non biased" of polls vs fansites, then ATI only has a 12% lead (with unknown error margin) in the fact of

a) unknown quality of final NV40s shipped
b) driver updates
c) future FarCry patches, and Doom3 performance
d) new SKUs to come this summer, plus NV41 and 44. Many people might be concerned over power/size/noise issues which might be rectified

I wouldn't expect 50-50 parity, but a 12% lead given the error margins and uncertainty of the market right now might just constitute an irrelevent or statistically insignificant data point.

We'll find out when NVidia and ATI announce report sales figures in 2 quarters.

Obviously pricing and availabilty plus the performance and availability of the mid range cards will determine the real winners.

But where does 12% lead come from? Nv40 votes count three times?

I am sure an I like pie option would have won out.

 

[DemoCoder]

Again, as I stated, similar results are being shown in other polls at other locations asking more or less the same questions - these same polls represented the market as it stood in terms of overall features and performance and hence where the sales are - I see no reasons why these don't stand today as the current user preferences for what they feel will best meet their needs.

First of all, do you really believe that these other sites represent completely separate user bases? People who frequent NVnews, also frequent rage3d, 3dgpu, firingsquad, anand, toms, et al. In fact, the really zealous make it a habit to visit other forums to cause trouble and vote in polls.

It's very simple if you've taken an elementary course in statistics. Opinion polls have to be randomly sampled among the target audience to be legitimate. Again, you seem to have no skepticism at all in this case, where skepticism is clearly warranted given the basic foundations about statistics and how to conduct polls.

If you were to put up a poll about the FX5200, I bet most members of these forums would say "it sucks" The reality is, it sold very well.

 

[DemoCoder]

But where does 12% lead come from? Nv40 votes count three times?

I'm not counting the lead over the NV40. I'm counting the lead over the default 50/50 position. If a poll was dead even between the cards, with 50 people voting NV40 and 50 voting R420, this poll would represent 12 people switching sides from NV40 to R420. For every person that switches votes, you get a 2x change.

Easy to manipulate statistics isn't it?

 

[T2k]

If you were to put up a poll about the FX5200, I bet most members of these forums would say "it sucks" The reality is, it sold very well.

...which has nothing to do with the fact: fx5200 IS sux, a piece of crap.

Edit: formatting

 

[AlphaWolf]

I'm not counting the lead over the NV40. I'm counting the lead over the default 50/50 position. If a poll was dead even between the cards, with 50 people voting NV40 and 50 voting R420, this poll would represent 12 people switching sides from NV40 to R420. For every person that switches votes, you get a 2x change.

Well if there was even a shred of sense in your method perhaps.

Easy to manipulate statistics isn't it?

Apparently not quite as easy as you thought.

 

[DemoCoder]

I'm not counting the lead over the NV40. I'm counting the lead over the default 50/50 position. If a poll was dead even between the cards, with 50 people voting NV40 and 50 voting R420, this poll would represent 12 people switching sides from NV40 to R420. For every person that switches votes, you get a 2x change.

Well if there was even a shred of sense in your method perhaps.

My method is a perfectly sensible interpretation. If I want to count the number of customers above "neutrality" who chose R420 over NV40, it is perfectly within reason to do so. Claiming that a non-randomly sampled poll is representative is not.

But perhaps you should spend more time trying to be reasonable about 3d cards, and less time trying to "score points" by sniping.

 

[ANova]

It's also possible the performance difference is small enough that nvidia would rather not bring it to attention. And we all know it offers no new visual enhancments.

False assertion. You keep pretending that the only thing the 6800 adds is dynamic branches. No matter how often you repeat it, it's still wrong.

You seem rather sure of yourself, have you personally seen the differences between the two? I'm merely being the devil's advocate.

I am also not pretending the 6800 only adds dynamic branches, it adds everything you listed a few posts back. The simple matter is the majority of those only result in performance improvments and make the work for developers somewhat easier, nothing more. The increased number of instructions can lead to better visual quality but the card will be incapable of running at decent speeds. UE3 proves this.

If you were to put up a poll about the FX5200, I bet most members of these forums would say "it sucks" The reality is, it sold very well.

It sold well because not everyone can afford a 5900 there are many people who still believe in the nvidia name and will buy their products for that reason and that reason alone. Of course the heavy marketing and touting the card as a cheap DX9 device helped matters as well.

 

[Bjorn]

...which has nothing to do with the fact: fx5200 IS sux, a piece of crap.

I wouldn't say that. It has the same performance as the R9200 and have more features. I'm disgregarding DX9 there but it has MSAA which is usable in older games.

But you of course get what you pay for.

 

[ANova]

...which has nothing to do with the fact: fx5200 IS sux, a piece of crap.

