Steam Hardware Survey shows big changes after Valve fixes counting

Game development requires a careful balance between keeping system requirements low enough to make the product accessible to a wide range of players while incorporating attention-getting features and a helping of eye candy. Steam's Hardware Survey is one of the tools that coders can use to make sure their software targets the the right hardware specs. The latest round of stats has some pretty big changes, but the wildest swings probably come from a correction in the way Steam counts internet café machines operated by many users. As it turns out, the last few months of stats may not have been entirely reliable.

The latest results come with a note explaining that the figures started changing in unexpected ways back in August of last year. Valve's engineers thought it had taken measures that would prevent PCs from net cafés from being counted multiple times, but the way administrators tend to those systems inadvertently circumvented Valve's workarounds. The company says it has taken these details into account and that the latest results are more representative of true state of the gaming-PC install base. As a result, we suggest taking the deltas presented in the latest set of results with a boulder of salt. The biggest change is a large swing away from the 64-bit version of Windows 7 to  Windows 10 64-bit. This change is almost certainly a reflection in the way that Valve now counts machines in East Asia.

Those wild swings in the charts coincide with a large increase in players using the Simplified Chinese language pack in Steam

Entry-level and mid-range graphics card from Nvidia's Maxwell and Pascal families appear to have lost a combined share of about 16% of the Steam player base, but we'd wager this large shift was also caused by the change in the counting of multi-user systems. The gains didn't seem to go to any one card in particular; the only card to grow its Steam hardware share by more than 1% was the almost-two-year-old Nvidia GeForce GTX 1070.

The high price of memory might be affecting the buying habits of PC gamers. The latest results show a 4.1% drop in the share of gamers with 12 GB or more of memory, a 2.7% drop in systems with 8 GB of RAM, and a 3.2% rise in players gaming on just 4 GB of system memory. The share of users with just 1 GB of video memory also increased, while the share of players with 2-5 GB of VRAM dropped by just over 12%. The changes in system and video memory could just as likely have been caused by the change in the way systems are counted.

The share of four-core processors dropped by a bit over 8%, while dual-core processors saw a big 5.8% jump, and six-core chips made a smaller 1.3% leap. The results also show a 5% increase in the number of AMD-powered systems, all at the expense of Intel's gaming CPU market share. Valve's notes specifically mention that the number of  four-core CPUs started rising in time with the growing numbers of players using the Simplified Chinese language within the Steam application, so we wouldn't suggest reading too much into these changes, either.

In the realm of VR headsets, Oculus' Rift continues to gain ground on HTC's Vive. Windows Mixed Reality headsets are also making inroads, though players with VR headsets of any type still make up only 0.3% of the player base.

In other Steam-related news, those without any XBox or Playstation controllers to use in Steam can download the beta version of the client and use a Nintendo Switch Pro Controller instead.

Comments closed
    • kvndoom
    • 1 year ago

    A painful blow was struck to the Windows 7 Master Race.

    Like a terrible fish.

    • Zizy
    • 1 year ago

    To verify GPUs sold, take a look at Mindfactory.de. They are nice to give their sales data and should be able to count stuff without errors, unlike Steam 😛

    There, I counted a total of 93.5k NV cards sold on the first page of 30 entries, out of 210. Giving about 655k cards sold in total.
    I counted a total of 25.5k AMD RX cards sold on the first page, 57 entries, and let’s say Vega fills those 3 spots and we have a total of ~51k AMD cards sold.
    Sure, numbers of cards sold drop with pages, but I guess relative numbers should still be somewhat in the same ballpark.

    If 51k corresponds to 1.8%, we would have about 23% of NV market share with 10xx = not that far off from their 33% on Steam survey, especially given AMD popularity with miners.
    So, it should seem Pancake numbers are actually reasonably close to reality, no matter how bogus Steam survey might seem. Except, they aren’t, for a crucial detail:

    All NV cards are listed there. Great, so far. But for AMD, only the new 5xx series are listed – the old 4xx aren’t. Correctly estimating sales of 4xx vs 5xx at this point is probably impossible.
    We can use 2 different wild guesses, both are likely wrong by various degree:
    1.) 4xx were sold for 22 months (July 2015 -> April 2017) and 5xx were sold for 13 months (April 2017 -> April 2018; May barely started), and sales are linear.
    This gives 4xx sales as 1.7x higher than of 5xx, leading to 137k AMD cards sold (vs 655k NV), and 1:5 sales ratio of AMD vs NV.
    2.) Steam survey. 0.63% vs 0.29% and 0.44% vs 0.19% leads to 2.2 and 2.3x higher sales of the old 4xx cards, leading to total of 166k AMD cards sold and a 1:4 sales ratio.

