Valve Hardware Survey

[junkey]

As I started up Steam today, Valve Hardware Survey popped up. After finishing it myself, there was a link that guided me to current results of a survey. I found it interesting, as they represent changes in different information (OS, VR HW, GPUs, CPUs and so on).
There was only one thing I couldn't wrap my mind around: there was a part of a survey where they represent a number of CPUs installed in users computers, represented in percents. What I think, what they mean is a physical cores installed in system instead of a physical CPUs (I don't think there are almost 46% percent of Steam users that have 4 physical CPUs installed in their gaming rigs). 

As a fun fact: we can see the change in the number of cores in CPUs as there is a deficit of -0.61% in dual cores as per July to August 2016 and popularity of 4 cores CPUs raised +0.55%, respectively. In the GPU segment, we can see a raise of DX12 GPUs (+1.02%) and graphically (no pun intended ) we can see the Team Green, also known as nVidia, is leading with 57.57%, the second obviously falls the Red Team (specified as ATI) with 24.44% and Intel comes as third with 17.56%. 

All the data is accessible on: http://store.steampowered.com/hwsurvey/

[PlayStation 2]

I believe it's just easier for Steam to identify cores as physical CPUs rather than cores.

[GDRRiley]

the 970 is king. lol

[GoodBytes]

Old? August...

We are in October.

How about waiting for the next one?

[Duskie]

It bugs me that I've literally never got one of these survey requests on either of my systems with an AMD GPU, but all three of our Nvidia laptops get them???

[The Pizza Thief]

6 minutes ago, Flitter said:

It bugs me that I've literally never got one of these survey requests on either of my systems with an AMD GPU, but all three of our Nvidia laptops get them???

Interesting, maybe they want Nvidia users to take the survey so their numbers grow larger compared against AMD users. 

[Prysin]

So, me and some others decided to "recreate" the Steam hardware Survey build...

 

THIS is what the steam hardware would give you (best parts we could get)

PCPartPicker part list / Price breakdown by merchant

CPU: Intel Celeron G1620 2.7GHz Dual-Core Processor  ($39.99 @ SuperBiiz)
Motherboard: Biostar H61MGV3 Micro ATX LGA1155 Motherboard  ($39.99 @ SuperBiiz)
Memory: Team Vulcan 8GB (2 x 4GB) DDR3-1600 Memory  ($33.99 @ Newegg)
Storage: Western Digital Caviar Blue 320GB 3.5" 7200RPM Internal Hard Drive  ($29.15 @ Amazon)
Video Card: Zotac GeForce GTX 970 4GB Video Card  ($259.88 @ OutletPC)
Case: Rosewill FBM-01 MicroATX Mini Tower Case  ($24.99 @ Newegg)
Power Supply: EVGA 500W 80+ Certified ATX Power Supply  ($36.89 @ OutletPC)
Operating System: Microsoft Windows 10 Home OEM 64-bit  ($85.70 @ My Choice Software)
Monitor: AOC E2429SWHE 23.6" 60Hz Monitor  ($99.99 @ Best Buy)
Total: $650.57
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2016-10-01 15:31 EDT-0400

 

 

This is what building it yourself can get you

PCPartPicker part list / Price breakdown by merchant

CPU: Intel Core i3-6100 3.7GHz Dual-Core Processor  ($110.88 @ OutletPC)
Motherboard: ASRock H110M-HDS Micro ATX LGA1151 Motherboard  ($51.99 @ SuperBiiz)
Memory: Avexir Core Series 8GB (2 x 4GB) DDR4-2400 Memory  ($39.99 @ Newegg)
Storage: Western Digital Caviar Blue 320GB 3.5" 7200RPM Internal Hard Drive  ($29.15 @ Amazon)
Video Card: Sapphire Radeon RX 470 4GB Video Card  ($184.99 @ Newegg)
Case: Fractal Design Core 1100 MicroATX Mini Tower Case  ($26.99 @ NCIX US)
Power Supply: EVGA 430W 80+ Certified ATX Power Supply  ($33.49 @ SuperBiiz)
Operating System: Microsoft Windows 10 Home OEM 64-bit  ($85.70 @ My Choice Software)
Monitor: AOC i2369V 23.0" 60Hz Monitor  ($99.99 @ Best Buy)
Total: $663.17
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2016-10-01 15:32 EDT-0400

 

 

PCPartPicker part list / Price breakdown by merchant

CPU: AMD Athlon X4 880K 4.0GHz Quad-Core Processor  ($84.99 @ SuperBiiz)
Motherboard: ASRock FM2A88M PRO3+ Micro ATX FM2+ Motherboard  ($56.89 @ OutletPC)
Memory: Team Vulcan 8GB (2 x 4GB) DDR3-1600 Memory  ($33.99 @ Newegg)
Storage: Western Digital Caviar Blue 320GB 3.5" 7200RPM Internal Hard Drive  ($29.15 @ Amazon)
Video Card: Zotac GeForce GTX 1060 3GB 3GB Mini Video Card  ($199.00 @ B&H)
Case: Fractal Design Core 1100 MicroATX Mini Tower Case  ($26.99 @ NCIX US)
Power Supply: EVGA 430W 80+ Certified ATX Power Supply  ($33.49 @ SuperBiiz)
Operating System: Microsoft Windows 10 Home OEM 64-bit  ($85.70 @ My Choice Software)
Monitor: AOC i2369V 23.0" 60Hz Monitor  ($99.99 @ Best Buy)
Total: $650.19
Prices include shipping, taxes, and discounts when available
Generated by PCPartPicker 2016-10-01 15:35 EDT-0400

