NVIDIA GeForce RTX 50-series GPUs Make a Dent in Latest Steam Hardware Survey

[Visible Noise]

people really buying a 5070 over a 9070...

Ummm, yes?

For a given budget people generally buy the better product.

 

[wolf]

I'd have thought for sure the 9070 series would be visible here already, I guess we keep waiting.

 

[oxrufiioxo]

I'd have thought for sure the 9070 series would be visible here already, I guess we keep waiting.

Bruh, people who buy AMD cards obviously don't game with their computers.... I mean they only use the Xbox app and Epic games app....

Ummm, yes?

For a given budget people generally buy the better product.

I think they are around the same price now but for a while the 9070 was 80-100 usd more which there is no way in hell the average consumer is going to spend that much more on an AMD gpu.

 

[DaemonForce]

Terrible. Try sorting your data.

That's some fair movement for an imaginary percentage of total volume in laptop and desktop but good.
My RX580 (-0.04%) is still slowly losing ground along with the 7900 XTX (0.01%).
Until we figure out the meaning of AMD Radeon(TM) Graphics, the 9070/9060 stuff is probably going to remain MIA for a while.
This is some very weird visibility.

 

[Sempron Guy]

But we thought if AMD would just price their cards competitively as actively emphasized prior to release of the RX 9000 series by tech tubers like Linus, JayzTwoCents and both Steves from Hardware Unboxed and GamerNexus then people would go for the RX9000 series. I guess they are wrong for 100th time as history says even decades ago where ATI would release better cards (price/performance) people would still go for geforce cards. That has always been the trend ever since.

 

[Divide Overflow]

But WHY?!

 

[Visible Noise]

But WHY?!

Because people want the product is the most logical answer to your question.

That, and without MSI they are essentially shut out of the OEM market. MSI dropping AMD had the effect some people were hoping EVGA getting out of the GPU business would have on Nvidia.

Actually I’m in a bad mood today, I’m going to reread those for a good laugh.

 

[yfn_ratchet]

The 9070 series makes an appearance in the Linux-Only charts, though lumped together into one entry. Given the total lack of presence on the Windows/all-OS charts and the shaky amount of representation on Linux, I'd assume that they're not being reported accurately at all. Nevermind that people who neglect to participate in the survey are a completely dark variable... we don't even know the turnout rate.

 

[TumbleGeorge]

Don't you think that the data for Nvidia 50* series cards on Steam is exaggerated because Nvidia fans are more likely to brag than AMD owners?

 

[Visible Noise]

Don't you think that the data for Nvidia 50* series cards on Steam is exaggerated because Nvidia fans are more likely to brag than AMD owners?

Facepalm. The Steam Hardware Survey doesn’t survey people. It inventories their hardware.

The 9070 series makes an appearance in the Linux-Only charts, though lumped together into one entry. Given the total lack of presence on the Windows/all-OS charts and the shaky amount of representation on Linux, I'd assume that they're not being reported accurately at all. Nevermind that people who neglect to participate in the survey are a completely dark variable... we don't even know the turnout rate.

So a survey that is randomly sampled from a base of well over 100 million systems is somehow biased against AMD. Because reasons.

 

[scooze]

I would expect the 5070ti to be the one leading the chart. It's kind of the "value" king of the lineup and this generation in general.

Much less valuable than 5070, +25% performance, +50% cost, terrible value

 

[Hecate91]

The 9070 series makes an appearance in the Linux-Only charts, though lumped together into one entry. Given the total lack of presence on the Windows/all-OS charts and the shaky amount of representation on Linux, I'd assume that they're not being reported accurately at all. Nevermind that people who neglect to participate in the survey are a completely dark variable... we don't even know the turnout rate.

Exactly, because there aren't any stats on how many people are polled, who knows how accurate the data is since 9070 & 9070XT only shows up in Linux charts, and the survey is randomized. IMO, the Steam hardware is greatly exaggerated when there isn't enough data. Valve could easily list stats of amount of users polled since they know the hardware configs of everyone using Steam.

