Steam players are reverting back to GTX 1660 graphics cards, despite Nvidia's best efforts

K

kac77

2[H]4U
Joined
Dec 13, 2008
Messages
3,320
I mean it kinda makes sense. For indie games and shooters, why sacrifice raw perf for some eye candy that's only in maybe a few of your games? Save a buck and get the same or better performance, why not?
 
I would say its people needing an upgrade... but also not wanting anything under a 4090 from this gen. So if your not willing to spend $1600 today... maybe spend $160 and wait.
Its too bad more of those people where not exploring the decently priced AMD cards. AMD needs a new marketing dept.

I guess there are a bunch of nvidia owners that are glad FSR is a thing. :) lol
 
Marees said:
Interesting. These cards can't run DLSS
Click to expand...
But DLSS is the best technology ever and not a muddy looking, laggy mess at all.

Ray tracing is the even more besterest thing ever since the beginning of ever! People say it's hard to spot in most games, but those people are AMD shills, I think we all notice the framerate shitting the bed.
 
There have been massive dumps of various 1660 models on the used and refurbished markets lately. The 1660 series was very popular with Ethereum miners back in 2020 and 2021 and you can get them for peanuts right now.
The card still makes sense for a lot of 1080p games and FSR only extended its life.
 
It
Armenius said:
Because we all know that Steam's random opt-in survey is 100% representative of the AIB market.
Click to expand...
might not be perfect but it is substantial. Valve claims something like 85% of the PC gaming market and they figure almost 90% of those users opt into the surveys. So even if you assume the numbers to be lower then that it’s still a good representation of 70% or more of the PC gaming market.

Certainly not a small margin of error, but good enough that when developers are working on projects they keep it as a reference point for what their target audience is capable of running.
 
Lakados said:
It

might not be perfect but it is substantial. Valve claims something like 85% of the PC gaming market and they figure almost 90% of those users opt into the surveys. So even if you assume the numbers to be lower then that it’s still a good representation of 70% or more of the PC gaming market.

Certainly not a small margin of error, but good enough that when developers are working on projects they keep it as a reference point for what their target audience is capable of running.
Click to expand...
Right, but that 90% sample are not all taking the survey at the same point in time. I just coincidentally got my first survey request after starting Steam for the first time since 2015 after I installed my 7800X3D. An 8-year period can see a lot of hardware changes.
 
Armenius said:
Right, but that 90% sample are not all taking the survey at the same point in time. I just coincidentally got my first survey request after starting Steam for the first time since 2015 after I installed my 7800X3D. An 8-year period can see a lot of hardware changes.
Click to expand...
Once you opt-in they pole monthly.
They prompt for opt-in when they detect a hardware change.

Regardless of opt-in Valve collects some basic hardware info as part of your account security, the opt-in is for additional software and service info.
 
Armenius said:
Right, but that 90% sample are not all taking the survey at the same point in time. I just coincidentally got my first survey request after starting Steam for the first time since 2015 after I installed my 7800X3D. An 8-year period can see a lot of hardware changes.
Click to expand...
Is the survey something you do? I thought it was simply something that the Steam program reported back to the mothership periodically
 
Interesting what happened with those 3060s was 10.67% of the users in March, then July 4.88% which if they represent 10s of millions of users (or 100s of millions?) that's a huge drop off for those cards to just disappear.
 
sfsuphysics said:
Is the survey something you do? I thought it was simply something that the Steam program reported back to the mothership periodically
Click to expand...
You choose to opt in or not when it randomly asks you.

Overall it's the closest thing we have to a large random sample of the PC gaming community.

People tend to take certain stances on it depending on how convenient the results are for their opinions.
 
Dont take steams charts with a grain of salt. Do you know how many internet cafe's in the world use lower end GPU's. Then everytime someone new logs into a PC at a cafe a survey pops up asking for the hardware info.

Thats why the results are skewed heavily.
 
sfsuphysics said:
Interesting what happened with those 3060s was 10.67% of the users in March, then July 4.88% which if they represent 10s of millions of users (or 100s of millions?) that's a huge drop off for those cards to just disappear.
Click to expand...
Valve reported there was an error in their report for the data validation. So it saw some huge swings in March and early April that were corrected for May.
Apparently they updated their software and buggered a few reports in the process.
 
Brackle said:
Dont take steams charts with a grain of salt. Do you know how many internet cafe's in the world use lower end GPU's. Then everytime someone new logs into a PC at a cafe a survey pops up asking for the hardware info.

