Nvidia RTX 3060 loses its crown as Valve's Steam survey returns to normal

[midian182]

What just happened? Did something go wrong in March's Steam survey? The number of significant changes that appeared two months ago suggested something wasn't right, but April saw the status quo restored with a returning top GPU and language while Windows 11 regained its lost share.

In March, the RTX 3060 saw an unusual 6.31% increase among Steam survey participants to take the top spot in the GPU chart, replacing the GTX 1650, which had held the number one position since last November. There were other unusual jumps, too, such as the RTX 2060 suddenly rising 3.42%.

It's unclear what caused the sudden change in March, but whatever it was, Valve seems to have addressed it. The most recent April results put the GTX 1650 back on top thanks to its 2.15% increase, while the GTX 1060 returned to second place. Elsewhere, the RTX 3060 and RTX 2060 saw losses of -6.01% and -3.6%, respectively. But one thing that hasn't changed is the absence of the AMD Radeon RX 7900 XT and XTX from the main GPU table.

April Steam survey GPU chart

Looking at the month's biggest gains, the GTX 1650 is top here, too. The best-performing Lovelace product is the RTX 4060 laptop GPU (a new entry), though all RTX 4000-series cards on the list made gains.

April Steam survey top-performing GPUs

Intel also saw its previous month's massive 9.01% increase over AMD in the CPU section wiped out in April. Chipzilla fell -9.04% last month, pushing team red's CPU share to 32.84%, the highest since January.

The massive 25.3% rise that saw Simplified Chinese become the top language on Steam was also wiped out last month, pushing English to the top spot again. Windows 10 64-bit saw its March gains (11.62%) turn into April losses (-12.74%) as Windows 11 closed the gap on its predecessor to 27.82% thanks to the former's 10.98% rise.

Other categories' unusually large changes seen in March were reset in April, including total hard drive space and number of physical CPUs

It was speculated that the drastic changes in March were due to more people taking part in the Steam survey, which is optional, though that wouldn't explain why the numbers returned to normal a month later. I've reached out to Valve and will update the article if the company responds.

Permalink to story.

https://www.techspot.com/news/98521-nvidia-rtx-3060-loses-crown-valve-steam-survey.html

 

[DSirius]

This survey is a gold mine for market and data research.
It can help us making a clear picture about diverse topics regarding computer hardware like processors, videocards, OS etc.

And here we can draw some astonishing conclusions about DLSS and FSR.

Steam Hardware & Software Survey

DLSS2 is present in 32.06% - average to 32%. Being exclusive to only RTX 2-3-4XXX cards.
AMD FSR is present in 64.12%. Average 64% from Steam survey videocards.

Here AMD FSR 28.26% are from Nvidia GTX 1XXX cards which are the most relevant.
Because Nvidia GTX do not support DLSS2, Nvidia GTX cards support only AMD FSR.
Another AMD FSR 1.98% are from AMD Radeon cards. Though I found more rows named AMD Radeon graphics, so the percentage may be higher.

Thus, in a direct competition, AMD FSR vs DLSS2 we have:
AMD FSR can count for 30.25% - from AMD Radeon gpus and GTX 1XXX gpus.
DLSS2 can count for 32.06%, RTX 2-3-4XXX cards.

And now let's draw some important conclusions.
On all Nvidia 1-2-3-4XXX videocards, DLSS2 is present in almost half of them - 56% average.
AMD FSR is present and working on almost half - 44% average - which are GTX 1XXX cards.
If we take into consideration that AMD FSR works, or can be implemented also on all RTX 2-3-4XXX cards, the percentage can be 100%. Just that Nvidia RTX 2-3-4xxx users will choose DLSS2 not FSR for obvious reasons, superior implementation as was supposed to be played.

