What's going on with the steam hardware survey?

[Zef]

AMD GPU's suddendly dropped from their usual 16% market share to 11% and their CPU's dropped at the same time from almost 38% to 31%?

Is steam hardware survey bugged again? I could understand the GPU drop but that CPU drop from 38 to 31% is massive.

[wesnef]

Tom's Hardware reported some big shifts in numbers in an article a few days ago, seemingly attributed to stats from China being added in. Not sure if that would be part of what you're seeing.

edit: https://www.tomshardware.com/desktops/gaming-pcs/huge-os-and-ram-usage-swings-in-steam-survey-likely-to-have-been-influenced-by-china-influx

[Takohashi]

Originally posted by Zef:

Is steam hardware survey bugged again?

Yeap. Incorrect statistics can be easily identified by the data of the language used.
The first thing you look at is the language and when you see "Simplified Chinese +20.88%", you can close the statistics page.

[Illusion of Progress]

It's another one of "those months" and you can basically ignore the results.

Every so often, Steam has one of these "outlier months" that is seemingly caused by higher than normal proportion of results coming from China and it skews the results. The telltale symptoms of one of these months is the following; the Chinese language, Windows 10, Intel CPUs, nVidia GPUs (recent x60s in particular), 1440p resolution, and 32 GB RAM all increase in use (and therefore other things see drops). Speculation is that more net cafes in China are being polled these months.

Look back towards the beginning of the graph on the CPU and GPU split and you see the same exact spike for one month around the end of 2023. These outlier results have happened at least a few times before (and will likely happen again until Valve figures it out, if they even know/care).

I almost guarantee it will be back to results similar to those of last month, next month (unless it has two outlier results in a row, which hasn't happened yet).

TL;DR: They are skewed results and you should use the results of last month and/or wait for results of next month. You can ignore the results showing through March.

[Bad 💀 Motha]

The AMD GPU drop could be a result of a rise on NVIDIA GPUs. More people have also been using and testing the Intel GPUs as well.

[Philco7a]

Did it for the first time. Was shocked to see most used is a 2ghz cpu, almost 1/4
What games are those people playing?

[Bad 💀 Motha]

Originally posted by Philco7a:

Did it for the first time. Was shocked to see most used is a 2ghz cpu, almost 1/4
What games are those people playing?

Laptops. Many Intel i3 are around 2Ghz bade clock. The Turbo doesn't count

[Tonepoet]

A.M.D. should be stronger than usual next month since A.M.D. sold more stock of the 9000 series cards than Nvidia did of the 5000 series according to Microcenter, but the launch was more recent so maybe it hasn't had a chance to be reflected in the results yet.

[nullable]

Originally posted by Tonepoet:

A.M.D. should be stronger than usual next month since A.M.D. sold more stock of the 9000 series cards than Nvidia did of the 5000 series according to Microcenter, but the launch was more recent so maybe it hasn't had a chance to be reflected in the results yet.

What's the percentage of PC gamers who bought new video cards this month? Might not be a significant impact even if the launch was successful. Changes are bound to be fairly incremental from month to month most of the time.

[Bad 💀 Motha]

Originally posted by nullable:

Originally posted by Tonepoet:

A.M.D. should be stronger than usual next month since A.M.D. sold more stock of the 9000 series cards than Nvidia did of the 5000 series according to Microcenter, but the launch was more recent so maybe it hasn't had a chance to be reflected in the results yet.

What's the percentage of PC gamers who bought new video cards this month? Might not be a significant impact even if the launch was successful. Changes are bound to be fairly incremental from month to month most of the time.

I doubt it's even 1%