A new graphics card tops the Steam survey for the first time since 2018

midian182

Posts: 9,734   +121
Staff member
What just happened? The latest Steam survey has arrived with a big change: a graphics card has knocked the GTX 1060 off the top spot for the first time since January 2018. November was also an excellent period for AMD CPUs, which reversed a months-long trend of declines to grab an almost 4% user share back from Intel.

The most popular graphics card among Steam survey participants had been the GTX 1060 since it replaced the GTX 750 almost five years ago. But the Pascal card has now been replaced by a Turing entry, though it's not the RTX 2060 as some might have expected—that card lost users last month—it's the budget GTX 1650 that first arrived in 2019.

November was another month that showed signs of conservative consumer spending. Barring a couple of exceptions, the top-performing GPUs (below) were all on the cheaper end of the scale. The GTX 1650 was second with a 0.66% increase, while the RTX 3050, GTX 1050/Ti, GTX 1650 Ti, and GTX 1660 Ti all made significant gains. The AMD Radeon RX 6500 XT, which was recently on sale for $99, also did well.

Not for the first time, the RTX 3060 Laptop GPU was the month's top performer, but the desktop version lost -2.06% of its user share; November wasn't a good month for the more expensive desktop cards.

Moving onto CPUs, AMD finally saw its fortunes turn around after four straight months of losing users to Intel. Team red snatched a 3.88% share from Chipzilla, moving it back above the 30% overall milestone—could the Zen 4 powered Ryzen 7000 series finally be starting to sell well after its slow start, possibly thanks to the recent reductions?

Elsewhere in the survey, Windows 11 returned to its usual state of regularly increasing users following an uncharacteristic decline in October. Microsoft's newest OS was up 4.61% in November as Windows 10 fell -3.31%, pushing the former closer to the 30% overall mark. And while it has been hanging on for a while now, Windows 7's share fell to just 1.88% overall.

Permalink to story.

 
Unsurprising. What it really shows is that nothing has changed at all regarding "the £150-£250 tier is still the most popular" despite BS attempts to skew it up with ripoff "up-tiering". The reason there's no newer GPU's is nVidia have simply stopped making them, so people just hold onto the older ones for longer:-

£70 tier = GT 1030 -> nothing at all.
£150 tier = GTX 1050Ti -> GTX 1650 -> GTX 1650S -> GTX 1630.
£200 tier = GTX 1060 -> GTX 1660 -> nothing at all.
£300 tier = RTX 2060 -> RTX 3050.

^ That's right, they "replaced" the GTX 1650S (1280 shader) with the GTX 1630 (512 shader) at virtually the same price. No surprise at all people have been snapping up the GTX 1650 (896 shader) as the lesser downgraded version instead (4x bus crippled RX 6400 completely falls apart on PCIe 3.0 boards). RTX 3050 has always been in a different tier.

tl:dr - Post 2020 low-end GPU's were and still are an absolute ripoff. If the RTX 4050 gets launched at more than $200, then expect the 1650 to remain in the same spot by 2025...
 
Last edited:
And the 1650 is slower than the 1060 it's replacing, big yikes
You're assuming that people are buying 1650s to replace 1060s, which is not happening.

This is the Steam survey after all, the one everyone says is inaccurate.

At best we can conclude that some people are decomissioning their 1060s, likely replacing them with a smattering of different GPUs based on their circumstances, or laptops as the rise of mobile GPUs suggests.
 
It seems the 1xxx is still king when playing games on the biggest platform on the planet!

So, where are those top of the line 3xxx and 4xxx and who is playing them and where?? Are they still trying to milk Crapto junk?? Even the 2xxx is very poorly represented!

Top of the line 3xxx and 4xxx seem to be for bragging rights only!!
 
Prediction: the 1060 regains the top spot next month or maybe the one after because noisy polls are noisy.

And oh no! AMD gained a lot of market share in both GPUs and CPUs! But they lost market share last month... what could have happened in just a months' time??

Other fun trends, lots of people:

uninstalled 16 and 32GB in favor of 8GB
downgraded from 6-core CPUs to 4-core
removed 6, 8, and 12GB video cards for 1GB ones
traded 1440p monitors for 768p as well as other ones
stopped speaking Chinese in favor of English and Russian
traded in 1TB drives for 250GB ones

conclusion: LOL
 
Last edited:
Prediction: the 1060 regains the top spot next month or maybe the one after because noisy polls are noisy.

And oh no! AMD gained a lot of market share in both GPUs and CPUs! But they lost market share last month... what could have happened in just a months' time??