    Now obviously all this misses the magnitude of mining cards – one could easily argue that arbitrary percentage of AMD’s new cards is used for mining instead of gaming and that Steam numbers are accurate…

      • stefem
      • 1 year ago

      Agree, but if you look better you will find entry for AMD 4xx series too

    • Pancake
    • 1 year ago

    Diving into the details of graphic cards actually used is very revealing.

    Basically, NVidia cards past and present are widely used in a very healthy even mix based on the various card operational lifespans.

    AMD is in a way scary position. Despite having being used in nearly 15% of systems running Steam, this is all the usage of the current gen cards on the list (excluding the vanishingly small quantities lumped into the “other” category)

    RX480 0.63%
    RX460 0.44%
    RX580 0.29%
    RX470 0.24%
    RX560 0.19%

    for a grand total of 1.79% which is somewhat less than the less-than-venerable older NVidia GTX950 which I raise merely as a mediocre comparison point. Much less the mighty GTX1060 at 12.29% all by itself. So, most AMD GPUs used are old legacy things.

    Wow. Just wow. Basically, practically for the last couple of years NOBODY is buying AMD cards for games! AMD – the Way it Isn’t Meant to be Played!

    There has just got to be some massive consequences for this disparity. Nobody’s going to bother to optimise their games for recent AMD GPUs. Why bother? Nobody USES the darn things. This is particularly crucial for DX12 where the radically different architectures of NVidia/AMD GPUs and low-level nature of the API requires significantly different optimal code paths. The driver ain’t gonna abstract it all away for you. This is not the case for CPUs where AMD/Intel run code pretty much the same way.

      • DPete27
      • 1 year ago

      Perhaps it’s because I use an AMD GPU, but I really have a hard time believing that number is correct.

        • EndlessWaves
        • 1 year ago

        Yeah, Steam’s numbers for AMD cards have always been highly dubious.

        For example the fifth most popular AMD card according to steam is the HD8800. If you don’t recognise it then it was a limited release OEM-only card. Steam claims it’s more popular than the 7850 and 7870 combined.

          • Pancake
          • 1 year ago

          I think you should apply Occam’s Razor. Maybe it’s just the truth. Why would Steam have any particular bias?

          Steam isn’t claiming the HD8800 – whatever it is – is more popular than 7850 and 7870 combined. It’s claiming that of the current PC’s using Steam for gaming the HD8800 is more used than the 7850 and 7870 (combined). So, to Occam’s Razor. Would you be gaming with a 7850 or 7870 in May 2018? Apparently not. It’s possible those GPUs are seeing out retirement in mum and dad’s web-surfing build. Or in the case of my 6950 sitting in a box for years as I have no use for it.

            • JustAnEngineer
            • 1 year ago

            My mother is using my old Radeon HD7870 today. She plays many Steam games, but they’re not very graphically challenging (e.g.: [url=https://store.steampowered.com/app/388450/Spooky_Bonus/<]Spooky Bonus[/url<]).

            • sweatshopking
            • 1 year ago

            I’ve got a 7870 in my sons pc. Brother has 7870xt, and a 7850. There is no reasonable way to think that the 8k series is in more pc’s today.

            • Pancake
            • 1 year ago

            Cool story, bro. You’re also smudging facts. The Steam survey is about PCs with Steam installed on it. Anyway, who cares about this ancient history old rubbish. Fact: 1.79%

            Question: do you dispute the Steam survey?

            • YellaChicken
            • 1 year ago

            If he does dispute the Steam survey are you going to act like an arse, wave your science degree at him and quote “facts” from a survey that has already been shown to be flawed in many ways countless times before?

            Over and over again in this thread you’ve ignored the fact that a lot of data in the Steam survey should be taken with a bucket of salt and while some changes have been made to correct some errors it doesn’t mean they’re actually fixed or that more errors aren’t lurking around the corner.

            You’re correct that analysis of the data as presented shows AMD cards aren’t popular. And undoubtedly they’re not as popular among gamers as Nvidia cards. But do you really fail to recognise the distinct possibility that the figures you keep shouting about are unlikely to be accurate? Fact: 1.79% is quite possibly incorrect.

            As an example, I wonder why the survey only shows Intel CPU speeds? Maybe from that data we can deduce that there isn’t a single Steam gamer using AMD CPUs? Or maybe, more likely seeing as AMD CPUs do get counted, the Steam survey just can’t read AMD CPU data correctly? Perhaps it can’t also recognise some AMD GPUs correctly either? Maybe the cards with dual bioses for mining don’t get picked up correctly? Maybe there could be issues with Steam reading the latest drivers? Or maybe it is accurate and you’re right but with the info in this article and all the past discussions on this site, and right across the web, about issues with the Steam survey there’s plenty of reason to doubt that figure.

            Question: do you blindly believe that everything about the Steam survey is 100% accurate?