Thats why the results are skewed heavily.
Click to expand...
They are a very small fraction of the user database but yes in any survey there will be garbage data and there exists some pretty good tools out there for identifying and filtering it.
In either event you end up with a decent bell curve that is good enough.
Valve can also poll data from developers on the platform. Many games do report back detailed hardware info, you probably would not be surprised how many games at least take a dump of the dxdiag or similar data set on first run or when you connect with their various launchers.
 
kac77 said:
Steam players are reverting back to GTX 1660 graphics cards, despite Nvidia's best efforts

The Nvidia GTX 1660 released back in 2019, but Valve’s latest Steam Hardware Survey figures suggest people are still buying the graphics card. Rather than necessarily jumping from lower spec favorites like the GTX 1650 to the new RTX 4060, players appear to be sticking with the GTX line of GPUs for the most part.
Click to expand...
All about price point for the performance. 1660 is fine for most uses, esp @ 1080p. And it's affordable. I got a 1660 Super for $130 new in May.

China and internet cafe's may have an impact. But, the chart is still relative. Who else publishes such a chart?
 
wareyore said:
All about price point for the performance. 1660 is fine for most uses, esp @ 1080p. And it's affordable. I got a 1660 Super for $130 new in May.

China and internet cafe's may have an impact. But, the chart is still relative. Who else publishes such a chart?
Click to expand...
Affordable and perfect for the big esports titles: League of Legends, Dota2, Starcraft 2, Team Fortress 2, CS:S, etc.
 
wareyore said:
All about price point for the performance. 1660 is fine for most uses, esp @ 1080p. And it's affordable. I got a 1660 Super for $130 new in May.

China and internet cafe's may have an impact. But, the chart is still relative. Who else publishes such a chart?
Click to expand...
No one, because its hard to know what people are playing with at home. If you got 200-300 people logging into steam at a single Internet Cafe on 1 day. Times that by 365 days, then you can easily see how the charts are skewed.

So, lets say 250 people login to steam at 1 single internet cafe and multiply that by 365 days (just an example). That would be 91,350 for 1 single internet cafe. Now lets say there are a 1000 internet cafe's in the world. That would be over 91,250,000 people using a 1660.....When they aren't really that many.

See how the chart can be easily skewed?
 
Brackle said:
No one, because its hard to know what people are playing with at home. If you got 200-300 people logging into steam at a single Internet Cafe on 1 day. Times that by 365 days, then you can easily see how the charts are skewed.

So, lets say 250 people login to steam at 1 single internet cafe and multiply that by 365 days (just an example). That would be 91,350 for 1 single internet cafe. Now lets say there are a 1000 internet cafe's in the world. That would be over 91,250,000 people using a 1660.....When they aren't really that many.

See how the chart can be easily skewed?
Click to expand...
Is it really skewed, though? Those are real users, using that hardware for gaming. You can filter that out if you want, but it is a real, large percentage of the user base.
 
Brackle said:
No one, because its hard to know what people are playing with at home. If you got 200-300 people logging into steam at a single Internet Cafe on 1 day. Times that by 365 days, then you can easily see how the charts are skewed.

So, lets say 250 people login to steam are 1 single internet cafe and multiply that by 365 days (just an example). That would be 91,350 for 1 single internet cafe. Now lets say there are a 1000 internet cafe's in the world. That would be over 91,250,000 people using a 1660.

See how the chart can be easily skewed?
Click to expand...
And people are still playing on the internet café equipment. So, it's relevant. Same as any cloud or streaming equipment they log in with and it gets captured. I have several systems I use and I routinely test gpu's. No telling what Steam used for my account last time it asked me. I'm a single data point at that moment in time. So are all those thousands of internet café users. They are single data points at that moment in time.

The real skewing is when I see all those igpu's dominating their chart. More relevant for the real world and what they offer on their platform, for sure.

I'd love to see Steam capture more data to let us dig deeper. Like, filter for individual games, streaming or cloud services, etc.
 
