The Radeon RX 480 has appeared on Steam Hardware Survey

Faceless Rebel

New member
This is of course the companion thread to the one I made in Other Graphics.

The RX 480 does not yet have enough share to appear in the page at http://store.steampowered.com/hwsurvey/videocard/

Instead you have to go to http://store.steampowered.com/hwsurvey/directx/ and use your F3 skills and search for cards one after another and add up their shares from the DX12 and DX11 lists. (The DX10 list is the same for these cards, simply look at the descriptions to understand why.)

Note that the numbers between the Videocard page and the DirectX page are not directly comparable. This means the numbers in the other thread!

From this page, we can determine that RX 480 share is 0.05% + 0.06% = 0.11%.

As a point of comparison with the new Nvidia cards, click only if you want to see:
The GTX 1070 has 0.26% + 0.31% = 0.57% and the GTX 1080 has 0.26% + 0.29% = 0.55%. The GTX 1060 doesn't appear yet.
 
  • Like
Reactions: rms
The Steam hardware survey really paints a bad picture for AMD. It looks like their last really successful card was the 7900 series. Fury doesn't even appear on the main list.

Another interesting thing is this does go to show how few people actually buy very high end cards. The 980 and 980 Ti's share is a lot smaller than the 970's. Although, interestingly the amount of 980 Ti's is the same as the 980, so it looks like people who have money to spend are just as likely to go all the way (similarly the number of 1070s and 1080s is very close).

AMD really need to turn things around with Vega.
 
Last edited:
Yeah, the 1080 being so close to the 1070 in share despite also being almost double the price does suggest that people who buy high end cards tend to go all the way. Go deep or go home is the motto at the high end, which probably bodes well for the $1200 New Titan X.
 
The Steam hardware survey really paints a bad picture for AMD. It looks like their last really successful card was the 7900 series. Fury doesn't even appear on the main list.

Another interesting thing is this does go to show how few people actually buy very high end cards. The 980 and 980 Ti's share is a lot smaller than the 970's. Although, interestingly the amount of 980 Ti's is the same as the 980, so it looks like people who have money to spend are just as likely to go all the way (similarly the number of 1070s and 1080s is very close).

AMD really need to turn things around with Vega.
That is because the 7970 was the last successful AMD product.
 
Take that thing with a grain of salt. RX 480 was sold in greater numbers than the 1080. 1080 was simply not available in great numbers and high end doesnt sell anywhere near that of mid range cards.

Many review sites stated the low availability of the 1080 while the RX 480 had strong availability at launch yet sold out.

I mean just look at some of the other numbers. R9 380 went from 0 to 1.14% in a single month? R5 335 went from 0 to 0.64% also in a single month? Its not making much sense. Both in July?
 
Last edited:
It looks like something weird went on with the classification of AMD cards.

What it looks to me is that 2xx series are now being identified as 3xx series. So, you see that the 200 series drops by a large amount, but the 390 series and 380 series go up a lot.
 
Ive gone months not logging into steam. Its looks like we are seeing a minority of steam users logging in for whatever reason in a given month. I suppose dependent on whatever games they were waiting on a sale on.
 
Ive gone months not logging into steam. Its looks like we are seeing a minority of steam users logging in for whatever reason in a given month. I suppose dependent on whatever games they were waiting on a sale on.

I'm logged into Steam all the time, but I only rarely get queried for the survey. Like, I can remember maybe once or twice in the past five years. Unless it's just a check box, one time decision and it does it automatically after that?
 
The Steam Hardware Survey polls a random sample of the total Steam user base. You might get invited to take the survey or you might not, because it's random.

Anyone who understands how statistics works will understand just why the SHS is so statistically sound and why it's regarded as the Bible for developers seeking to tune their game system requirements so they hit the largest number of their target gaming population. To question the validity of SHS is sheer ignorance and nothing more.
 
You dont think those sudden big % increases of archaic gpus is odd?

Personally Ill take sales numbers anyday. Itd be more interesting to see telemetry of steam users as to what gpus they have instead.

I can see it having value for devs due to the wide variety of gpus out there but to gauge sales of recent gpus is another thing altogether.
 
I already explained why there might be sudden % changes in very old GPUs which are presumably only used by a very small number of the total Steam user population. A basic understanding of statistics is all that is necessary to understand this.
 
Yes but by polling only a small sample of willing volunteers (people with modest gpus might not care to while those with epeen gpus would gladly do it) its very possible to miss a completely new gpu that has only been out a month. Or sample a large number of otherwise rare gpus.

This kind of sampling only makes sense when you tally months worth of data.

Not just one month.

This brings to mind the census in Canada. Where one form, the short form was voluntary, and the other a long form was obligatory. The head of StatsCan said by the tories removing the obligatory long form census StatsCan could no longer vouch for the accuracy of its data. He made quite a strong argument for that issue as the census was one time every 5 years. By steam being only voluntary you needs a lot of polling to gauge an accurate gpu presence in the market. If its polling was mandatory such as under a steam agreement on installing the app they would not need to do such repetitive polling.
 
Last edited:
According to the numbers on this pipeline article on Anandtech, desktop discrete graphics card sales have dropped from 68 million in 2011 to 44 million in 2015.

But, worse for AMD, three quarters of that decline came out of their share: from 27 million units down to 8.88 million units. Which is why Nvidia now has 80% market share. Nvidia's sales have also dropped, but they managed to hold on better than AMD.

Now if Nvidia is at 80% share, that looks basically about right with what we see in the Steam Hardware survey. Basically four times as many people buy Nvidia cards.

Still just totally brutal, as AMD is getting killed on two fronts. Either they turn it around within the next year or so, or they're toast, and sadly we're all probably toast soon thereafter (or at least our wallets will be).
 
It looks like something weird went on with the classification of AMD cards.

What it looks to me is that 2xx series are now being identified as 3xx series. So, you see that the 200 series drops by a large amount, but the 390 series and 380 series go up a lot.

It's a mess due to all the rebadging AMD did with the R9 300 series. GPU's could be GCN 1.0, 1.1, or 1.2

AMD's focus on mobile APU's means basically any "gaming" laptop will sport an NV GPU.
 
May be sad the FuryX doesn't appear on the main list, but I bought it at launch, and damn I love this card! Doom runs like crazy on it! I'll buy the heck out of Big Vega whenever it appears.
 
The thing that concerns me about AMD's dropping market share is it means that developers are less likely to optimize for their hardware. I guess with the consoles this is sort of balanced out, but it's still a problem with small developers like in VR. I feel like some games run much worse on my Fury than they should, because a small team with limited resources will just focus on making sure the game runs well on Nvidia and just make sure it's halfway playable on AMD.
 
AMDs market share is not dropping, its increasing.

not sure why you think it is.,

It may be increasing compared to last year (honestly, couldn't get much worse), but it is way down from a few years ago. When AMD/ATi were 40% of the market, developers couldn't really ignore them. When they're at 20%, it's a different story. They still shouldn't, but some unfortunately will just treat Nvidia as the default target and have AMD as an after thought.
 
Back
Top