            • Pancake
            • 1 year ago

            Umm, I suppose I have to wave my credentials around because there doesn’t seem to be very much objectivity in the arguments thrown against me.

            Fact: I’m unbiased and objectively quoting from the Steam survey – perhaps the world’s most reputable and important PC gaming hardware survey. They change their methodology to suit changing circumstances and are quite transparent about it. You’re just throwing around a whole lot of FUD like a few other people here to slander them. FUD not based on any quantifiable fact.

            Fact: 1.79% in the highly reputable Steam survey.

            Who knows why they’re only measuring Intel CPU speeds. Maybe because it’s a consistent measure of something. I don’t think it’s indicative of a deep bias or hatred of AMD CPUs as you might suspect. And CPUs are not GPUs. Both are quite trivial to identify through various system calls. Source: used to be video card driver developer who’s won a major international award for his work.

            Question: why do you doubt Steam? It’s a highly reputable company with some of the world’s smartest software engineers working there. And they are the largest distributor of PC games in the world. The fact that they put out their survey data, including caveats, is a very generous service to the gaming community. They don’t have to do it. Thank you, Valve. Thank you very much.

            • YellaChicken
            • 1 year ago

            Could you miss the point any further than you already are?

            Fact: There’s nothing biased in anything anyone’s saying to you, quite simply many of us doubt the accuracy of the data. It’s quite simple (like you).

            Fact: Your analysis of the data is absolutely sound. However if you put crap in, any analysis you do will give crap out. Bad data will skew results. Simple

            Fact: I doubt Steam because after the next survey they’ll find something else wrong with it. It happens all the time. Being the biggest survey doesn’t make it reputable.

            Fact: I didn’t imply any bias over the Intel CPU speeds. I just suspect that they’d report all CPU speeds if they were reliably getting accurate data from other CPUs. And as trivial as your incredibly convenient experience in the field tells you it is, I’ve seen mistakes get made by software when trying to read hardware data, with my own 2 very reputable eyes. I’ve seen apps like 3d mark, gpu-z, cpu-z, occt misread CPU types and speeds, GPU types and speeds, memory, voltages and other types of hardware data. And yet these are all reputable applications. I’m sure they’d all love to hear you think this stuffs trivial, but I guess you know better than them.

            There’s tons of objectivity being presented to you. You’re just so busy fawning over your “reputable” data you’re not taking anything else in.

            • stefem
            • 1 year ago

            What are you talking about? there are also AMD CPU speed listed in the survey results

            • sweatshopking
            • 1 year ago

            um, all those pcs have steam installed.

            • Pancake
            • 1 year ago

            That’s even scarier. It further emphasises the point that AMD GPUs aren’t being used by gamers playing the latest games.

            1.79%

            • Waco
            • 1 year ago

            My buddy uses a 7870 today thanks to his GTX 780 dying. It play Far Cry 5 well enough, surprisingly.

            It’s temporary but it’s totally doable if you’re not trying to max out settings.

            • synthtel2
            • 1 year ago

            On top of HD 88## cards never having been available through nearly as many channels as HD 78## cards, it brought a lot less (nothing but marketing) new to the table to drive sales, and only existed as latest-gen in any marketing for 7 months (January to August 2013) as compared to 10 months for 78## (March 2012 to January 2013). This doesn’t directly make Steam’s numbers bogus or anything, but it’s weird enough to warrant some further thought.

            (Also, while Pitcairn isn’t going to set any records, it remains a perfectly competent chip today.)

            • Pancake
            • 1 year ago

            The facts are as they are. FWIW, for anybody technically curious it’s rather a trivial thing to query your PC for what graphics card is installed. I have no doubt the highly competent programmers at Steam can get that right.

            1.79%

            • Voldenuit
            • 1 year ago

            Since the 88xx series was just a rebadge, it was probably an issue of how Steam identified the cards, or how the driver reported the hardware to Steam.

            If you add up the 88xx and 78xx cards in steam hwsurvey, the sum total was probably somewhere in the ballpark of real-world systems.

            Also, we at TR tend to over-represent DIY builds. There are probably more people gaming on OEM systems than we would infer by extrapolating from TR users and our close associates.

        • Pancake
        • 1 year ago

        You have your own little anecdote – your little microcosm. And then you have bulk statistics carefully gathered and analysed by the largest PC gaming distribution company in the world. I’m not saying you should be happy with the truth – absolute facts. But don’t fall into the trap of creating your “own truth”, your own set of “alternative facts”, your own narrative. That way lies madness. If you look at my hardware purchases over the years you could only conclude I am a rabid AMD fanboy. But facts are facts. The truth is the truth. It doesn’t help to crawl into a world of self-delusion. Are you man enough to face the truth?