The article seem a bit strange, that 3.2% figure was not special at all (in their screenshot it was way lower than the 3060-2060 around the 3070....) and a bit strange to use it will be extremely noisy (1.26% this month). And the 4060 was released june 29, new model always take a while to show up on steam, the 4060TI just appeared

A good way to see if people are still buying 1660 is looking the pre-built the big sellers sales like Walmart or bestbuy:
https://www.bestbuy.com/site/hp-ome...super-512gb-ssd-white/6535732.p?skuId=6535732
https://www.walmart.com/ip/ViprTech...-Keyboard-Mouse-Black/2930568881?from=/search
https://www.walmart.com/ip/ASUS-ROG...GTX-1660-Ti-512GB-SSD/1304837396?from=/search

It is still an option on the inspiron dell line with 13th gen cpus and I would imagine more popular in some part of the world than the US. We can imagine most sales are not from people buying new GPU at the store but via the purchase of PCs.

Brackle said:
No one, because its hard to know what people are playing with at home. If you got 200-300 people logging into steam at a single Internet Cafe on 1 day. Times that by 365 days, then you can easily see how the charts are skewed.
Click to expand...
Data would probably be better if steam would intrude like this, scan the pc and phone home about it (with the mac address they would know it is the same pc even if it is different account) but I thought they used our answer when they ask us when we want to participate, fill a survey and sent it to them ?

Nvidia did give us their Geforce Experience users numbers (which at least would be correct as long there is no a large bias for the average nvidia user that do not use that program with the type of card they own) and steam hardware survey did not look out of place at all, same for south korea discrete gpu sales numbers.
 
Last edited:
Brackle said:
No one, because its hard to know what people are playing with at home. If you got 200-300 people logging into steam at a single Internet Cafe on 1 day. Times that by 365 days, then you can easily see how the charts are skewed.

So, lets say 250 people login to steam at 1 single internet cafe and multiply that by 365 days (just an example). That would be 91,350 for 1 single internet cafe. Now lets say there are a 1000 internet cafe's in the world. That would be over 91,250,000 people using a 1660.....When they aren't really that many.

See how the chart can be easily skewed?
Click to expand...
I would sure hope someone like Steam would record hardware ID's and remove duplicates to prevent this.

I game on a 1660ti and a 2070. 1080p and without all the acronym nonsense (DLSS, FSR, RT, blah blah} to just make the game slightly more pretty has no appeal to me. That and I refuse to ever spend more than $200 on a card. I don't opt in to the survey so I'm one case not counted. Just in my "circle" or players, I am not alone in this. To those who spend $1500 on a GPU and PSU, more power to ya, not for me.
 
That_Sound_Guy said:
I would sure hope someone like Steam would record hardware ID's and remove duplicates to prevent this.

I game on a 1660ti and a 2070. 1080p and without all the acronym nonsense (DLSS, FSR, RT, blah blah} to just make the game slightly more pretty has no appeal to me. That and I refuse to ever spend more than $200 on a card. I don't opt in to the survey so I'm one case not counted. Just in my "circle" or players, I am not alone in this. To those who spend $1500 on a GPU and PSU, more power to ya, not for me.
Click to expand...
Since Valve has never come out to say how it compiles the statistics, it is totally possible that it does not go by hardware ID's.

But, we have no way of knowing. That is why you cannot use the steam hardware survey results as proof that 1 GPU is used more than another.
 
Brackle said:
Since Valve has never come out to say how it compiles the statistics, it is totally possible that it does not go by hardware ID's.
Click to expand...
Don't they say that they ask a random subset every month and those who accept send back to them a report of their machine hardware ?

Do you mean if they and what the correction algorithm used to take account say people with nice hardware are more likely to want to share it or something ?
 
Brackle said:
No one, because its hard to know what people are playing with at home. If you got 200-300 people logging into steam at a single Internet Cafe on 1 day. Times that by 365 days, then you can easily see how the charts are skewed.

So, lets say 250 people login to steam at 1 single internet cafe and multiply that by 365 days (just an example). That would be 91,350 for 1 single internet cafe. Now lets say there are a 1000 internet cafe's in the world. That would be over 91,250,000 people using a 1660.....When they aren't really that many.

See how the chart can be easily skewed?
Click to expand...
I can almost guarantee Valve filters those result out. If they have an IP with 600 user accounts logging in only during business hours for 365 days a year in countries where LAN cafes are popular it’s fairly obvious to anyone to remove those data.

EDIT: also as pointed out, Hardware IDs among many other identifying features.
 
Brackle said:
No one, because its hard to know what people are playing with at home. If you got 200-300 people logging into steam at a single Internet Cafe on 1 day. Times that by 365 days, then you can easily see how the charts are skewed.

