Steam Hardware and Software Survey (May 2017)

I feel like this has actually been addressed repeatably. The fact that the survey behaves EXACTLY like one would have forecast with the influx of Chinese users is actually the testament to its statistical validity.

 

In order to make such claims, you should present verified data on the respective sizes of
a) the OEM market compared to the DYI market and the respective attachment rates of the specific models in question
b) the actual sales numbers over the time in contrast to the time, products were sold

What if a) was 8:1 and b) was not able to compensate? A way.

 

Steam hardware survey is all but reliable...

 

It's perfectly reliable as long as you understand the methodology of the survey and the market (people) that are using Steam.

Steam is less about AAA games as it is about indie games and casual games. Hence, older machines and hardware tend to stick around for a very long time. That means then when you start opening up the storefront to China and SEA countries, the survey will start to get flooded by a massive influx of older machines and hardware.

That is as expected with Steam opening up the store to be more friendly to developers trying to self publish. It also leads to abuse by some devolopers in flooding the store with titles that barely qualify as a game.

Regards,
SB

 

The fact that we are debating here 80 % of the time about methodology and wether there is any accuracy at all in this survey is a testament to its reliability.

Can't we just forget we have this thread?

 

Not necessarily - could be that some people do not like the results they see and in turn try to discredit the methodology or deniying accuracy based on anecdotal evidence. Me, I don't know whether it's accurate or not, simply because I do not know about the exact methodology. But based on just that, I could try and downtalk half all hardware reviews where they do not specify exactly what they're testing.

If it's an influx of older hardware, how come the massive increas of 1060s?

 

All right, except their data fluctuate like a boozer at the October Fest.. It''s not only about China playing CSS with 3yo hardware... Just look at the graphs of the last month http://store.steampowered.com/hwsurvey
Steam hardware survey was accurate and reliable some years ago.. Then they started to fucked it up months after month (incorrectly detected dedicated mobile graphics card, no real data about region markets, incorrectly detected directx version capable machines, drivers that could force Steam to take 2-3-4 another surveys just using a url...).
Moreover putting all the country zones all together makes those data completely useless since different country zones have different (free-speach) laws and different economics. E.G.: what can you really sell legally in the RPC market? What is the medium gaming machine at Rio? How different are the connection speed of Japan and Australia? How many Antartica gamers are penguins in real life? (Yes, a joke about another serious issue of steam data: https://steamspy.com/country/AQ)

 

Windows 10 was not allowed to sell in China, even the Jack Sparrow Pirates editions (C), until this year when Microsoft released a version completely tailored for the great shit wall of the RCP government. Moreover, as far I remember, the paladins of the democracy, free-speech and free-market banned AMD, Intel and NVIDIA to sell high-end chips in the RPC area, closing even more the Chinese market.

 

What specifically are the wild flactuations that we are supposed to be looking for?

 

In the past month the DX12 capable systems had many different spikes which cannot called "noise". Moreover despite the opening in the PRC market of October, the DX12 capable systems jump from 68.3% to 74% and that's cannot be possible since Windows 7 gained a +22.59% thanks to the PRC market (with a 63.6 market share including PRC). DX12 capable system should be Windows 10 users with a DirectX 12 driver detected.



It would be also interesting to analyse the AMD data, both GPU and CPU. Moreover looks like the Intel iGPUs has lose a lot of shares, even with the PRC market.... But this looks more another matter of discussion, as is another matter of discussion the bugs of the Steam user localization as the absence of per-market localized data.

 

RPC = remote procedure call?

If you mean China, that it absolutely, definitely was NOT *JUST* opened in October (where did you get this idea?), there is a chart in this thread showing the language trends over the last 6 months.

 

Sorry I meant "PRC" . Anyway that's does not explain the DX12/11/10 graph vs the OS share data. And those public data remains completely useless form most of developers.

 

I just took a look at the Steam Survey and it appears that they've just gotten the colors incorrect in the legend. If you look at the detailed breakdown of OS/GPU share for DX10/11/12 Systems, you'll see they correctly have Win 10 + Dx12 GPUs with low overall shares compared to other OSes or other GPUs.

Someone should report that to Valve as the overview for DX10/11/12 Systems is giving incorrect information compared to the more detailed breakdowns (which appears correct). DX12 + Win10 should be at 16.x% while Dx12 + Pre-Win10 should be at 73.x%

Regards,
SB

 

Oh this explains a lot (like the low). Good observation!

 

Too bad china's(like my Brazil) have a too little minimum pay. It would push DX12 levels up and beyond. China market deserves more atention by the Multinationals AMD, Intel and Nvidia.

 

Does Brazil still have insane prices on hardware?

 

My RX580 costed me ~450 dollars. Zen pricing is fine, for a country that's not US.

Problem is the minimal pay, but that's history for another time.

 

Well, the Chinese takeover of Steam is picking up steam (awful pun intended).

Chinese language, Win 7, Nvidia GPUs and Intel CPUs are yet again significantly up. Observed installment base rate for NVidia’s lower-end products across the last 3 generations (Kepler, Maxwell and Pascal) are up ~4-fold in the last 5 months.

 

Long live chinese PUBG botters.

 

Dunno what's the problem. "Everyone" wanted a globally representative distribution. Seems we're getting there. We only need to get the african residents on board now. /halfsarc