Golden question for chat forum and TechSpot editors too:
Which is "better" in general? AMD FSR or DLSS?
And here we can include the most 2 important factors: availability and quality.
Availability is crucial because a technology which works has the "best" quality against another technology which does not work at all for 50% of them, not being "supported" by Nvidia.*
I would say that in this situation the most relevant factors are:
1. Availability; 2. Quality; 3. Performance; 4. Price. We can add and 5. Consumer friendly scale. (Including how easy and friendly is to implement).

AMD managed to bring a technology, FSR, which works on all videocards including competition while Nvidia FAILED to give to their own customers a technology, DLSS, which could work on the SAME Nvidia cards on which AMD FSR is available and works.
And AMD FSR is working on PS5 and XBOX-X too if we're looking for % in videogame market. AMDFSR has a good, decent quality. Nvidia DLSS is slightly better, some would say that is superior. Performance is almost on par on both AMD FSR and DLSS2, some will say that is slightly better for DLSS. AMD FSR price is close to free. DLSS price is high for DLSS2, some would say that is too expensive.

Can we call, or is it fair to say AMD FSR is better? Or can we call or is it fair to say DLSS2 is better? I invite everybody, Nvidia, AMD and Intel users, to respond.

* Oh, we can make a disclaimer at the bottom of the page stating the paradox that it is obvious that NVIDIA could have made DLSS 2 work on GTX 1xxx (or DLSS3 work with Ampere and Turing). NVIDIA didn't make it available because of market segmentation.

But I think that this disclaimer will only worsen NVIDIA DLSS position.
For those who truly believe that NVIDIA couldn't, this also means that AMD is more technically advanced for being able to make it work not only on AMD, but on NVIDIA and Intel hardware too across all their generations.

And unfortunately, here we have some reviews, which despite having this golden mine data, despite investing many hours in hard quality work, in the end, they find difficult or some of them fail to draw basic and common sense conclusions.

This I pointed out in my previous posts, let's make a clear picture and do not let us becoming impaired, by any corporation PR, to the point of not seeing the real PATH TRACING (or RAY TRACING) in the forest (video gaming market) because of some shiny but dry branches.
(Pun intended)

In 2-3 years DLSS technology may have the same dying fate as Nvidia GSync.
Gsync was another Nvidia technology which claimed, through Nvidia PR, and their army of "reviewers", as having "stratospheric" superiority against AMD FreeSync. In the end Gsync had an embarrassing death (while being expensive during it's agony) and AMD FreeSync was and is the winner and prevailed as the "better" or "superior" technology. The same dying fate may outlines for DLSS too, because of Nvidia's same fatal flaws.

Let's make things strait, do not get me wrong, DLSS is a great technology, like Gsync was, but both have fatal flaws. It is an Nvidia closed proprietary technology instead of being open or free, and is too expensive while not worthing the money vs AMD FSR competition.
As a good thing, Nvidia DLSS2 can be "slightly" better than AMD FSR only when it works, and from this data, it is working on only 50% - half of Nvidia cards.
Anybody relate to another DLSS market segmentation? DLSS2 works only on RTX2-3XXX, DLSS3 only on 4XXX. Next will be DLSS4 only on RTX 5 or 6 XXXX cards?

The future looks confusing for Nvidia gaming customers, in too many darky shadows for playing games with Nvidia DLSS, Ray and Path Tracing. (Pun intended).

Nvidia, make DLSS open, and lower the price! Doing so you can "win" this competition against AMD FSR, and all customers, both Nvidia and AMD will be pleased!

Until than, half of Nvidia gamers are pleased by DLSS while the other half are alienated by the same Nvidia lack of their own DLSS support, but are pleased by AMD FSR.
How embarassing or foolish is that for Nvidia?
Because, for intelligent or informed customers, it is evident how ANTICONSUMER Nvidia is, regardless if they are Nvidia or AMD users or Intel.
(I hope that Intel, or others will manage to compete in the future, in videocard market, so we can have a better and real competition.)

P.S. And here we are, I think I wrote, or at least my post can be considered like, a new starting point for an interesting TechSpot article and it is open available on TechSpot, my favorite techsite on the Citadel

 

[takaozo]

Maybe the data is from last summer, no 4000 or 7000 series here now .