          • Goty
          • 1 year ago

          *reads article and makes his way to the comments*

          Pancake: THE STEAM HARDWARE SURVEY IS INFALLIBLE!

          Wait, what?

            • Pancake
            • 1 year ago

            So, you can’t handle the truth? I’ll let you speak to everyone about your character and intellect – or lack thereof. Will you be cancelling your Steam account because they “dissed yo team”?

            Reality is what it is. I base my opinions purely on objective fact. Statistics.

            Reality: AMD is freakin’ doomed with their GPU gaming usage statistics particularly for their current gen cards. Now, let’s accept that fact and move on to a productive discussion on how they might get their way out of it. I don’t know how they can. As a former developer of GPU graphics card drivers (OpenGL and DirectX) I can’t see their way out of their current DX12 conundrum. AMD can keep throwing cash at developing drivers that nobody (statistically speaking) is using and hope that something will change in the future but how does that make sense from a business perspective? If I was in charge I’d focus development on cryptocurrency apps and hardware. THAT will make money in the short to medium term.

            • synthtel2
            • 1 year ago

            Even if we gloss over the weirdness in some of the numbers and accept what we’ve got there as pure fact, I don’t think that “AMD is freakin’ doomed” is fact (even after tossing in some qualifiers like “in PC gaming graphics”).

            (a) Consoles are really not trivial. There’s still a lot of work involved in taking the hardware that can do that and making it useful on PC, but nobody in the industry is forgetting (nor will they be anytime soon) that AMD hardware is out there and competent. A substantial amount of the graphics work done to get a game running well on consoles will help out PC AMD graphics too.

            (b) AMD hasn’t been messing with their architecture much (meaning much lower driver development costs), and has been rolling out new dies pretty conservatively (saving a lot on hardware). If something GCN-like won’t cut it, then maybe it’s time to rethink, but I don’t think GCN is dead yet. (It looks pretty dead because they’re always cranking voltages/clocks as far as they’ll go, but that’s not inherent to it.)

            (c) Crypto is unreliable, gaming isn’t. If they orient everything towards crypto, where’ll they be next time crypto crashes? Assuming they’ve got design wins in 9th-gen consoles and aren’t ditching the fundamentals of GCN, staying in PC gaming may be low-margin at times (and people will continue to wonder why they release partial lineups), but seems unlikely to be unprofitable. When crypto crashes, gaming sales get better. Either way, they get sales.

            • Pancake
            • 1 year ago

            Oh wow. A substantive well considered post. Allow me to retort:

            a) Consoles are not trivial. They also come with their own APIs and unique tool sets and development environments. Just as NVidia and AMD provide their own tool sets for profiling and optimising code for their GPUs. However, consoles (Auxy – pay attention) ARE NOT PCs. At the end of the day it’s not what nostalgic engineers think. It’s about market forces. Who’s going to care too much about the 1.79%. Also, game development is a chaotic and pretty terrible form of software development where horrible hours burnt to get to a deadline is the norm. It isn’t a well-ordered software development process (like what I’m used). Code quality is terrible and documentation non-existent. I know this. I know many game developers – smartest people on earth, the best people – but they got burnt out and chewed up in this terrible industry. Chewed up husks burnt out in their 40’s. Terrible. I myself was offered a job for an Atari branded game developer (premium high end when they were in the Infogrames stable) but turned it down because I didn’t like the work culture.

            1.79%

            b) this is a devilish conundrum for AMD. Basically, GCN SUCKS! It’s horribly energy inefficient compared to NVidia architecture which is evolving at a rate of knots. AMD are kind of stuck in a bifurcated situation. They have to continue supporting console architecture for compatibility and long life reasons. But they also have to evolve in the PC space. Situation catastrophe: resources divided and diminished.

            1.79%

            c) Crypto is most definitely a wild ride. But it’s also where most AMD GPU product is being used. Gaming is the – to use a pun – the long game. But to be in the long game you need consistency in execution and consistency in market uptake which is ultimately what drives software support. Which brings us to:

            1.79%

            • Anonymous Coward
            • 1 year ago

            Your attitude stinks.

            • Pancake
            • 1 year ago

            If you have a problem with unbiased objective facts I can’t help you. Did you do a liberal arts course? Me – I have degrees in engineering and science.

            • Brother Michigan
            • 1 year ago

            Ah yes, that nebulous “science” degree! So distinguished!

            • Pancake
            • 1 year ago

            It’s not about being “distinguished” it’s about having the necessary skills to think critically and logically. And yes, a degree in electronics engineering and physics does trump a journalism or liberal arts major when it comes to a technical discussion.

            I mean, you might be better equipped than I to talk about – I dunno – analysing movie plots or pop culture trends. I’m sure I’m hopelessly outclassed in web design or social networking.