So, lets say 250 people login to steam at 1 single internet cafe and multiply that by 365 days (just an example). That would be 91,350 for 1 single internet cafe. Now lets say there are a 1000 internet cafe's in the world. That would be over 91,250,000 people using a 1660.....When they aren't really that many.

See how the chart can be easily skewed?
Click to expand...
How about when the GTX 1060 rules Steam as #1 since 2016? Is it hard to believe that the majority of GPU buyers are interested in cheap GPU's and couldn't care less for Ray-Tracing and DLSS? This isn't anything new as most people buy $200 and under graphic cards, and right now a GTX 1660 is less than $200 new, and above $100 used on Ebay. What's more interesting from Steams current Hardware survey is that Intel's Xe is above AMD by quite a bit, which their most popular graphics card is well Radeon graphics and then the RX 580. Then it's Vega 8 which means laptops, and then the 5700 XT, probably because those are cheap like slightly above $100 cheap. Even if we are to ignore the GTX 1660, it's not like other cheap Nvidia cards aren't at the top. GTX 1650, 3060, 2060, and 1050 Ti's are all right up there. I think it's safe to say that Nvidia has priced themselves right out of the market, and if they hope things will change by the time they release the RTX 50 series, then good luck. AMD has very much priced themselves out of the market, as nobody is even thinking about buying their GPU's. You know it's a problem when Intel's Xe is beating AMD by a significant amount.

Another thing to point out is that Linux has surpassed Mac OSX on Steam.
 
DukenukemX said:
What's more interesting from Steams current Hardware survey is that Intel's Xe is above AMD by quite
Click to expand...
I am not sure what the sentence above AMD mean ?

Intel Xe graphics (which also I am not sure how many different models it include of how many laptops) seem below AMD Radeon Graphics (which I am not sure either), Intel does not seem to have a discrete gpu card to make the list yet.

DukenukemX said:
used on Ebay
Click to expand...
As a general rule every used card sold on ebay mean a new more expensive one was bought by that sellers and were bought at an high price back in the days at a 1:1 ratio.


DukenukemX said:
This isn't anything new as most people buy $200 and under graphic cards
Click to expand...
Yet most of the Top 10 models were probably sold above that, 3060, 2060, 3060ti, 3070, 1660 super, 3050, for part of the 1060-1650 life time it got up there I had to pay mine a 1060 $420 + tax canadian for the cheapest Zotac model, it got to $300 USD, there some bias as more expensive card get pass around in gaming machine instead of aiming in your NAS or trash, but if you look at top 20 most popular models you will not see that many $200 or under sku (eveb the 1660 super or Ti were they ever at those price ?)

There was literal digital 8-14 month wait lines for $500 and $700 GPU not so long ago, selling 10 of millions of them to gamers.
 
Last edited:
LukeTbk said:
I am not sure what the sentence above AMD mean ?

Intel Xe graphics (which also I am not sure how many different models it include of how many laptops) seem below AMD Radeon Graphics (which I am not sure either), Intel does not seem to have a discrete gpu card to make the list yet.
Click to expand...
I'm not sure what Radeon graphics is either, but Vega 8 is what you see in laptops. My guess is that Radeon graphics is from older AMD built in graphics. Something like an AMD A6 APU would probably just show up as Radeon graphics for Steam. The Intel Xe graphics are also probably from laptops. It still doesn't look good on AMD.
As a general rule every used card sold on ebay mean a new more expensive one was bought by that sellers and were bought at an high price back in the days at a 1:1 ratio.
Click to expand...
The Ryzen 5700X that I bought off Ebay was from a seller who claims he decided to get an Xbox to replace his gaming PC. People could have sold their desktop for parts to buy a laptop as well, or just have no money and just decided to sell their PC. Could also be that these GPU's were used for mining and they're left over mining cards.
Yet most of the Top 10 models were probably sold above that, 3060, 2060, 3060ti, 3070, 1660 super, 3050, for part of the 1060-1650 life time it got up there I had to pay mine a 1060 $420 + tax canadian for the cheapest Zotac model, it got to $300 USD,
Click to expand...
The 1060 you bought at that price would have been around 2016-2017 when the crypto market boomed and GPU prices soared. The RTX 2060 was so popular during the pandemic that Nvidia had to start making more of them, probably due to how much cheaper these were compared to the RTX-30 series. Right now a lot of these GPU's are probably coming from someone's crypto farm they had in their basement.
there some bias as more expensive card get pass around in gaming machine instead of aiming in your NAS or trash, but if you look at top 20 most popular models you will not see that many $200 or under sku (eveb the 1660 super or Ti were they ever at those price ?)
Click to expand...
Regardless the most popular GPU's are still cheap GPU's, and most of those are much older GPU's from a generation or three ago. If people are buying current generation GPU's then why is Nvidia stopping production?

