GeForce RTX 3060 debuts at number 84 in Steam HW Survey

[HEXUS]

And Auz & NZ retailers have listed Gigabyte and MSI GeForce RTX 3080 Ti graphics cards.

Read more.

[mikeo]

and the madness continues ......................

[Iota]

1. NVIDIA GeForce RTX 3070 1.38%
2. NVIDIA GeForce RTX 3080 0.85%
3. NVIDIA GeForce RTX 3060 Ti 0.39%
4. NVIDIA GeForce RTX 3090 0.37%
5. NVIDIA GeForce RTX 3060 0.17%

3.16% of the Steam Hardware survey users are using a 3xxx card from Nvidia. Pretty poor showing for some of the cards that have been out since 2020. Looking great for AMD 6xxx series though! (/s)

[3dcandy]

1. NVIDIA GeForce RTX 3070 1.38%
2. NVIDIA GeForce RTX 3080 0.85%
3. NVIDIA GeForce RTX 3060 Ti 0.39%
4. NVIDIA GeForce RTX 3090 0.37%
5. NVIDIA GeForce RTX 3060 0.17%

3.16% of the Steam Hardware survey users are using a 3xxx card from Nvidia. Pretty poor showing for some of the cards that have been out since 2020. Looking great for AMD 6xxx series though! (/s)

But steam is in no way shape or form indicative of what is being used....

[Tabbykatze]

But steam is in no way shape or form indicative of what is being used....

There's not really any other better way though...

[Iota]

But steam is in no way shape or form indicative of what is being used....

Well it is amongst Steam users currently active....

[LSG501]

But steam is in no way shape or form indicative of what is being used....

Considering steam is one of the major gaming platforms and the geforce 3xxx series are supposed to be for gamers, with lower than previous gen pricing, I'd say it's pretty indicative of the current gpu climate for such a low share....

Also as said it's not like we can find out anywhere else about their users.

[SlickR]

But steam is in no way shape or form indicative of what is being used....

Well it is amongst Steam users currently active....

And taking the survey, say third of steam users are active that month when the survey is going on and say 1/4th take the survey, that is only a small percentage of steam users being represented.

I think steam has something like 200 million accounts, but a lot of those are double or triple account for just 1 person, so say the real number is 100 million, and about 40 million are active in a month and say a 4th take the survey, that is "only" 10 million people's hardware being represented.

[DanceswithUnix]

, that is "only" 10 million people's hardware being represented.

10M samples would be pretty good. Surely if the sample number was too low, then the results would jump around wildly from one month to the next, yet they never have.

I think Steam survey is riddled with selection bias problems that make it hard to interpret, but I don't have an issue with sample size.

A good showing from the 3070 there, I wonder how many of those are in laptops.

[3dcandy]

Whilst agreeing to some of the points raised it's still the best but way off the mark generally.

But of course it's the best as really there isn't anything else out there

[kompukare]

Considering steam is one of the major gaming platforms and the geforce 3xxx series are supposed to be for gamers, with lower than previous gen pricing, I'd say it's pretty indicative of the current gpu climate for such a low share....

Surely the relevant question then is: how do the 3000 series compare to previous generations at a similar point in time?
Then we can make some allowance that the 2000 series was overpriced.

[3dcandy]

Surely the relevant question then is: how do the 3000 series compare to previous generations at a similar point in time?
Then we can make some allowance that the 2000 series was overpriced.

Supply restrictions make that comparison really useless.....
No supply issues for the 20xx series at launch or in the 6 months afterwards

[LSG501]

Surely the relevant question then is: how do the 3000 series compare to previous generations at a similar point in time?
Then we can make some allowance that the 2000 series was overpriced.

At originally listed retail prices they wouldn't be overpriced (in relative terms) and you can't really compare the current inflated prices due to scalpers/crypto/short supply to the 2xxx series prices.

Also my comment about price was regarding rrp being lower, not actual retail (should have made that a bit clearer reading it back this morning)

Also I'm pretty sure that Nvidia have said they were producing more 3xxx versus 2xxx at the same period (and we still have a shortage....) so it's not really something you can compare either, unless you're purely interested in how badly the shortages have impacted purchases by gamers, although that wouldn't be completely accurate either.

[philehidiot]

It's not accurate but it's useful as a trend. Therefore it is useful as those inaccuracies will be fairly consistent over time or slow enough changing to model out. So if you want to compare the market penetration of two company's gaming products, you can do that. You can't use absolute numbers but you can definitely use it as a rough proportion. If you want to compare take up by gamers between different product launches, you can do that too - look at the slope of the curves of each model or group you are interested in and compare one group's curve to the other.

That would give you a good idea of how many cards are making it into the hands of real gamers and allow an assessment of the narrative. Are they all going to miners or is it pure demand as some companies are saying?

If you wanted to be really clever, you could use data from other launches about numbers sold and so on and adjust for Steam user growth. This would help give you some real numbers to work with as a guide.

There is plenty you can do with data like this, even if the absolute values are cack.

[3dcandy]

It's not accurate but it's useful as a trend. Therefore it is useful as those inaccuracies will be fairly consistent over time or slow enough changing to model out. So if you want to compare the market penetration of two company's gaming products, you can do that. You can't use absolute numbers but you can definitely use it as a rough proportion. If you want to compare take up by gamers between different product launches, you can do that too - look at the slope of the curves of each model or group you are interested in and compare one group's curve to the other.

That would give you a good idea of how many cards are making it into the hands of real gamers and allow an assessment of the narrative. Are they all going to miners or is it pure demand as some companies are saying?

If you wanted to be really clever, you could use data from other launches about numbers sold and so on and adjust for Steam user growth. This would help give you some real numbers to work with as a guide.

There is plenty you can do with data like this, even if the absolute values are cack.

Indeed there is, but so far (as far as I can see) nobody has....
Kinda makes it all a bit moot and the article just rehashes it

My mates in US of A are saying that miners are snapping up more cards now, not less as ETH is going to change shortly to make gpu mining less effective. I only know a handful who have managed to get 30xx cards for gaming worldwide and even less of AMD's latest

However stocks of 5xxx cpus appear better slightly and in the US I'm hearing Intel are having to discount massively to get sales. People appear willing to wait for the cpus but not the gpus

[kompukare]

It's not accurate but it's useful as a trend. Therefore it is useful as those inaccuracies will be fairly consistent over time or slow enough changing to model out. So if you want to compare the market penetration of two company's gaming products, you can do that. You can't use absolute numbers but you can definitely use it as a rough proportion. If you want to compare take up by gamers between different product launches, you can do that too - look at the slope of the curves of each model or group you are interested in and compare one group's curve to the other.

That would give you a good idea of how many cards are making it into the hands of real gamers and allow an assessment of the narrative. Are they all going to miners or is it pure demand as some companies are saying?

If you wanted to be really clever, you could use data from other launches about numbers sold and so on and adjust for Steam user growth. This would help give you some real numbers to work with as a guide.

There is plenty you can do with data like this, even if the absolute values are cack.

Yes, a bit of work but between the Steam data and the hashrate data DanceswithUnix looked at yesterday, we could make some pretty good guesses.

I suspect, that

1. demand is crazy
2. volumes aren't actually that bad
3. plenty of gamers got hold of cards even if they overpaid
4. plenty of miners got hold of cards even if they overpaid

So both groups have overpaid for cards and driven up prices.
I had expect mining to die down by now, but yesterday when responding to Dances' post I looked up the crypto prices again: ETH has gone up to over $3000 from around $1000 in January, so I guess the ponzi scheme hasn't collapsed yet.