            • Redocbew
            • 1 year ago

            If you care so much about being scientific, then may I suggest you find a way to share that enthusiasm and commitment with other people. This is not it. This is exactly what could turn someone away from a scientific field should they be so unlucky as for this to be their first impression.

            Let me also make a prediction. When the crypto bubble pops it will not form a vacuum that sucks every buyer towards one particular vendor. We have yet to see if Ryzen is just a flash in the pan, and it’s going to be an uphill battle winning back market share in GPU-land regardless of the strength of their future products, but if there’s one thing AMD is good at it’s staying alive in the face of much larger competitors.

            • stefem
            • 1 year ago

            Who are those much larger competitors of AMD? I can think only one, the other has roughly the same size

            • Redocbew
            • 1 year ago

            If you use market cap to estimate, then Nvidia is about 13 times the size of AMD while Intel is about 21 times their size.

            • stefem
            • 1 year ago

            Are you really saying that market cap is a good metrics to evaluate and compare the size of a company? Revenue would be a much better metric even while remaining quite misleading…
            Both company have roughly the same number of employees, they are almost equal in size

            • cegras
            • 1 year ago

            How old are you? Because you’re acting like a sophomore. Highschool or university – take your pick.

            • ludi
            • 1 year ago

            Funny, in my engineering curriculum they required us to take a semester or two in Engineering Statistics. You may have missed it, but it’s this nifty course where they teach you the basics of quantifying technical data with numbers, including the limitations of what those data actually mean.

            • Waco
            • 1 year ago

            Stop making the rest of us logical thinkers look bad with your crap attitude.

            The survey is not reputable just because it’s the biggest no more than McDonalds food being sold more than anything else makes it good.

            • synthtel2
            • 1 year ago

            That 1.79%, I don’t think it means what you think it means.

            For a start, it’s missing the 540, 550, 570, Vega 56/64, 480M, Ryzen iGPU parts, and Kaby-G. Those aren’t going to double it or anything, but if you’re going to throw around a singular number in that manner, making said number ~1.5 decimal places more precise than it deserves to be while simultaneously rolling with a known-low estimate just makes you look like you’re pushing an agenda (badly).

            More importantly for the actual content, Polaris just isn’t that different from the original round of GCN cards as far as things game devs need to pay attention to. Vega has a couple more tricks available, but if you don’t use them it isn’t gonna complain. When setting up a game to support different cards correctly, most of the work done for anything both AMD-graphics-specific and PC-specific is going to apply to every GCN part ever made. This doesn’t mean devs shouldn’t bother testing on everything under the sun, but (as you said) game dev is a mess; it’s already common for some hardware or other to go insufficiently tested, and it isn’t the end of the world.

            Given the above, the business choice is not “do we support Polaris and Vega”, it’s “do we support GCN”. Considering consoles are already forcing devs to be in the right general ballpark of GCN-friendliness, nobody in their right mind is going to decide to ignore all of GCN.

            //////////////////////////////////////////////////////////////////

            GCN is a bit behind, but mainly just looks bad because it’s usually seen OC’d to the moon. Tuning it so hot is probably good for AMD in the short term, but I doubt it is in the long term, because everyone’s so used to it that they think it’s a problem with GCN. As far as the bit they are legitimately behind, there are plenty of ways they can continue to evolve it, and having to continue to support consoles is a problem for few of them.

            Consistent execution helps, but this market shifts quickly when one or the other comes out with a better product. R300, G80, RV770, Maxwell – it isn’t at all unheard of for one side or the other to look unbeatable, yet they’re both still here.

            • Pancake
            • 1 year ago

            The 1.79% is exactly what it means – the current gen AMD cards listed on Steam’s usage charts. Now, their list bottoms out at 0.15% with everything below lumped into the “other” category. So, even assuming 540, 500, 550, Vega 56/64 etc are at 0.15% each (highly improbable) you’re barely gonna get to 3% which would be somewhat less than the GTX1070 all by itself.

            I make no comment at all to the quality of either AMD or NVidia GPU solutions. I am purely commenting about the market situation by analysing Steam’s statistics and the upcoming “demographic” hole of doom AMD is facing. If you want a healthy market then the present situation of 32% vs 1.79% is absolutely unbalanced and dire. That’s what merits discussion. Not trying to divert by saying Steam’s statistics are unreliable or that _I_ am somehow an anti-AMD agent. The truth hurts sometimes. TR will not develop if the readership cannot evolve to a factual discussion and instead rally around like partisan monkeys flinging poop (Goty I’m looking at you).

            • synthtel2
            • 1 year ago

            Yet you’re still discussing the things you say are irrelevant (and that I didn’t actually say) and not discussing the things you say are relevant (beyond repeating what’s already been said). How is that supposed to lead to a factual discussion again?

            • Redocbew
            • 1 year ago

            +3 for princess bride reference.