I wouldn't say that. It has the same performance as the R9200 and have more features. I'm disgregarding DX9 there but it has MSAA which is usable in older games.

But you of course get what you pay for.

Alot of people bought the card specifically for it's DX9 support. It may be as fast as the 9200 in DX8 and lower but once it tries to run DX9 it's performance drops in half or more. It's just a selling point for nvidia, and it ultimately ends in screwing the consumer.

 

[AlphaWolf]

My method is a perfectly sensible interpretation. If I want to count the number of customers above "neutrality" who chose R420 over NV40, it is perfectly within reason to do so. Claiming that a non-randomly sampled poll is representative is not.

um ya... right.

But perhaps you should spend more time trying to be reasonable about 3d cards, and less time trying to "score points" by sniping.

Do as you say, not as you do?

 

[Bjorn]

Alot of people bought the card specifically for it's DX9 support. It may be as fast as the 9200 in DX8 and lower but once it tries to run DX9 it's performance drops in half or more. It's just a selling point for nvidia, and it ultimately ends in screwing the consumer.

I know people that bought the R9200 and thought it had a bit more in common with the R9700 then it actually does.

But i agree, it's hardly a DX9 card for the gamer. But neither is the 9600 SE from what i've seen.

 

[jimmyjames123]

As I said, we don't necessarily know that the new instructions added to ATI's VS have been utilised by their compilers yet.

And we don't necessarily know that the new instructions added to ATI's VS have not been utilized by their compilers yet. Anyway, you specifically stated that the NV40's VS performance appears to be lower than the R420's, and I think that it is much to early to really jump to any conclusions about VS performance between the two cards. That's just my opinion.

Generally speaking its assumed that we're comparing the X800 XT PE against the 6800 Ultra.

The statement as stated said NV40 vs R420, and not X800 XT PE vs 6800 Ultra. Since there are variants for both the NV40 and R420, then I think it is more appropriate to be specific about each card when talking about how they compare in terms of performance. This again is just my opinion.

And this is an in-quantifiable factor. You can't necessarily bank on ATI not gaining further performance as well as there are general architctural changes, a memory bus that clearly needs some learning and a new shader compiler on the way - I can't quantify what, if any, performance increased these may give for ATI, can you?

Well of course it is unquantifiable. That doesn't mean that the issue doesn't exist, as it's quite obvious that the NV drivers are very raw in relative terms. Time will tell, I assure you.

 

[DemoCoder]

um ya... right.

Roll your eyes all you want. You have still provided no rebuttal to the criticism that online polls are bunk and polls without random sampling are unscientific and inaccurrate. Try searching on "straw poll" or "literary digest" to learn some history, or take a basic course in statistics. It simply doesn't matter if a previous straw poll turned out correct.

If you don't like my counting strategy, you can do more to criticize it then an emoticon can't you? I choose as my starting point, equal bias, e.g. the position that "both cards are about the same". Let people decide which video card to buy by flipping a fair coin. Heads, ATI, tails, NVidia. Now, let's bias this coin so that 12 out of 100 flips, a tails will become heads. With no bias, after 100 flips, you will have 50 for NVidia, and 50 for ATI. The baseline for our experiment is that if the cards were equivalent, roughly half the time people would pick ATI. Now we want to measure how far we deviate from perfect equality. With the bias in decision making, 12 of NVidia's flips will become ATIs, giving you 38 vs 62.

I can continue to extrapolate in this manner. Anytime a consumer has to choose between the two cards, roughly 12% of the time, he will choose against NVidia and for ATI. My model has the same predictive power as your "24%" model, and will agree with experiment, but without the "figure inflation"

Now, I'm being somewhat facetious here on purpose to show you how the statistics are bogus and won't map to real sales figures, especially given the sample bias.

But you know what, if you're going to argue, could you atleast argue intelligently?

Do as you say, not as you do?

I am being reasonable. I haven't changed my position, period. I have consistently said "let's wait and see" and "I think they are mostly evenly matched". You're the one who's prematurely committed. I'm merely responding to people who are prematurely making conclusions.

You, on the other hand, did nothing but join the thread to make a personal quip, followed by information free emoticons. Wow, you make such great arguments.

 

[jimmyjames123]

And are differing polls around the web such as the one displayed in this thread not a reflection of the balance of features many people are looking at right now? I'd assume so.

Oh come on. You realize how short-sighted this poll is, don't you? Saying that one card "wins" over the other when comparing NV40 to R420 is totally silly, and you know it. It really boils down to each individuals preferences, what games they play, what feature set they want, what AA or AF quality that they want, etc.