Other fun trends, lots of people:

uninstalled 16 and 32GB in favor of 8GB
downgraded from 6-core CPUs to 4-core
removed 6, 8, and 12GB video cards for 1GB ones
traded 1440p monitors for 768p as well as other ones
stopped speaking Chinese in favor of English and Russian
traded in 1TB drives for 250GB ones

conclusion: LOL
Valve’s surveys could do with a bit more transparency behind the sampling, I think. St the very least, the sample size would be a useful indicator has to how many Steam users actively agreed to sending the information and for a more robust data set, survey statistics over 6 and 12 month periods. There again, game publishers tend to use other data samplers, such as UL, to get a better understanding of potential userbases.
 
And the 1650 is slower than the 1060 it's replacing, big yikes
You've somehow managed to miss the point completely, with Nvidia's numbering system.

The 1650 was never intended to surpass the 1060.
The 1650 was intended to surpass the 1050 ti, which it does.
The 1660 was intended to surpass the 1060, which it does handily..
The 1660 ti version manages to approximately equal the 1070.

This is intended to be a strictly "apples to apples", comparison. I don't care what, I, or anyone else, coulda, woulda, or shoulda bought. in the Radeon lineup, in lieu of any one of those cards.
 
St the very least, the sample size would be a useful indicator has to how many Steam users actively agreed to sending the information and for a more robust data set,
The sample size would be virtually useless, since the people participating obviously vary from month to month.

The differential between the users who maintain uploading their data on the same machine without "upgrades", is really only a portion of the respondents.

I'm sure the user who wants to tell the world they have installed Windows 11, a new video card, or whatever, is more likely to respond, which slews the survey in a profound, (too much adjective?), rather short term way.

Them there's the user who still is soldiering on with the same rig. Perhaps he or she is thinking, "I told them last month, nothing has changed, so why bother"?.
 
What amazes me about this list is how many different gpus there. I mean the most used is only 6%.
Optimizing games for all of these gotta be hard.
We need more of the cheaper gpus, gpus that behave like a weaker middle end but cost like low end.
I am hoping very much amd will offer something this year, something that plays comfortably all new games on high in 1080p and costs 150 less than nvidia

Next top: gtx 1070. Or rtx 2060
 
I asked Gabe many years ago, why he calls it a "survey", when not everyone is included... He also got cocky when someone suggested Valve could just do a "real" survey of GPU's using a newly released title (ie: Cyberpunk 2077), or using a particular DirectX, or Vulkan API. He was asked why that would hurt... his annoyance of that question, and non-answer told the whole story.


I run 3 rigs for gaming, my last survey on any of them was 2 years ago... (I game every day).
 
I asked Gabe many years ago, why he calls it a "survey", when not everyone is included...

"... why he calls it a "survey", when not everyone is included."....

Really?? So, you think any survey about XYZ should include ALL the households in a country, (in this case every gamer or anyone who has a GPU) otherwise, it's not a "real survey"??

Yes indeed, every survey should include all and take months or even years, otherwise, they're not "real"!!
 
Since these numbers are relative to each other and the list doesn’t account for 100% of the cards, couldn’t the 1650 also have lost users?
 
"... why he calls it a "survey", when not everyone is included."....

Really?? So, you think any survey about XYZ should include ALL the households in a country, (in this case every gamer or anyone who has a GPU) otherwise, it's not a "real survey"??

Yes indeed, every survey should include all and take months or even years, otherwise, they're not "real"!!

No.

Do you even have a Steam account..? If you only survey people who have certain GPU's and not everyone, you are going to get skewed results. And if you ONLY publish data sets, and time frames that don't include just released GPUs.. u can skew data sets.

Why does steam constantly over sample old accounts, instead of say... push a random survey starting next week, to all people who log into steam over the next week..? To capture the actual CURRENT usage of GPU's.

That^ is a survey...


But, doing basic data collection over time.. and then compiling it into a report, is not a survey. It's ONLY a survey, when everyone in the survey is parsed within the same time/time frame. (You can not add data sets together...)
 
These Numbers Make NO sense. I cannot believe for 1 second that people are dropping 2060/3060/3060ti and 3070 cards, and we are seeing an increase of 1050ti/1650/1605ti.

Nope do not buy it. More like, nobody submitted of Thanksgiving break.
 
The GeForce RTX 3060 Laptop GPU is more respectable than the GTX 1650. A quick eBay search shows USED 1650's selling for 110 US Dollars. I thought they would be much cheaper than that.
 
OK, first not everybody wants a used video card.
Second,, xx60 has always been where Nvidia places the first card that could actually qualify as an, (ostensibly), "real gaming card".
And lastly, were you just window shopping, so you could have something to say when you came back here?
Because if you actually wanted one, but weren't willing to pay the asking price, then there's a "make offer" button for that. Soooo.....?