            And for science, and numbers, and stuff.

            • Goty
            • 1 year ago

            [quote<] I'll let you speak to everyone about your character and intellect - or lack thereof.[/quote<] I think you're speaking enough to your own character and intellect for the both of us.

            • steelcity_ballin
            • 1 year ago

            posted link to wrong chain, removed. Need a “delete post” option maybe?

          • sweatshopking
          • 1 year ago

          Let’s just remember when claiming “facts are facts”
          that steam just updated this thing BECAUSE IT WAS TOTALLY COUNTING WRONG.
          🙂

            • Pancake
            • 1 year ago

            Science is about improving the ability to measurement. To move forward with improving techniques. Steam get credit for improving their technique – and in a way which favours AMD products. Well might you wallow in the stinking pit of fear, uncertainty and doubt. It’s not a good place for you and doesn’t lend you any credence or respect except from the (small) echo chamber of deniers.

            • sweatshopking
            • 1 year ago

            i’m waiting for you to come out and say this entire thread you’ve been trolling and don’t actually believe what you’re writing.
            before you qq at bias, my pc is running a 1060 6gb, which I quite like. I also have an r9 290, which I also quite like.

        • maxxcool
        • 1 year ago

        I do not think it is accurate either. *BUT* I do believe the red/green divide is dramatically unbalanced in Nvidia’s favor… we just cant get a clear measure of it due to poor data collection techniques, and to large a field of what can qualify as a gaming pc.

        I believe the red teams superior compute/currency ability has landed large overall percentages into rigs that will never see steam.

        • steelcity_ballin
        • 1 year ago

        Darnit, meant to post this to your thread, not Pancake’s.

        [url<]https://imgur.com/a/JZEm81d[/url<]

      • auxy
      • 1 year ago

      Don’t forget that every major game console in service now is based on AMD hardware. That’s a very large install base. Don’t worry, games will continue to be optimized for AMD graphics. (‘ω’)

        • FuturePastNow
        • 1 year ago

        Yeah. Microsoft and Sony are using GCN.

      • Chrispy_
      • 1 year ago

      I think you may be overreacting, but agree with you to some extent; The cryptominers dried up the supply of Polaris cards within 6 months of their launch, and the prices went out of range for most gamers shortly after that.

      For a very long time, AMD had nothing gaming-capable available at reasonable prices, whilst Nvidia were still selling 1060s like hotcakes, with only the 1070 starting to climb in price for several months once all the potent AMD cards were unavailable for Etherium. Even now, Vega is priced insanely, whilst both the 1070Ti and 1060 6GB are available at something near MSRP and in stock at multiple retailers.

      Essentially, Nvidia was hit by crypto at the high end (1070, 1080Ti) which is a very small part of their overall marketshare. AMD was hit by crypto right in the heart of their key market demographic.

      At some point, every single Polaris and Vega card that the Eth miners removed from the gaming pool will appear on the used market, and it’ll hurt both Nvidia and AMD’s new sales figures, but again – AMD will be making money from the consoles, and Nvidia will be making money from the Tesla/Grid/Quadro end of the market, so both will probably be affected less than you think.

        • Pancake
        • 1 year ago

        Yeah, cryptomining undoubtably has had a huge effect. But, I think the interesting discussion is what happens post-mining. The situation is absolutely dire for AMD because their new GPUs simply ARE NOT used for gaming. Fact.

        No doubt there will be a glut of somewhat used Polaris and Vega cards in the near future but I don’t see that having much of an impact excepting low-end gaming. By the time they’re sold they will be undesirable and anybody wanting an upgrade will be out for the RX680 or GTX 1180. A perfect storm. More like Stormy Daniels 2018 vs Perfect Stormy Daniels 2006.

          • anotherengineer
          • 1 year ago

          I use my RX 480 for gaming 😉

      • LostCat
      • 1 year ago

      Counting only modern capable GPUs from AMD against the rest of the Steam market without cutting out the other vendors older or incapable GPUs is kind of silly.

        • Pancake
        • 1 year ago

        Well, the discussion is about AMD’s dire “demographic” challenge. But if you want the NVidia roll call:

        GTX 1060 12.29%
        GTX 1050 Ti 8.60%
        GTX 1050 4.59%
        GTX 1070 3.69%
        GTX 1080 2.23%
        GTX 1080 Ti 1.17%
        GTX 1070 Ti 0.40%

        which makes for a total of 32.97% of the entire Steam GPU surveyed PC installation.

        In my view that’s a pretty healthy “demographic” mix. In years to come the old NVidia GPUs (such as my ancient GTX 970) will be replaced by a healthy younger population of 1000 series GPUs. One can only imagine the same will roll on with the 1100/1200 series. The roadmap looks sunny and good. Meanwhile, in the AMD team they’re struggling with how to evolve the rubbish (for gaming efficiency) GCN architecture.