It is also much too simplistic to say "R420 wins" or "NV40 wins", especially since there are different cards that make up the NV40 and R420 lineup. I would think that most people would agree that the X800XT PE has the slight edge in general at the moment over the 6800 Ultra with respect to sheer performance. However, this is entirely dependent on what game is being tested, and what AA/AF and resolution setting is used, naturally. The NV cards seem to be particularly strong with OpenGL games, and they should be particularly strong with shadow-intensive games like Doom3 too. At the same time, I think that most people would agree that the 6800 GT has the slight edge in general at the moment over the X800 Pro. Here again, it depends on what game is tested, and what AA/AF and resolution settings are used.

Again, as I stated, similar results are being shown in other polls at other locations asking more or less the same questions - these same polls represented the market as it stood in terms of overall features and performance and hence where the sales are - I see no reasons why these don't stand today as the current user preferences for what they feel will best meet their needs.

Think more carefully about what you are saying here. You argument does not hold up this time for one main reason: almost no gaming enthusiast has been able to test the 6800 Ultra, 6800 GT, X800 XT PE, or X800 Pro in their gaming machines! We also have had little to no exposure to many of the new features on the next gen hardware, such as SM 3.0, Ultrashadow II technology, 3dc, etc. To top it off, it is very clear that the initial set of NV 6x.xx series Forceware drivers are still relatively raw and buggy. It is very closed-minded IMHO to put a lot of stock into these overly simplistic polls that have been showing up around the internet based on a group of cards that almost no one has had the priviledge of gaming with or testing or owning.

 

[madshi]

If you don't like my counting strategy, you can do more to criticize it then an emoticon can't you?

Let's say we had:

NV40: 50%
R420: 0%
equal: 50%

If I understand your counting strategy right, NVidia would have a 0% lead in that specific situation, or am I misunderstanding you?

 

[ANova]

I know people that bought the R9200 and thought it had a bit more in common with the R9700 then it actually does.

I'm interested as to what they were expecting. ATI never claimed that it had DX9 support nor that it was based on the R300 core. Why would anyone automatically think it similar to the 9700? That sounds like the result of sheer ignorance to me.

However, this is entirely dependent on what game is being tested, and what AA/AF and resolution setting is used, naturally. The NV cards seem to be particularly strong with OpenGL games

This is due to bad code, it has nothing to do with the hardware. And being based on code it can be changed and/or rehauled, as ATI is currently doing.

 

[DemoCoder]

Pretty much. In such a bizarre distribution, it's definately subscribe most of the NVidia votes to selection bias. If the poll were scientifically conducted, I'd might suspect different. In other words, if this were a presidental election, I'd suspect a percentage of the NVidia 50% figure to be "diehard partisans" and another part of it to be made up of independents who chose NVidia.

My point is: I believe a big chunk of the voters on Nvidia and ATI sides to be partisans, and I don't believe they represent the vast market, which by and large, isn't overly hardcode geek loyal to an IHV.

The question isn't whether you like the aesthetics of the counting technique, the question is whether it will reproduce the same results.

You could argue that everyone who checked "Nvidia" or "ATI" will never budge, and that the only "50/50" coin-flip you can ascribe is to the fence-sitters. I just don't believe this accurately represents the situation of the market. I'll be honest, part of the reason I came up with the count is to induce such an argument. People want to argue that the poll need not be scientific to be accurate, but then they insist that any interpretation of those numbers must be scientific.

Statistics can lie, and you can lie with statistics.

 

[madshi]

There are currently 222 votes for R420. I really don't believe they are all (or most) just partisans. I think maybe 20 are, maybe 40 are, but not 222. DC, let's forget for now that you seem to favor NV a bit, don't you see things a bit from the developer point of view? I understand that you prefer the NV40 from that point of view. But tell me as a *gamer*, what advantages does the NV40 bring to me? It might gain additional speed once SM3.0 is cleverly used by games. What more? I don't expect much (if any) IQ improvements - even NVidia sais SM3.0 is mainly for ease of programming and for gaining speed! R420 on the other side has the advantages for me as a gamer, that it offers usable 6xAA (not sure about temporal AA yet, haven't seen it with my own eyes yet) and it runs a lot cooler. Speed wide I think you can't go wrong with either card, so that is not of much concern for me. Think about it: The cards are so damn fast that the R420 will enable me to use 6xAA for most older games! <drool>

 

[jimmyjames123]

This is due to bad code, it has nothing to do with the hardware. And being based on code it can be changed and/or rehauled, as ATI is currently doing.

And what's your point? Software and hardware go hand in hand. We all know that ATI is hard at work trying to overhaul their OGL code. This is a good thing. But at the same time, isn't this just another "unquantifiable" element that DaveB alluded to earlier? Whether you like it or not, NV has the clear edge in OpenGL performance at the moment. This point is hardly debateable.