Because people want the product is the most logical answer to your question.

The average consumer buying a gaming PC doesn't know what they want besides seeing Nvidia in the specs list, Nvidia has marketing and mindshare.

That, and without MSI they are essentially shut out of the OEM market.

Sure, if you ignore Gigabyte, Asus, Sapphire, XFX, and Asrock. MSI isn't the only supplier for gaming PC's.

 

[londiste]

I still would like to know what these entries mean under the GPU section:

AMD Radeon Graphics 2.05%
AMD Radeon(TM) Graphics 1.99%
AMD AMD Custom GPU 0405 0.45%
AMD Radeon Vega 8 Graphics 0.4%
AMD Radeon Graphics (RADV VANGOGH) 0.35%
AMD Radeon (TM) Graphics 0.29% (second one)
AMD Radeon Vega 3 Graphics 0.23%
Other 9.45%

Some of the above I can guess but the Steam Survey seems to be unable to identify some AMD graphics correctly.

AMD Custom GPU 0405 is one of Steam Deck drivers, a custom one if I remember correctly. RADV VANGOGH is also one of Steam Deck drivers.
Vega 8 and Vega 3 are self-explainatory, Vega 8 is in 9-10 different APUs, same for Vega 3 but that one is in even more APUs.

You are right about the rest of them, kinda. It has been a long-running problem. Steam HW Survey takes what the driver says its name is and does not do any deeper detection. And AMD drivers have across time returned nondescriptive names. The same Radeon Graphics rows could basically be anything AMD.

Exactly, because there aren't any stats on how many people are polled, who knows how accurate the data is since 9070 & 9070XT only shows up in Linux charts, and the survey is randomized. IMO, the Steam hardware is greatly exaggerated when there isn't enough data. Valve could easily list stats of amount of users polled since they know the hardware configs of everyone using Steam.

What makes you say it is exaggerated or not enough data? Statistically the amount of responses you need for a representative sample is not that big. Depending on what you expect it might be surprisingly small. Statistics is like that. Anecdotally, I seem to get HW Survery often enough with all the different systems I happen to run Steam on.

Didn't Valve say they did the survey for their own purposes. Which kind of implies that it should have decent enough data, as well.

 

[nguyen]

Looks like the 5060 and 5070 are the best sellers for RTX5000 series since Nvidia was able to keep these closer to their MSRP across various regions.

 

[Vayra86]

All those AMD GPUs are all iGPUs and APUs, mostly from laptops. "AMD Custom GPU 0405" and VANGOGH are the Steam Deck OLED and base Steam Deck APUs, respectively.

"Other" is of course the catch-all for all GPUs not popular enough to be listed sperately.

So, the paltry amount of RX 9000 GPUs that have been sold would be grouped under "other."

As an interesting aside, if you look at the Linux-specific numbers, you find that the Steam Deck models comprise 40% of all Steam users on Linux. Not very encouraging for Linux adoption when 40% of your userbase is on an extremely niche handheld.

There is no linux adoption, there are just devices with an OS on it, let's be real here.

The vast majority is never going to install Linux by themselves, they never did that for Windows either.

"Everything except mindfactory.de sales numbers are a hoax." Sure, buddy, sure. You go have fun with Frame Smear Rendering FidelityFX Super Resolution, the big boys are using their real GPUs.

Well its pretty clear Steam's data are... puzzling to say the least. Part of having a trustworthy dataset is having a full explanation / transparency of how the numbers are generated. We don't have that and the Survey is not geared to provide that either.

Mindfactory.de sales say about as little as the Steam Survey. You can, at best, spot a market-wide trend from the Survery if you zoom out over periods of 10 years. If you zoom in, you're just wasting time thinking you're being smart with data because you're reading a list of numbers. You don't know what you're looking at.

 

[W1zzard]

Until we figure out the meaning of AMD Radeon(TM) Graphics, the 9070/9060 stuff is probably going to remain MIA for a while.

Those are the IGPs. Steam takes the name straight from Windows Device Manager, same as GPU-Z for the first row entry in GPU-Z.