        32.97% vs 1.79%

        How ya feelin’, Lostcat? Feelin’ better now I’ve done the maths?

          • LostCat
          • 1 year ago

          So that’s over 5%. Add the 970 and 980 and the 280, 285, 290, 380, 390, and whatever Vega has and I’ll think about it.

          I’ll give you that NV does have a strong PC market right now. As has been said though, ignoring that a lot of AMDs optimization has already been done on consoles doesn’t make sense.

          I’d like to see how Vega would’ve done if it was actually available.

            • Pancake
            • 1 year ago

            Umm, no. I’m not going to add previous gen cards. The point I’m making is about current gen cards and about the near complete absence of AMD in gamers’ rigs.

            And it’s not 5%.

            It’s 32% vs 1.79% (OK, OK, let’s round it up and call it 2%).

            • LostCat
            • 1 year ago

            Right, which, if you make that 100% of the cards we’re talking about is 5%.

            You were talking about math, so I did some. Fascinating, I know.

            • Pancake
            • 1 year ago

            Ahh, you coulda said. Nice work, kitty. Have a bowl of cream and tummy-tum rubs. It is another way of looking at the numbers so let’s put it out there.

            So, of the current gen gaming GPU market:

            NVidia: 95%
            AMD: 5%

            O_o
            O_O
            x_x
            o_O

            Wow. Just wow.

            • LostCat
            • 1 year ago

            You don’t look like a catgirl. Don’t touch me.

            When I consider the 120m or so consoles AMD has going on I think they can afford to let PC slide a bit, and we’ll have to see how their 7mm and Navi stuff does.

            With the trouble I’ve had with my current machine I could probably get a 7nmVega or Navi +Ryzen 2 build going when the time comes.

            • Pancake
            • 1 year ago

            So, without so much as a bang or a whimper AMD has effectively capitulated the PC gaming space and left almost all of it (95%) to NVidia. That is just absolutely fascinating.

            Yes, AMD has the console space at the moment. Which basically means 2 major GPU gaming customers. It is a precarious position especially as console manufacturers seem to completely change things every generation in a space where compatibility doesn’t seem to matter.

            I could see some evolution of Ryzen + beefier IGP being in the next gen consoles. But, equally, if NVidia wanted to spoil AMD’s party they could throw a bit of their bountiful piles of cash and win over one or both Sony and MS. Heck, NVidia could take over Sony if they wanted to.

            Interesting times.

            Edit: imagine NVidia coming out with some sort of ARM/GPU combo scaleable from low-power mobile to plugged-in console. Like a super-Tegra. Not one chip but a series of chips but software compatible. So you could have a transferable gaming experience from a small hand held device with reduced details/quality to the full-fat 4K full-effects experience. Sony, please make it happen.

            AMD would be standing there with a sad face.

            • LostCat
            • 1 year ago

            Compatibility? I don’t know, I can play tons of Xbox and Xbox 360 games on my Xbox One X with much better quality than the original systems were capable of.

            Also worth noting we don’t have numbers for Origin or battle.net. Probably not too different given how late Vega was to even be available at reasonable price points (it seems to have come down, but still isn’t quite where it should be.)

      • Krogoth
      • 1 year ago

      People have been getting AMD GPU SKUs for gaming purposes and do some mining on the side too. It has been a small minority ever since GCN 1.0/Kepler-era.

      It is simply dwarfed by number of Nvidia sales. Their 970, 1050 and 1060 sold like hotcakes and make up the bulk of discrete GPUs on the market. 1070 is by far the most popular “high-end” SKU.

      Current titles are actually optimized well for GCN archtecture since PS4/Xbox One utilize it and gaming consoles are used as the baseline for game developers.

        • Pancake
        • 1 year ago

        Agree. It’s all relative. I mean, even 0.63% out of a population of 100 million (pulled figure out of my thin air) is 630,000. A huge number. However, compared to 12% out of that same population then there’s no wonder why NVDA is worth $145B compared to $10B for AMD.

      • lemonhead
      • 1 year ago

      And still we have tons of releases and reviews of Freesync monitors. I get Gsync is expensive to license, but Freesync is useless to me. I don’t see Nvidia adopting it anytime soon.

        • Chrispy_
        • 1 year ago

        I wouldn’t mind quite so much about GSync if when you paired an Nvidia card with a 75Hz or 100HZ monitor, it didn’t just ignore the monitor’s refresh and render at 60Hz anyway.

        Almost every Nvidia GPU and >60Hz monitor have failed [url=https://www.testufo.com/frameskipping<]this test[/url<]; It's really sad because 75Hz is noticeably smoother, but you're better off just dropping decent monitors down to 60Hz if it's not GSync, just to avoid the hideous frame-skip micro-stutter. At 120Hz, Nvidia's problem goes away again, but many of the nicest screens are 75Hz or 100Hz.

