December Steam Survey Numbers Points Towards Slow Death of the Quad Core Gaming CPU

[dirtyferret]

The problem you describe when it comes to looking at the Steam Survey is that there is absolutely no context whatsoever to these numbers

I agree whole heartily with your post but I do want to say there is context to the numbers but it is extremely broad context and people flip out trying to take that broad context and narrowing it down to themselves.

I personally do not like peppers. I don't like them cooked, raw, stuffed, in soups, or dishes. If someone was to do an unbiased random sample survey of people's favorite vegetables would I expect peppers to do well on the list? Sure, 100%. Why? Because ive been to various grocery stores across the USA. I used to travel a lot for work and have lived in four different states so every time I entered a grocery store, I see fresh peppers up front in the produce isle and they are not rotting away so people are obviously buying them. Does that make me want to buy one? No.

 

[seth1911]

Not really a news, its still a low Corecount in the mass against nowdays are 16 Cores aviable for Consumer.....,
im too in the chart wih my A4 4300M (1 Modul) and my A8 3800.

11,88% use 2
35,3% use 4 (= still here 47,18%)
33,25% use 6
--------------
80,43%

 

[Keullo-e]

If you click on https://store.steampowered.com/hwsurvey/videocard/ there is DirectX 8 GPUs and below in the top section.


It definitely is a great source of HW info for Steam users, but that is it. It's impossible to generalize it to PC gaming as a whole, because it's a bubble. Here is a fun read on why such surveys are problematic: https://www2.math.upenn.edu/~deturck/m170/wk4/lecture/case1.html

Much more trustworthy source is: https://www.jonpeddie.com/

I should fire up my AGP system with GeForce 3 Ti 200 (DX8 card) on it

 

[Chrispy_]

It has been a while since I saw a test but IIRC the 4C/4T CPUs have been suffering in games for a couple of years now, and the 6C/6T CPUs were okay last time I checked, just like the 4C/8T CPUs.

Definitely with the 10400F spending most of the last 15 months at $170 or less (as low as $145 and currently closer to $160) it's really hard to see why you wouldn't just get a 6C/12T these days.

 

[Minus Infinity]

I see the RX 6X00 cards still barely make a dent too

Hold your horses, what's an RX 6X00?

 

[watzupken]

The decline in quad core CPU users should not come as a surprise. I feel the real reason is because CPU makers, Intel and AMD, have neglected this segment for many years. AMD released the R3 3300 series which are out of stock most of the time. They never followed up with a Zen 3 quad core other than the APUs, which is generally not available in retail, unless you get it from OEMs. Intel as well have not updated their i3 series for some time since it is still using the same old Skylake architecture even till now. So essentially you want the latest architecture, you need to go with the Ryzen 5 or i5.

 

[AlwaysHope]

I see the RX 6X00 cards still barely make a dent too

Lol... 1st gen RDNA series barely 1%!

 

[TheinsanegamerN]

Lol... 1st gen RDNA series barely 1%!

RX 5x00 are 1st gen. RX 6x00 is 2nd gen. It does really reflect on how few of those 6000 series are selling that, despite being better at mining, nearly all of the RTX 3000 series cards individually hav emore marketshare then all the 6000s combined.

Hold your horses, what's an RX 6X00?

All RX 6000 series. 6900xt, 6800xt, 6800, 6700xt, 6600xt, 6600, and soon 6500.

 

[Mysteoa]

I see the RX 6X00 cards still barely make a dent too

Of course, they are not AMD's top priority. Why should AMD make 250mm+ gpu chips, when they can make 100mm+ CPU chips and sell it for more.

 

[kruk]

1. I did not to intend a direct attack towards you and apologize if you felt this way.
2. It is extremely frustrating discussing the STEAM hardware survey every single time a thread is posted about it, almost on a monthly basis. I also worked in this field for a long time; Nielsen ratings, MRI market index, Arbitron (now Nielsen audio), quantitative and qualitative analysis, etc., etc., so the countless posts of people not understanding what the numbers say or think it's voodoo magic can get rather tiresome.
3. If you want a civilized discussion feel free to send me private message but I stopped trying to have an education discussion about the STEAM survey years ago. Too many people simply toss a hissy fit because their favorite brand is not at the top or that "inferior" hardware is too well represented. What either one has to do with anyone's PC and how they take that as an insult to their ego is beyond me and maybe that should be the real discussion.

Apologies accepted. The thing that I always doubted the most is if Steam really a representative sample of PC gaming? It obviously has a lot of users, but this doesn't necessarily mean it's a good sample. For example: we have data for other sources that the market share of dGPUs in Q3 2021 should approximately be 1:4, but it's more like 1:10 according to Steam survey. So, why such a big discrepancy?

I also dislike their representation of GPU models and inconsistent grouping. Example: GTX 1060 3 GB and 6 GB are grouped into a single product, but RX 580 2048SP and RX 580 are not and many more examples. That is not how data should be represented and people draw wrong conclusions from it ...

 

[dirtyferret]

The thing that I always doubted the most is if Steam really a representative sample of PC gaming? It obviously has a lot of users, but this doesn't necessarily mean it's a good sample.