[Centurius]

10 minutes ago, Brynjar said:

Interesting, maybe they want Nvidia users to take the survey so their numbers grow larger compared against AMD users. 

Doubtful, the results are likely proper statistically just this one single person happens to have had them a lot on Nvidia systems. On the entire sample the law of big numbers is almost certain to apply.

[Enderman]

Yeah, physical CPU cores, so an 6600k and 6700k both have 4.

[Prysin]

Just now, Enderman said:

Yeah, physical CPU cores, so an 6600k and 6700k both have 4.

no, it doesnt understand difference between physical and threads from what ive noticed.

[The Pizza Thief]

2 minutes ago, Centurius said:

Doubtful, the results are likely proper statistically just this one single person happens to have had them a lot on Nvidia users. On the entire sample the law of big numbers is almost certain to apply.

Just throwing out some conspiracy theories lol 

[Notional]

 

Let's see what the September survey says. But do bear in mind, not all steam users are probed every month.

[WkdPaul]

-moved to PC Gaming-

[kirashi]

1 hour ago, Dan Castellaneta said:

I believe it's just easier for Steam to identify cores as physical CPUs rather than cores.

Not sure how hard it is to rewrite their survey strings to say CPU cores instead of CPUs... But sure.

[Enderman]

1 hour ago, Prysin said:

no, it doesnt understand difference between physical and threads from what ive noticed.

So you're saying that only 4.55% of people have an i7???

[Prysin]

2 minutes ago, Enderman said:

So you're saying that only 4.55% of people have an i7???

wouldnt surprise me really. Just look at the LTT forums.

 

people reccomend fucking i5s even with 1080s, despite there being clear evidence from several sources, one of the most respectable being Digital Foundry, that "just 4 cores/threads" isnt enough. A 6600k has been proven to slightly bottleneck a 1070. Let alone what it would do to a 1080.

But LTT and the rest of the internet seems to think that "i5 is all you need for gaming". So i keep seeing idiots suggest i5s even with 1070 and 1080 SLI systems.

 

So no, it doesnt surprise me AT ALL.

 

Although, i could be wrong, and i do suspect that AMD FX CPUs are "read" wrong, that it reads the module count, as that is what windows consider "physical" and not the actual "ALU core" count.

[Enderman]

5 minutes ago, Prysin said:

wouldnt surprise me really. Just look at the LTT forums.

 

people reccomend fucking i5s even with 1080s, despite there being clear evidence from several sources, one of the most respectable being Digital Foundry, that "just 4 cores/threads" isnt enough. A 6600k has been proven to slightly bottleneck a 1070. Let alone what it would do to a 1080.

But LTT and the rest of the internet seems to think that "i5 is all you need for gaming". So i keep seeing idiots suggest i5s even with 1070 and 1080 SLI systems.

 

So no, it doesnt surprise me AT ALL.

 

Although, i could be wrong, and i do suspect that AMD FX CPUs are "read" wrong, that it reads the module count, as that is what windows consider "physical" and not the actual "ALU core" count.

Uh no, there are WAY more people with an i7 than 5%.

http://imgur.com/QvQ1LBk

[GanstaKingofSA]

Why are we talking about this if this was release like 1 yr ago?

[Prysin]

6 minutes ago, Enderman said:

Uh no, there are WAY more people with an i7 than 5%.

http://imgur.com/QvQ1LBk

that is only from 12600 people, one can assume that the wast wast majority of those who responded to the reddit poll was probably "above" you average consumer in terms of dedication and hardware.

I dont think too many of the filthy casual laptop plebs who "ruin" the steam survey each month posted in that reddit poll.

[Enderman]

6 minutes ago, Prysin said:

that is only from 12600 people, one can assume that the wast wast majority of those who responded to the reddit poll was probably "above" you average consumer in terms of dedication and hardware.

I dont think too many of the filthy casual laptop plebs who "ruin" the steam survey each month posted in that reddit poll.

If it was not accurate then why is it 80:20 intel:amd in both surveys?

Clearly a sample of 12k is statistically representative of the population.

 

So yeah, sorry, but it's pretty obvious that steam counts 4 core 8 thread i7s as a 4 core CPU.

Which is true, because they are 4 core CPUs.

