It doesn't represent sales, it is simply adoption of a certain user base to certain products.
It doesn't make sense to use percentages.
They don't post how many people took the survey anymore.
There are still certain anomalies that can take place, like back in the day I switched out my 8800gt to a 4850 and was benching/ocing them it read my GPU as a Nvidia card.
Hell, I'm an avid gamer but haven't signed into any of my Steam accounts for over 6 months.
It is nothing more than a very broad overview of hardware trends and trying to establish/compare sales numbers for specific products from the survey is not a good idea.
It doesn't make sense to use percentages.
They don't post how many people took the survey anymore.
There are still certain anomalies that can take place, like back in the day I switched out my 8800gt to a 4850 and was benching/ocing them it read my GPU as a Nvidia card.
Hell, I'm an avid gamer but haven't signed into any of my Steam accounts for over 6 months.
It is nothing more than a very broad overview of hardware trends and trying to establish/compare sales numbers for specific products from the survey is not a good idea.
Final Destiny
Newcomer
Since trinibwoy seems to be in the know/have insider info on the number of participants to the STEAM survey, maybe he can shed some light on it....
No... it doesn't poll every single one of those 5 million. Not even remotely close. Of course, Steam doesn't release the sample sizes, so it's not even a reliable source anyways.
UniversalTruth
Veteran
Charlie doesn't think like this.
http://semiaccurate.com/2012/06/08/computex-2012-amd-to-launch-tahiti-2-next-week/
And still, real world situation kind of confirms it- only 2 out of 20 GTX680s available in stock at newegg.
Um, not polling the full user base does not mean it's unreliable. It very much depends upon specifically how the subset of the user base was selected.
The first thing I'd point out with regard to sample size, however, is that the rather consistent results between months of even the very low marketshare GPU's definitely shows that they have enough sample size to get a precise measurement. The total number of people in the sample is quite clearly not a problem here.
Whether that measurement is also accurate depends entirely upon whether or not the sample is biased. And that is very difficult for us to know from the outside. But since the overall marketshare results don't seem to be all that out of line, well, I would tend to expect that there isn't that much bias in the results.
Well see that's the beauty of a large sample size - one person doesn't matter at all
5 million is an instantaneous count. The total number of unique users that log into Steam in a given year is obviously far higher than that. You don't need to poll every member of a population to generate reliable numbers - that's one of most basic tenets of statistical/probability theory. Given the ease of data collection in this case the sample/population ratio is probably far higher than most other use cases (e.g. presidential election polls usually include only about 1000 people)
Steam apparently asks 1/12th of their member base each month to participate in the survey. I think it's pretty clear that the sample size is very large even if only 10% of those polled opt-in.
You guys do understand that you only need to sample a relatively small and random subset of a universe to get a reliable estimate for a statistic right? As long as Steam's polling methodology doesn't favor one IHV or one particular SKU I see no reason to think it's not reliable - at least from a mathematical standpoint :smile:.
If you want to discredit Steam you have to identify sources of selection bias or sampling error. Everything else is just pointless blustering.
Well, not being steam surveys and thus having a rather limited number of participants are two quickpolls in german enthusiast forae are those two quickpolls. But apart from having a strong tendency of over emphasizing the high-end portion of the market, both communities see GTX 680 roughly on par with HD 7970 despite being on the market for a significantly shorter amount of time.
http://www.forum-3dcenter.org/vbulletin/showthread.php?t=528393
http://extreme.pcgameshardware.de/q...nutzt-du-derzeit-zum-spielen-juni-2012-a.html
http://www.forum-3dcenter.org/vbulletin/showthread.php?t=528393
http://extreme.pcgameshardware.de/q...nutzt-du-derzeit-zum-spielen-juni-2012-a.html
Babel-17
Veteran
"opt-in" and "random sample" sort of contradict each other.
It would be like trusting the results of a survey to determine how often people bathe or brush their teeth or donate to charity, few who rarely do so would choose to opt-in to such a survey, I expect.
But I'm not saying the data Steam collects is insignificant. I'd give some weight to the numbers they collect.
It would be like trusting the results of a survey to determine how often people bathe or brush their teeth or donate to charity, few who rarely do so would choose to opt-in to such a survey, I expect.
But I'm not saying the data Steam collects is insignificant. I'd give some weight to the numbers they collect.
20% faster than a 7970?! This forum needs a "banging head against a wall" emoticon. You can get 20% more perf by moving a slider! 20% is not a new product. Please people, STOP BUYING the crap put out by these companies, and force them to release decent upgrades!
Err, what? So now we should have nothing but contempt for anything that brings less than a 30% improvement? More?
That makes no sense.
Well, all surveys by necessity are opt-in. Doesn't mean randomization isn't possible, or can't help. Nor does it necessarily mean there's a problem: there has to be a bias towards or against opting in for some of the survey information."opt-in" and "random sample" sort of contradict each other.
And I don't think we have any reason to believe that people who like buy parts made by one manufacturer are more or less likely to opt-in than those who buy different parts.
Hé they dont call it HD8000 and dont say this will be a new architecture and sell it as a new series they will just use the 28nm progression on quality of Asic for throw a 7970 who will be clocked at higher rate ... I have offtly tell AMD have been really conservative ( for some reason ) with the core and memory speed of the 7970, this have let Nvidia a big margin for set the 680 performance.
anyway, it is 20% on clock progression not 20% more perf.. ( even if the 7970 architecture scale really good with core speed ).
See it as the X1950XTX for the X1900XTX,... nothing more.
Ofc, you can buy allready a MSI lightning ( excellent card ) or a gigabyte at 1050mhz... but i can understand marketing wise AMD will not let Nvidia play his game alone.
This explain too the report of the 7990 ( or 7970x2 ) as ofc the card will be based on the new revision.
Last edited by a moderator:
AMD users are known to be more likely to opt out because that gives them a reason to argue against the validity of the Steam results.Chalnoth said:
Or so I've been told...
Last edited by a moderator:
"opt-in" and "random sample" sort of contradict each other.
Wow terrible teeth brushing analogy lol. Owners of a particular brand of video card aren't more or less likely to opt-in to the survey.
Well see that's the beauty of a large sample size - one person doesn't matter at all
Well I think I do.
Was talking about D3, lots of PC gamers have tunnel vision when something new comes out.
At the prices they are releasing them at? Yes. And I'm letting you be conservative at 30%. I would say new generation > 50%. It's obvious both the 7970 and 680s were "soft" balls, and could have and should have been better. Just because people bought them isn't a case for us as the buyers to say "oh well, it seems to be the right perf/cost". We got screwed this time around.
I'm currently borrowing a friends 670 OC (hes on holiday, and a very good friend). The fan on my 4870x2 is finally dying (4 years constantly turned on, helluva playa!). This is a ~£380 card, 4 years after I bought the 4870x2 for £360. It's not a great deal faster @ 2560x1600. Certainly nothing that was unplayable before is now playable. I would say at best, stuff is 50-100% faster. After FOUR years. Insane.
DavidGraham
Veteran
Didn't you know that Charlie is SEMI-accurate ?
Similar threads