Yes, in fact it is an excellent representation but STEAM is an account on a gaming PC not a user. So one user can have multiple accounts and/or multiple PCs/laptops (but at may not even opt into the survey so the difference made is minuscule) as well as multiple users on one PC (also minuscule stat). That said STEAM has over one billion accounts and 120 million active monthly users. There are roughly 1.7 billion PC gamers (more than half in Asia) so STEAM clearly has a dominating market penetration with PC gamers making the two communities practically one and the same. You only need a poll of several hundred STEAM accounts to get your margin of error down to 5% or lower for PC gamer representation.

For example: we have data for other sources that the market share of dGPUs in Q3 2021 should approximately be 1:4, but it's more like 1:10 according to Steam survey. So, why such a big discrepancy?

I'm not sure where you see 1:10? Also the article you linked is dGPUs shipped and market share of shipped units. Regardless, in December STEAM showed a split of 77% Nvidia, 15 AMD, 9 Intel for GPUs ( I rounded up). Your article shows a 79/21 split of share for Q3 2021 among Nvidia and AMD dGPUs. This gets back to my previous statement of trying to take info from one poll (that includes integrated graphics) into another poll (that only shows dedicated graphics). They are both within the same ballpark of one another so you can use both polls to see Nvidia has a larger market presence compared to AMD but that is it. One poll is separate from the other poll.

I also dislike their representation of GPU models and inconsistent grouping. Example: GTX 1060 3 GB and 6 GB are grouped into a single product, but RX 580 2048SP and RX 580 are not and many more examples

I agree but it's not my poll.

That is not how data should be represented and people draw wrong conclusions from it ...

People draw the wrong conclusions from data all the time. Too often it is, "I already made up my mind and I just need to find the data to back it up."

 

[Chrispy_]

Lol... 1st gen RDNA series barely 1%!

They're all mining, not playing steam games.

1st gen RDNA cards were commanding ridiculous premiums. Even before Nvidia introduced LHR versions, the 5700 and 5700XT were incredilby good ETH miners, matching the hashrate of an original (non-LHR) 3070 but at lower power draw. Now, thanks to LHR a 5700-series card is about 40% better than a 3070!

I sold my used 5700XT in the UK for (adjust to US pricing) $1150 in May last year, back when new RTX 3080 cards were selling for $1350 (if you could find one in stock back then).

I wouldn't be surprised if many gamers with 5700-series cards sold them on ebay to fund a 3060Ti or similar upgrade and still have more leftover money from the sale than they originally paid for the 5700-card in the first place!

 

[AlwaysHope]

They're all mining, not playing steam games.

1st gen RDNA cards were commanding ridiculous premiums. Even before Nvidia introduced LHR versions, the 5700 and 5700XT were incredilby good ETH miners, matching the hashrate of an original (non-LHR) 3070 but at lower power draw. Now, thanks to LHR a 5700-series card is about 40% better than a 3070!

I sold my used 5700XT in the UK for (adjust to US pricing) $1150 in May last year, back when new RTX 3080 cards were selling for $1350 (if you could find one in stock back then).

I wouldn't be surprised if many gamers with 5700-series cards sold them on ebay to fund a 3060Ti or similar upgrade and still have more leftover money from the sale than they originally paid for the 5700-card in the first place!

Appreciate the insight, mining has no interest for me, but explains a LOT.

 

[Chrispy_]

Appreciate the insight, mining has no interest for me, but explains a LOT.

Mining is probably the largest single factor that explains why cards are hard to get and stupidly overpriced. Component shortages increase the price a bit, but the biggest factor by far is demand that far outstrips supply.

It made economic sense for me to start mining and I bought 24 cards, making me a very small-scale miner. Even so, I don't think I've bought 24 cards for personal use across 2 concurrent PCs in my entire life. I've only bought 28 cards for personal use in 26 years, until ETH mining when I bought a pallet of 20 direct from my sysint distributor, and sourced 4 more at retail. If I'd had hindsight and enough spare capital I probably would have bought three more pallets of 20 in Feb/March 2021. By June they were worth twice what I'd paid back then.

S3 Virge, Voodoo Rush, Voodoo Banshee, Riva TNT2 ,Geforce 2 MX, Geforce 3, Radeon 9200, Radeon 9700 Pro, Radeon 9550, Geforce 6800 GT, Radeon X1800 XT, Radeon X700 Pro, Geforce 8800 (G80), Geforce 8800 GTS-512 (G92), Geforce 7600GS, Radeon 5850, Radeon 6850, Radeon HD 7950, Radeon R9 290X, GTX 970, Radeon RX480, RX Vega56, 2060 FE, 5700XT, 2060S, 2070S, 3060, 3060Ti

 

[lexluthermiester]

You know you can... click and get those details? Physical CPU's is based on how many full cores are reported to the OS, because of the scandals in the past from the CPU makers disagreeing on what defines a core

View attachment 232115
View attachment 232116
View attachment 232117

What I find interesting is that there are 2 core CPU's in fourth? That's a bit mind boggling these days. 4 and 6 core CPU's represent the bulk of the numbers with 8 cores coming in third.

 

[AlwaysHope]

What I find interesting is that there are 2 core CPU's in fourth? That's a bit mind boggling these days. 4 and 6 core CPU's represent the bulk of the numbers with 8 cores coming in third.

If Intel, could be running Pentiums or Celerons with HT disabled?

 

[Mussels]

What I find interesting is that there are 2 core CPU's in fourth? That's a bit mind boggling these days. 4 and 6 core CPU's represent the bulk of the numbers with 8 cores coming in third.

Laptops. Many many years of intel laptops being dual core with HT, from i3's to i9's all basically being the same for over 5 years -.-