Yes a real value this survey.

 

[DSirius]

Maybe the data is from last summer, no 4000 or 7000 series here now .

Yes a real value this survey.

It is from April 2023. I extracted the data from Steam survey webpage.
https://store.steampowered.com/hwsurvey/videocard/?sort=pct
You can find 4090 and all other 4xxx cards there.
AMD 7XXX cards, having small percentage, were included in AMD FSR 1.98% which are from all AMD Radeon cards. Though I found more rows named AMD Radeon graphics, so the percentage may be higher.

 

[toooooot]

That must have been a mistake. There got to be a lot of people who can barely afford a lowest end gpu that just barely beats Intel's iGPu.

 

[Lew Zealand]

So the numbers are back to being somewhat in line with 2 months ago. Which are only relevant with error bars of what, about 25%? Maybe one standard deviation? More uselessness.

 

[B Double]

There have been 3 gaming computers in my house for years. All running AMD hardware. I've only ever been asked to participate in the steam survey once. That's the problem is it relies on people opting in. Making it very flawed and potentially biased, which if it is would make it far less relevant to data purchasers, possibly even misleading. Do they reveal the source code or the selection process???

 

[B Double]

This survey is a gold mine for market and data research.
It can help us making a clear picture about diverse topics regarding computer hardware like processors, videocards, OS etc.

And here we can draw some astonishing conclusions about DLSS and FSR.

Steam Hardware & Software Survey

DLSS2 is present in 32.06% - average to 32%. Being exclusive to only RTX 2-3-4XXX cards.
AMD FSR is present in 64.12%. Average 64%.

Here AMD FSR 28.26% are from Nvidia GTX 1XXX cards which are the most relevant.
Because Nvidia GTX do not support DLSS2, Nvidia GTX cards support only AMD FSR.
Another AMD FSR 1.98% are from AMD Radeon cards. Though I found more rows named AMD Radeon graphics, so the percentage may be higher.

Thus, in a direct competition, AMD FSR vs DLSS2 we have:
AMD FSR is present on 30.25% - from AMD Radeon gpus and GTX 1XXX gpus.
DLSS2 is present in 32.06%.

And now let's draw some important conclusions.
On all Nvidia 1-2-3-4XXX videocards, DLSS2 is present in almost half of them - 56% average.
AMD FSR is present and working on almost half - 44% average - which are GTX 1XXX cards.

Golden question for chat forum and TechSpot editors too:
Which is "better" in general? AMD FSR or DLSS?
And here we can include the most 2 important factors: availability and quality.
Availability is crucial because a technology which works has the "best" quality against another technology which does not work at all, or works only for half of them.*
I would say that in this situation the most relevant factors are:
1. Availability; 2. Quality; 3. Performance; 4. Price. We can add and 5. Consumer friendly scale. (Including how easy and friendly is to implement).

AMD managed to bring a technology, FSR, which works on all videocards including competition while NVIDIA was NOT ABLE to give to their own customers a technology, DLSS, which could work on the SAME Nvidia cards on which AMD FSR is available.
And AMD FSR is working on PS5 and XBOX-X too if we're looking for % in videogame market. AMDFSR has a good, decent quality. Nvidia DLSS is slightly better. Performance is almost on par on both AMD FSR and DLSS2. AMD FSR price is close to free. DLSS price is high for DLSS2, some would say that is too expensive.

Can we call, or is it fair to say AMD FSR is better? Or can we call or is it fair to say DLSS2 is better? I invite everybody Nvidia, AMD and Intel users to respond.

* Oh, we can make a disclaimer here at the bottom of the page stating the paradox that it is obvious that NVIDIA could have made DLSS 2 work on GTX 1xxx (or DLSS3 work with Ampere and Turing). NVIDIA didn't make it available because of market segmentation.