9070 XT is "AMD Radeon RX 9070 XT", so it would be easily visible in the table

https://imgur.com/FfP7VXs

 

[Selaya]

Steam Hardware & Software Survey

protip.

 

[DaemonForce]

You're clicking on the directx table to find it?
I can see the vanilla 9070 and a GRE version in there but the 9070 XT is gone and the 9060 is missing. That's crazy.

 

[dyonoctis]

Well its pretty clear Steam's data are... puzzling to say the least. Part of having a trustworthy dataset is having a full explanation / transparency of how the numbers are generated. We don't have that and the Survey is not geared to provide that either.

Mindfactory.de sales say about as little as the Steam Survey. You can, at best, spot a market-wide trend from the Survery if you zoom out over periods of 10 years. If you zoom in, you're just wasting time thinking you're being smart with data because you're reading a list of numbers. You don't know what you're looking at.

The only way to get a correct set of data would be mandatory telemetry about the hardware being used, along with another set of data that would avoid the same computer from being counted twice (more on that later) but nobody would like that (?). I don't think that Valve is somehow avoiding to report Radeon users, it's just it's an opt-in survey... that wasn't really designed to represent marketshare to begin with. The survey prompt seems to be random; a hardware change won't always trigger it. You can trigger it manually, but that's an effort that many will not do.

The other issue is that valve had to tweak their algorithm because computers where several people might log into a different Steam session could trigger the survey. So a single configuration could be reported several times and inflate the numbers

Imagine if you had 10 of your best friends over to play Steam games on your PC. Every single one of them logged in to their own accounts -- which led to 10 copies of your system config being uploaded to the platform and counted as separate submissions.
That's basically what happened with Steam. And according to AMD, while the company made some corrections to its data, Valve has never been particularly concerned with making sure its numbers track real-life market share. AMD, meanwhile, is drastically underrepresented in iCafe gaming.
"They did change their algorithm a little bit, but they really aren't motivated to go in and change this," Herkelman said, "because the purpose of their data is not for market share. The purpose of their data is to show general trends to game developers... it definitely doesn't track our real share.... you can see the same thing actually happen in our CPU share. It's still under-represented, it's the same exact curve, and it's all related to iCafes."

TLDR, blame Nvidia and Intel for being too popular in internet café, and therefore being overrepresented because of too many consecutive login on the same PC by different people.

https://www.extremetech.com/gaming/295513-amd-explains-why-steam-doesnt-accurately-measure-market-share

 

[Grimling]

people really buying a 5070 over a 9070...

I build a Gaming PC for my dad some weeks ago, and i wanted the 9070, but it was 100 bucks more than the 5070. so it was the 5070 then.

 

[Tomorrow]

I still would like to know what these entries mean under the GPU section:

AMD Radeon Graphics 2.05%
AMD Radeon(TM) Graphics 1.99%
AMD AMD Custom GPU 0405 0.45%
AMD Radeon Vega 8 Graphics 0.4%
AMD Radeon Graphics (RADV VANGOGH) 0.35%
AMD Radeon (TM) Graphics 0.29% (second one)
AMD Radeon Vega 3 Graphics 0.23%
Other 9.45%

Some of the above I can guess but the Steam Survey seems to be unable to identify some AMD graphics correctly. So this statement:



Cannot be made if AMD graphics are not being identified correctly.

No. It seems the Steam Survey cannot identify RDNA4 cards by their model name. They could be in any of the categories I listed in my comment above.

I think detects the name correctly but in my case earlier this week it was unable to determine the VRAM. So it's possible 9000 series are thrown into the "Other" section. Tho the 16GB section has grown too.

Also three things the survey collects but does not report: it collects the panel diagonal size and refreshrate. And the CPU model name.
It would be very useful to see, but they're not reporting it despite collecting it. We could see actual CPU models and what refreshrate and sizes of monitor people are running.