[Dackzy]

12 minutes ago, Enderman said:

Uh no, there are WAY more people with an i7 than 5%.

http://imgur.com/QvQ1LBk

0.3% Phase change and 0.3% LN2, those numbers are a bit high, I think that we can safely put those numbers to 0.0003%. I think it is safe to say that this survey is mainly answered by PCMR people. Also 12.6k people are not a lot.

[Enderman]

2 minutes ago, Dackzy said:

0.3% Phase change and 0.3% LN2, those numbers are a bit high, I think that we can safely put those numbers to 0.0003%. I think it is safe to say that this survey is mainly answered by PCMR people. Also 12.6k people are not a lot.

The phase and LN2 are mostly trolls, so that tells you that about 0.5% of the survey is margin of error.

12k people is a statistically significant sample size.

[sgtcool]

2 hours ago, Notional said:

 

 

Let's see what the September survey says. But do bear in mind, not all steam users are probed every month.

Hmm what happened in Q2 10?

[Notional]

4 minutes ago, sgtcool said:

Hmm what happened in Q2 10?

NVidia released Fermi in April. Before that the 300 series sucked, and well Fermi made it possible to forge the ring of Sauron in your computer.

[Prysin]

2 minutes ago, Enderman said:

If it was not accurate then why is it 80:20 intel:amd in both surveys?

Clearly a sample of 12k is statistically representative of the population.

 

So yeah, sorry, but it's pretty obvious that steam counts 4 core 8 thread i7s as a 4 core CPU.

Which is true, because they are 4 core CPUs.

i dont think that survey is actually representative.

Just because SOME of the facts are the same, does not mean the rest is.

 

let us look at some statistics, shall we?

http://store.steampowered.com/stats/?l=norwegian

https://en.wikipedia.org/wiki/Steam_(software)

Quote

The service has over 125 million registered accounts.^[a][6] Steam has had as many as 12.5 million concurrent users as of November 2015

Meaning the Hardware Survey is based off AT LEAST 100m+ users.

 

Do you really think out of 100m users, that 0.3% uses Phase Change, DICE or LN2?

no, not a chance. Not a chance in fucking hell.

Do you think 33.6% of users has a GPU in the 300-600$ Range? NOPE.

 

Listen to some stats from the vendors.

AMD and Nvidia alike has BOTH stated that the "most sales by numbers happen in the 199-350$ segment. With the majority of those being on the "lower half".

That is what BOTH the vendors state. and has stated on more then one occasion.

 

Do you think 14% of users has a 500$+ card?

no, not even close. Not even REMOTELY CLOSE. And those numbers doesnt add up at all, if those numbers were real, Nvidia and AMDs graphic sections would be reporting much higher revenues then they really are. At 100m+ steam users, 14% being 14m or above. At 500$ x 14m = 7 billion USD.... NEITHER OF THEM MAKE ANYWHERE NEAR THAT MUCH MONEY PER YEAR.

 

If AMD and Nvidia "split" is according to your reddit thread, then out of 7 billion dollars in sales, 67.7% goes to Nvidia. Who gets how much per 500$ sale? i dont know, but let us assume 30% margin at least. 30% of 500 = 150$ in revenue to Nvidia.

 

67.7% of 14 = 9.478 / 14 GPUs are Nvidias.

 

9.478% of 100m users = 9.478 million 500$ Nvidia GPU users (GTX 970 which was a 299 USD card hold 5% of steams numbers. So already your "evidence" is solid AF right?)

Meaning assuming 150$ profit per 500$ GPU, with 9.478m GPUs at 500$ sold, Nvidia should make 1,421 BILLION USD from just the 500$ segment.

 

Let us look at Nvidias fiscal year of 2015 http://www.anandtech.com/show/8973/nvidia-releases-q4-fy-2015-and-full-year-results-record-revenue-for-quarter-and-full-year

Quote

NVIDIA released their earnings report for the fourth quarter of their fiscal year 2015, which ended January 25th, 2015. FY 2015 was a record for the company, with revenues coming in at $4.68 billion

 

A HUGE important fact here is the following:
If 500$ GPUs alone (a section of the market that Nvidia openly admitts is NOT the highest volume or highest profit section. It is the highest profit MARGIN section however) generates nearly 25% of all your income, assuming they sold nearly 9.5m 500$ cards, then you'd be right, but i find that hard to believe, as both AMD and Nvidia both state that they sell a lot more in the cheaper segments. ANd thus gain much more money due to volume. Laptop GPUs for example is grossly overpriced, but the majority of them are GTX 940m, 950m, 960m, 965m and 970m... None of which go above the 350$ mark in cost to the OEMs, which actually have to buy the card before putting it into a laptop.

 

Since we know the laptop segment is larger then the desktop segment due to AMD offering some insightful research on the split in types of computers last year, we also know that the laptop segment that sells the most is the 300-800$ ones, which is where you will find the 940m at best... NOT the higher end stuff. Thus volume sales of these cards alone make your evidence weak.

 

 

TL;DR

Your picture isnt remotely close AT ALL as to being scalably relatable or in any other way representative. Any similarities in the surveys are pure coincidence and nothing else.