But I think that this disclaimer will only worsen NVIDIA DLSS position.
For those who truly believe that NVIDIA couldn't, this also means that AMD is more technically advanced for being able to make it work not only on AMD, but on NVIDIA and Intel hardware too across all their generations.

And unfortunately, here we have some reviewers which despite having this golden mine data, despite investing many hours in hard quality work, in the end they find difficult or some of them are not able to draw some common sense conclusions.

This I pointed out in my previous posts, let's make a clear picture and do not let us becoming impaired to the point of not seeing the real PATH TRACING (or RAY TRACING) in the forest (video gaming market) because of some shiny but dry branches. (Pun intended)

In 2-3 years DLSS technology may have the same dying fate as GSync.
Gsync was another Nvidia technology which claimed having "stratospheric" superiority against AMD FreeSync. In the end Gsync had an embarrassing death (while being expensive during it's agony) and AMD FreeSync was and is the winner. and The same dying fate may outlines for DLSS too.

Let's make things strait, do not get me wrong, DLSS is a good technology, like Gsync was, but both have fatal flaws. It is an Nvidia closed proprietary technology instead of being free, and is too expensive while not worthing the money vs AMD FSR competition.
As a good thing, Nvidia DLSS2 can be "slightly" better than AMD FSR only when it works, and from this data, it is working on only 50% - half of Nvidia cards.
Anybody relate to another DLSS market segmentation? DLSS2 works only on RTX2-3XXX, DLSS3 only on 4XXX, next will be DLSS4 only on RTX 5 or 6 XXXX cards?

The future looks confusing, in too many darky shadows for Nvidia DLSS, Ray and Path Tracing gaming customers. (Pun intended).

Nvidia, make DLSS open, and lower the price! Doing so you can win this fight against AMD FSR and all customers, both Nvidia and AMD will be pleased!

Until than, half of Nvidia gamers are pleased by DLSS while the other half are alienated by the same Nvidia lack of their own DLSS support, but are pleased by AMD FSR.
How embarassing or foolish is that for Nvidia?
Because for intelligent or informed customers it is evident how ANTICONSUMER is, regardless if they are Nvidia or AMD users or Intel.
(I hope that Intel, or others will manage to compete in the future in videocrad market so we can have a better and real competition.)

P.S. And here we are, I think I wrote, or at least my post can be considered like, a new starting point for an interesting TechSpot article and it is open available on TechSpot, my favorite techsite on the Citadel

Great work but you left out the fact that the survey is opt in. I've been a steam gamer for years. We have 3 all AMD rigs in the house and only ever been asked to participate once. So the selection process is open to bias not to mention being inaccurate to useless depending on how that selection of a user is made. Is that data transparent? I don't think so.

 

[Kam7r]

Great work but you left out the fact that the survey is opt in. I've been a steam gamer for years. We have 3 all AMD rigs in the house and only ever been asked to participate once. So the selection process is open to bias not to mention being inaccurate to useless depending on how that selection of a user is made. Is that data transparent? I don't think so.

it's nothing new, this survey is a pile of BS, and I don't even know why it's being reported... oh wait... I clicked on this news...

 

[AnilD]

Great work but you left out the fact that the survey is opt in. I've been a steam gamer for years. We have 3 all AMD rigs in the house and only ever been asked to participate once. So the selection process is open to bias not to mention being inaccurate to useless depending on how that selection of a user is made. Is that data transparent? I don't think so.

So that's how statistics and surveys work. You take a representative sample out of the population. In this case, the Steam survey should be representative of hardware used by PC gamers. I see no conflict of interest from Valve to mess up the stats numbers one way or the other. There's likely some randomness of who gets opted-in based on geographical data and usage.

 

[Inkwe]

So that's how statistics and surveys work. You take a representative sample out of the population. In this case, the Steam survey should be representative of hardware used by PC gamers. I see no conflict of interest from Valve to mess up the stats numbers one way or the other. There's likely some randomness of who gets opted-in based on geographical data and usage.