There was literal digital 8-14 month wait lines for $500 and $700 GPU not so long ago, selling 10 of millions of them to gamers.
Click to expand...
Not sure what "not so long ago" is to you but the demand for GPU's crashed last year. We are still dealing with that crash as AMD and Nvidia both refuse to bring prices back down to affordable levels. Nvidia is betting the farm on AI while AMD is waiting to see if they should do the same. BTW the 7900 XT is now $750, which is a lot cheaper than at release and still overpriced.
 
I'm pretty sure with such a weak card these GTX 1660 owners are not buying

Ratchet & Clank Rift Apart at $59.99 on Steam, so why the BS chart?​

 
1_rick said:
"Radeon Graphics" is how the 4900hs in my laptop shows up.
Click to expand...
Yeah, I am pretty sure that Radeon Graphics is basically Vega and below.
The RX580 was a super popular mining card and I would bet that they are popping up as refurbished all over right alongside all those 1660s
 
DukenukemX said:
Regardless the most popular GPU's are still cheap GPU's, and most of those are much older GPU's from a generation or three ago. If people are buying current generation GPU's then why is Nvidia stopping production?
Click to expand...
Nvidia and AMD have pretty much always stopped their consumer card production around now so they can deal with the Q1 Enterprise demand and let consumer stocks dwindle so they are hopefully nearly gone for January, because consumer spending drops off a cliff between Jan and April during that time they will have announced their refresh cards and then why would people be wanting to buy those when a refresh and price drops are right around the corner. This is normal inventory management, and this video of "Nvidia cutting production to manipulate the market" gets posted every year by somebody, and every time it is the same thing, a big nothing, and business as usual.
 
DukenukemX said:
The 1060 you bought at that price would have been around 2016-2017 when the crypto market boomed and GPU prices soared. The RTX 2060 was so popular during the pandemic that Nvidia had to start making more of them, probably due to how much cheaper these were compared to the RTX-30 series. Right now a lot of these GPU's are probably coming from someone's crypto farm they had in their basement.
Click to expand...
And were sold much more than $200 during the pandemic
DukenukemX said:
Regardless the most popular GPU's are still cheap GPU's
Click to expand...
In since 2016 cheaper rather cheap could convey more what it is meant but yes same for tvs, cars, cpus, couch, dining table, monitors, flashlight.... obviously ? With iGPU getting quite good, cheap new GPU virtually does not exist anymore.

DukenukemX said:
If people are buying current generation GPU's then why is Nvidia stopping production?
Click to expand...
The rumors is shifting production to Hopper as much as they can, I am not sure they stopped the yet to release 4050 production and it is from a distributor saying not sure will get many 4080-4090 skus in the future (from the part I went in the video), those are nearly a year old rather niche product by now and for the demands level of when they planned to make them I can see having a lot of them to go throught before a 2024 refresh with not enough Hopper going out. The price points of all the above 4060 do not scream wanting to sales that much of them.

DukenukemX said:
Not sure what "not so long ago" is to you but the demand for GPU's crashed last year.
Click to expand...
As recently has 2020, make a 5080 that 65% stronger than the 4080 with 20gb of vram at $750 and a 5070 than is 55% stronger than a 4070 with 16gb at $550 a 5060 that 70% stronger with 12GB at $330 and we will see how no one care about GPUs above $200.
 
Last edited:
crazycrave said:
I'm pretty sure with such a weak card these GTX 1660 owners are not buying

Ratchet & Clank Rift Apart at $59.99 on Steam, so why the BS chart?​

Click to expand...
Or : https://www.vg247.com/remnant-2-steam-top-seller-peak-concurrent-record

Remnant 2 has officially graduated from a niche game to a big deal. The game is now Steam's second-best selling game, beating every other game with the exception of Counter-Strike: Global Offensive.

It need a 6900xt or better to play at 1080p max setting at a low fps of 60
 
LukeTbk said:
The game is now Steam's second-best selling game
Click to expand...
Honestly doesn't mean much because things get knocked around on that list all the time. If we look at just number of reviews or thread activity it's really not super hot. Overall sales for a certain period of time would be more telling.
 
You must log in or register to reply here.
Back
Top