All the 1650 was supposed to accomplish on its own recognizance, was to be better than the 1050 ti. Which it is

In other news, I think I heard, ("out of the corner of my ear", as it were), that the import tariff exemption for graphic cards was about to expire. If that's true, then y'all better have yer best pissing, moaning, and whimpering acts, primed and ready to go.. They won't do anything about any potential further price hikes, but it might do you some good to get that that frustration out of your systems.
 
Last edited:
Depends what you want to use that 1650 for. A media center or light/older games in systems without pcie power (Dell or HP) should do the trick. I see some passive models and half height also. Performance is more or less equal to Rx470 and about 20% more than 1050ti.
 
Depends what you want to use that 1650 for. A media center or light/older games in systems without pcie power (Dell or HP) should do the trick. I see some passive models and half height also. Performance is more or less equal to Rx470 and about 20% more than 1050ti.
Staying within he brand, (no Radeon comparisons). the later GDDR6 models are alleged to be about 10% zippier than GDDR5, but the OC'd offerings do require a 6 pin PCI-E tap. They do kind of stomp the 1050s, even the ti models.

IMO, the sweet spot is the 1660 ti. I think it has about 130 more CUDA cores than the stock 1660 or 1660 Super. Assuming, of course you can manage to live your life without ray tracing or 4K gaming. (Or I suppose, the day after tomorrow's AAA titles).

Newegg was selling an Asus "Tuff" OC'd 6 GB for $230., that I managed to grab for $210. (long story). They sold out of that lot, it was back ordered, and the incoming cards they have listed at $360. That's against an original release MSRP of $280. So, at the $210 price, (again staying within the brand), $230.was a relative bargain. But $360.? No way, the card isn't worth anywhere near that.

Not sure WTF is going on there. The tariff exception for VGAs is supposed to expire., so maybe we should expect video card prices to be on the rise again, across the board

Not to be political, but after deservedly blaming Huang for the DIY ,from the factory. scalping, (capitalism at work), it was, after all, "the Donald", who started the trade war with China.
 
Last edited:
Staying within he brand, (no Radeon comparisons). the later GDDR6 models are alleged to be about 10% zippier than GDDR5, but the OC'd offerings do require a 6 pin PCI-E tap. They do kind of stomp the 1050s, even the ti models.

IMO, the sweet spot is the 1660 ti. I think it has about 130 more CUDA cores than the stock 1660 or 1660 Super. Assuming, of course you can manage to live your life without ray tracing or 4K gaming. (Or I suppose, the day after tomorrow's AAA titles).

Newegg was selling an Asus "Tuff" OC'd 6 GB for $230., that I managed to grab for $210. (long story). They sold out of that lot, it was back ordered, and the incoming cards they have listed at $360. That's against an original release MSRP of $280. So, at the $210 price, (again staying within the brand), $230.was a relative bargain. But $360.? No way, the card isn't worth anywhere near that.

Not sure WTF is going on there. The tariff exception for VGAs is supposed to expire., so maybe we should expect video card prices to be on the rise again, across the board

Not to be political, but after deservedly blaming Huang for the DIY ,from the factory. scalping, (capitalism at work), it was, after all, "the Donald", who started the trade war with China.

Who in their right mind, would buy a 4 year old GTX for $200+ ..? When the 6500xt cost less..? And for the SAME PRICE you can get a 6600 which has over 30% more performance...

I would say, that people blindly buying nVidia's GPUs and over paying, is the reason GPU prices are so high. Why would nVidia bother to lower their 4 year old GTX technology, if people are buying them instead of a 4 year newer technology.
 
Who in their right mind, would buy a 4 year old GTX for $200+ ..? When the 6500xt cost less..? And for the SAME PRICE you can get a 6600 which has over 30% more performance...

I would say, that people blindly buying nVidia's GPUs and over paying, is the reason GPU prices are so high. Why would nVidia bother to lower their 4 year old GTX technology, if people are buying them instead of a 4 year newer technology.
It's all about needs and preferences. I would get one, if I was in the market for this, since it's a good HTPC option, no separate power needed and low for factor.
That 6500xt is pcie4 x4 and will not work so well in a pcie3 board. That card makes no sense to me now, maybe later when all systems are pcie4+.
You can't blame/judge someone for having different opinion. His money, his choice.
Me I go with the best value for my money no matter the brand. I'm looking for an GPU upgrade and until now no option for me but wait better days.
 
Who in their right mind, would buy a 4 year old GTX for $200+ ..? When the 6500xt cost less..? And for the SAME PRICE you can get a 6600 which has over 30% more performance...

I would say, that people blindly buying nVidia's GPUs and over paying, is the reason GPU prices are so high. Why would nVidia bother to lower their 4 year old GTX technology, if people are buying them instead of a 4 year newer technology.
I know right? I'd probably get beat up in a bar or pub for telling someone I paid that much money for that card.
 
Back