 

[TheGeekn°72]

There are no 9070 9070xt near msrp sadly, i ended up buying the 5070 for 610$

I've had a look on most of the "NATO" countries' stores and noticed the 5070Ti was almost constantly significantly pricier than the 9070XT
Even if Nvidia outsold them, I genuinely cannot believe there isn't enough 9070XT units sold to not show up on the survey, no idea how that happens

So, the paltry amount of RX 9000 GPUs that have been sold would be grouped under "other."

considering the fact that AMD has otherwise significantly more stock than Nvidia at launch and they still went out of stock extraordinarily quickly and the subsequent restocks after that also did, I wouldn't exactly call that "paltry"

I know my single contribution to AMD wouldn't appear because in the just over 3 months I've owned it I haven't been f'in asked by Steam.

there should be the option to manually submit your hardware data within steam without using a Run command I think

`people really buying a 5070 over a 9070...`

Ummm, yes?
For a given budget people generally buy the better product.

Bruh, people who buy AMD cards obviously don't game with their computers.... I mean they only use the Xbox app and Epic games app....

I really, really hope you were being sarcastic.

I would expect the 5070ti to be the one leading the chart. It's kind of the "value" king of the lineup and this generation in general.

it's priced 900-1000€/$.

I'll leave it at that, I'll let you extrapolate the paragraph I'd have written otherwise.

 

[TheDeeGee]

"...desktop gamers eager for improved performance and AI-driven features"

Are desktop gamers eager for AI-driven features? Wasn't there a TPU survey awhile back that indicated otherwise? Maybe I imagined it...

I'm driven by a feature called Native Resolution.

Glad this is still possible with Indie games that are far more enjoyable to play anyway.

 

[TheGeekn°72]

I don't think that Valve is somehow avoiding to report Radeon users, it's just it's an opt-in survey... that wasn't really designed to represent marketshare to begin with. The survey prompt seems to be random; a hardware change won't always trigger it. You can trigger it manually, but that's an effort that many will not do.

That's the kind of stuff I genuinely don't get from Valve : they do absolutely insane batshit crazy stuff like SteamVR and the performance overlay and then have stuff like the hardware survey completely untouched and lacking SO MUCH pertinent information that could help so many game studios refine their graphics performance for the hardware that people have...

The other issue is that valve had to tweak their algorithm because computers where several people might log into a different Steam session could trigger the survey. So a single configuration could be reported several times and inflate the numbers

they could fix that by going off motherboard serial number or detecting the Steam client ID (assuming it got one)

TLDR, blame Nvidia and Intel for being too popular in internet café, and therefore being overrepresented because of too many consecutive login on the same PC by different people.

I'll just refer to the line above

I'm driven by a feature called Native Resolution.

Truly the best thing since the invention of video games :kek:

Glad this is still possible with Indie games that are far more enjoyable to play anyway.

yeah, Big Games is such in a sorry state of disrepair that it really saddens me studios we used to look up to to give up quality, enjoyable content we could all have fun with get us retextured garbage year over year, now including AI garbage that no one asked for just to save on the salary of a few artists...
I'll link one of my recent posts on that, I think you'll agree with me on the decline of corporate produced video games

 

[Onasi]

That's the kind of stuff I genuinely don't get from Valve : they do absolutely insane batshit crazy stuff like SteamVR and the performance overlay and then have stuff like the hardware survey completely untouched and lacking SO MUCH pertinent information that could help so many game studios refine their graphics performance for the hardware that people have...

1. Valve can only use the information that the GPU driver exposes, as W1zz said above. Their survey can only go off that.

2. Not a single major game studio would give a flying fuck about “refining” anything based on users hardware - they put together what they can for the corporate deadline and, these days, even if it runs like absolute dogshit the excuse would be “just use upscaling bruh”. Frankly, the information that’s already there is more than enough - they know that the majority of gamers on PC are on x60/x70 tier cards and their AMD equivalents (nobody cares about Intel GPUs, even Intel themselves). Unsurprisingly, this doesn’t stop said developers from releasing peak masterpieces that look like games from 5 years ago and bring down 5090s to their knees.