That's the thing. It's evident that Valve's sample size is too small in the wild variances we see on these surveys. You'd have to survey a very large portion of users given the amount of differing hardware configs possible. Like others have pointed out, I get asked to opt into the survey only once in a blue moon. And then only from the PC I'm currently sitting at: two PCs in different rooms, for different use cases and with large changes in capability in my case. So depending on when (or even if) Valve asks me, I either have a Ryzen 5 1600, RX580 and 1080p 60Hz screen or a Ryzen 5 5600, 1080ti and QHD Ultrawide at 144Hz.

 

[b3rdm4n]

Can we call, or is it fair to say AMD FSR is better? Or can we call or is it fair to say DLSS2 is better? I invite everybody, Nvidia, AMD and Intel users, to respond.

In the here and now, it's clearly DLSS being better*, I can elaborate, but this topic has been covered many times, including recently by HUB/techspot. It has undeniably better results (*in image quality), across a wide game selection, and users of cards capable of DLSS, especially when considering in context to the steam survey, the people/cards that are capable and willing to run new AAA releases that will get FSR/DLSS etc, there is an unignorably large subset of people who can and will take advantage of it, compared to FSR. A majority even.

Naturally, this cannot and will not last forever, like I doubt DLSS will be the premier upscaling technique in 10 years time. But lets say for argument's sake FSR is destined to kill DLSS, the decline hasn't even begun yet, DLSS is still on the upswing, gaining momentum and improving in quality. FSR is too don't get me wrong, but they move somewhat in parallel and given the scale at play here, I'd wager DLSS is at least 1 if not 2+ generations from being supplanted by something that's not only better visually, but perhaps universally accessible too. I do not think FSR's path to success involves AMD sponsorships requesting or necessitating the exclusion of DLSS, get FSR into every game sure, not a problem, but don't block dev's using what is little more than a checkbox to give other users DLSS. Do your best to increase the quality and adoption, and let the tech stand on it's merits against competing ones, blocking it, if that's what has happened even once, is anti-consumer (against the aforementioned massive subset of RTX users that, nothing to do with a hardware purchase, want to make a game purchase) and could hurt AMD and game developers by dis incentivising RTX users from buying said games.

It seems Tim and Steve agree, the part 3 QnA that HUB released less than 48 hours ago talks about this, and FSR had a chance to 'kill' DLSS and be the only one chosen in all titles because everyone can use it, turns out that sales pitch alone clearly wasn't enough 12-18+ months in, and it needs to also be as good or better IQ/Perf wise, not just acceptably close but clearly falling short. The window FSR had to do that and 'strike while the iron is hot' has passed. Here's hoping FSR 3+ (and 2.x feature set with improvements) can do it. Bear in mind Dev's barely give a crap whether something is open source or not, they use licensed middleware's, engines, and tools left and right, being accessible to everyone is their biggest strength, and that alone isn't enough to 'kill' DLSS.

* Oh, we can make a disclaimer at the bottom of the page stating the paradox that it is obvious that NVIDIA could have made DLSS 2 work on GTX 1xxx (or DLSS3 work with Ampere and Turing). NVIDIA didn't make it available because of market segmentation.

This is readily explainable when we consider the content on XeSS. XeSS has 2 codepaths, the generic 'everyone can enable it' DP4a path, and the XMX path only ARC cards can use. It is legitimately a detriment to the name XeSS to have both paths be called the same thing, testing demonstrates that the XMX path not only has superior image quality, but considerably better performance too. Yet, with such little market penetration, the vast majority of possible XeSS users would be using the DP4a path, giving more average-to-good results, rather than the very-good-to-excellent results the XMX path can give, which is a net-negative to the name XeSS. This would apply with DLSS, if it had the same name, but meh performance and meh visuals for everyone not on an RTX card, but RTX users could get better of both, it's a detriment to the brand name that is DLSS - just as lumping in FG into that namesake has become, it adds confusion and negativity.

