GTX 1650 Takes Top GPU Spot in Steam Hardware Survey

The Steam Hardware Survey (opens in new tab) has updated its results for November 2022, and there are some interesting tidbits to pull from the data. Not only is there a new top desktop card (opens in new tab), but it’s the first time this has happened for a long time. Caveats, of course, apply. Elsewhere, we see Intel and AMD integrated GPUs rising up the ranks, while there are more gamers using Linux than ever before.

The Geforce GTX 1650 (opens in new tab) now tops the GPU popularity contest. The low-end Turing card from 2019 had a 0.66% month over month increase, rising to the pinnacle of the graphics card rankings with a 6.27% share of gamers. Meanwhile, the GTX 1060 that was the former leader sank by 1.85% to 5.77%. The 1060 was launched in 2016 and had reigned since January 2018, the Pascal card proving popular for its price to performance ratio and relatively low power consumption.

There are of course lots of disclaimers we need to toss out. First, those are some large swings, which leads us to think Valve might have changed the way it samples, or it just got a very different sampling pattern. Second, certain GPUs have their figures combined while others do not — the Steam Hardware Survey combines figures for the 3GB and 6GB variants of the card, along with mobile and desktop variants, for example. That applies to the GTX 1650, which includes GDDR5 and GDDR6 variants, along with mobile and desktop versions.

If we do that same grouping of entries, the RTX 3060 would technically be in first place. The desktop card sits at 3.41% and the laptop variant accounts for another 4.63%, for a combined 8.04%. Still, if you ignore the finer details, this is the first time we’ve seen a different GPU atop the list in quite a long while.

Steam GPU results, November 2022

(Image credit: Valve)

The RTX 2060, a former contender for the top spot, also saw a rather massive drop in popularity of 1.46%, with 4.46% of respondents having one in their rig. Again, that’s a rather massive swing for a nearly four years old GPU to have in a single month. In fact, many of the top 20 GPUs show rather large single-month swings, so it’s more than likely that Valve’s sampling methodology is to blame. Maybe it updated things to be more accurate; we can only hope.

As usual, the top 20 cards is a seriously Nvidia-heavy zone, with AMD’s Radeon Graphics integrated chip its only entry at number 13. Intel is also represented by integrated GPUs, the Iris Xe used on 11th and 12th-gen chips, and the UHD Graphics that came in with 2017’s Kaby Lake Refresh.

The only really top-end card (opens in new tab) in the top 20 is the RTX 3080 (opens in new tab), which saw a slight bump to 1.84% of respondents to the survey. AMD’s RX 580 (opens in new tab) sits at number 21, just ahead of the integrated Vega 8 GPU. The latest cards from all manufacturers (RTX 4090, RTX 4080, and Intel Arc A770/A750) are nowhere to be seen, though it usually takes several months for newcomers to show up.

Steam Hardware Survey, CPUs November 2022

(Image credit: Valve)

In terms of CPUs, Intel still reigns with 67.1% of the users, though this has dropped by almost four percent from its peak in October. AMD gained all that market share, rising to 32.87%. The most popular Intel processor speeds were in the 2.3 to 2.69GHz range, suggesting laptops (opens in new tab), while AMD saw its largest cohort running 3.3 to 3.69GHz CPUs.

Apple Silicon chips took over from Intel for the first time in Mac Land, a 1.04% rise just pushing the Arm-based CPUs over the 50% mark. Linux gaming also saw an increase in popularity, likely because of the Steam Deck (opens in new tab), with a 0.16% rise meaning 1.44% of gamers who responded use the open-source OS, with the most popular distro being Ubuntu, though Arch Linux, on which SteamOS is built, sits in second place.

Overall, there are some interesting changes, but Valve’s statistical methods remain, at best, nebulous. With no explanation on how data gets collected (some of us suspect “new” or “changed” hardware gets sampled more often than existing hardware, for example), we can only present the results as they stand and suggest people not read too deeply into things.