But I think that this disclaimer will only worsen NVIDIA DLSS position.
For those who truly believe that NVIDIA couldn't, this also means that AMD is more technically advanced for being able to make it work not only on AMD, but on NVIDIA and Intel hardware too across all their generations.

I certainly believe Nvidia could, but I can understand and appreciate why they didn't, they wanted their product to be the gold standard for upscaling, so if it got worse perf and IQ on GTX cards, it quickly muddies the waters and dilutes the reception of the accelerated version.

Some facts in there, some opinions too, and hey especially here on this very forum I don't expect to have many agree with me, but that's my limited take and lay of the land on DLSS and FSR. Happy to provide sources to claims where made, if necessary.

 

[tancabean]

That's the thing. It's evident that Valve's sample size is too small in the wild variances we see on these surveys. You'd have to survey a very large portion of users given the amount of differing hardware configs possible. Like others have pointed out, I get asked to opt into the survey only once in a blue moon. And then only from the PC I'm currently sitting at: two PCs in different rooms, for different use cases and with large changes in capability in my case. So depending on when (or even if) Valve asks me, I either have a Ryzen 5 1600, RX580 and 1080p 60Hz screen or a Ryzen 5 5600, 1080ti and QHD Ultrawide at 144Hz.

The number of times that you have been asked is pretty irrelevant. It’s possible that some people are never asked in their entire lifetime. That’s how random sampling works. Given the vast size of the population Valve only needs to survey a tiny % in order to get a useful representative sample.

 

[thewelshbrummie]

Considering that the last survey article had the subheadline "not an April fools' joke" when the % changes were huge and statically unusual... it's frustrating that you chose to question the stats in this article and not the previous one, when Win 11 % apparently dropped 25%... only to regain that a month later.

It was extremely suspicious and basically obvious that something wasn't right with the previous stats and yet wasn't challenged.

"It was speculated that the drastic changes in March were due to more people taking part in the Steam survey, which is optional, though that wouldn't explain why the numbers returned to normal a month later."

Who provided that speculation? To me (and the other comments on the 3rd April article) it was obvious that the March stats weren't valid for whatever reason, be it an attempt to game the stats or Valve making a mistake. If there was speculation (and there wasn't any on Techspot), a source would be appropriate.

 

[ID10T]

Great work but you left out the fact that the survey is opt in. I've been a steam gamer for years. We have 3 all AMD rigs in the house and only ever been asked to participate once. So the selection process is open to bias not to mention being inaccurate to useless depending on how that selection of a user is made. Is that data transparent? I don't think so.

Pretty sure you are only asked after a hardware change, at least that's been the case for me since I started using Steam over a decade ago.

 

[ID10T]

Gsync sucked imo, had odd issues with frame pacing in certain titles and you'd have to disable it and lock it, I figured it was better until I bought a new monitor with FreeSync and it just works and requires zero thought or tweaking.

 

[Lew Zealand]

Pretty sure you are only asked after a hardware change, at least that's been the case for me since I started using Steam over a decade ago.

It's not even that as I change my hardware pretty frequently and have at least 10 machines with Steam installed. I should be getting the survey all the time but I don't. It's just random and not very often.

My new/old GT 635 is underrepresented!

 

[ID10T]

It's not even that as I change my hardware pretty frequently and have at least 10 machines with Steam installed. I should be getting the survey all the time but I don't. It's just random and not very often.

My new/old GT 635 is underrepresented!

Huh well that’s what I get from blanketing one sample case

 

[B Double]

Pretty sure you are only asked after a hardware change, at least that's been the case for me since I started using Steam over a decade ago.

Will see. Just dropped a 7900xt into my system as price dropped. Now that they make 4070/ti poor performance and Vram for the price. Makes the 7900xt very attractive. Oh and I didn't have to get a new Power supply.

 

[takaozo]

If premium GPU buyers do not exist, then the used GPU buyers wouldn't have what to buy.