      • anotherengineer
      • 1 year ago

      I wonder if crypto miners, individuals and businesses have steam installed on their mining rigs??

      As for the low steam %, ya who would pay $600 for a $200 gaming card??

      I also typically block the steam survey on my pc, so you can add 1 more RX480 if you like 😉

      • uni-mitation
      • 1 year ago

      Your underlying assumptions based upon the results of this poll are flawed at the least.

      1- You may be advised that you should never add [url=http://mathforum.org/library/drmath/view/57604.html<] or work with percentages [/url<] unless you know for sure that they are based upon the same ratios. 2- Statistics, or in this case, distribution of things in a population: You make the claim based upon this one poll that it is the current state of current gen usage. Yet, you fail to do any statistical analysis, or anything like that. You pull this number as it was given by God himself. You make the mistake and you don't realize that this poll is a sample population of ALL video game cards in the whole world. Therefore, you need to do some math and come up with a distribution of the population and a margin of error. Just like they do in Presidential elections. 3- You have no qualms in accepting this one poll as truth yet you don't talk about the method in which this data was collected. Was it an opt-in to the poll? Was it totally sampling of the data? Please explain. 3.5- People that work on this field compare and contrast the accuracy and precision of one poll with other "well-established" polls, and measure their deviation, variance, and adjust the results with the other polls to make the data better. Please look at this [url=http://fivethirtyeight.com/features/how-were-tracking-donald-trumps-approval-ratings/<] this site's top-notch methodologies [/url<] to see how people in the profession do it. It should be very easy with someone of your educational background. I don't claim any adequate training in statistics & probability except for a course titled "Statistics & Probability for Engineers. " So you have a leg on the competition. I am looking forward to your take down of how we are all wrong. 4- You have the burden to prove your claim about the current distribution of current Gen cards. You have not met that burden, and at least I don't feel the need to do anything but say your case is sorely lacking. uni-mitation

        • Pancake
        • 1 year ago

        1) I don’t even know what you’re trying to say. Can’t parse it. OFF COURSE you add percentages when it comes to a proportion of a population. You really don’t math, do you?

        It’s like this – there’s 100 fruit in Gametown. Johnny AMD has 2 apples and a banana. THEREFORE Johnny has 3 fruit (2 + 1) or 3% of the fruit in Gametown. What don’t you understand??

        2) Steam done the math. Yes, I have acknowledged it’s a sample population of ALL video game cards. But it’s the population of ALL Steam users – people actually likely to be actively gaming. As opposed to the old retired 7850 you might have given to your grandma for her web surfing hand me down PC.

        3) Steam poll isn’t an opt-in poll. You install and use Steam you’ve participated in it. Thanks for adding to the picture. It is – as far as I understand it – total sampling of all Steam users. I can’t even imagine what you’re thinking in your further points – what sort of nightmare scenario. Do you somehow think NVidia users are trying to rig the Steam poll? There’s some sort of Russian conspiracy behind this?

        Really, why should the burden of proof be on me? I’m not getting paid to do some deep analysis of Steam’s methodology or results and I don’t have access to the raw data.

        Like many people reading this article you’re deep in self-denial and flinging FUD at Steam – the leading PC game distribution company in the whole wide world.

        The results are as presented. There is something unbelievably scary and fascinating going on. I hope some journo will do some investigation into it. It really merits investigation rather than denial. Might not be TR though – unfortunately – as it will drive too much of the one-eyed readership into a frenzy.

        1.79%

          • uni-mitation
          • 1 year ago

          It is plain to see that you have failed to back up your claims. Thank you for the exchange.

          uni-mitation

            • Pancake
            • 1 year ago

            Backed up what? A simple analysis of the world-renowned Steam Hardware Survey? Your rambling discursion in your previous post showed little but innumeracy and an inability to string together a bunch of coherent thoughts.

            Why don’t you go prove there’s something deeply flawed with it? I’ll leave you with something to ponder about:

            1.79%

            • rechicero
            • 1 year ago

            Yeah, I think “simple” is a good definition…

      • Kretschmer
      • 1 year ago

      It’s fairly likely that the mining tax has depressed AMD’s gaming adoption over the past few years, but <2% seems rather low.

      I can’t search at work, but has any game dev recently posted a blog or AMA where they mention the vendor breakdown? That would be a good place to sanity check the Steam numbers.

      • BurntMyBacon
      • 1 year ago

      Disclaimers (feel free to skip directly to discussion points):
      [i<]For the purposes of this discussion, I'm going to put aside the finer points of statistical analysis as I don't believe a finer analysis will change the trend vs the current gross analysis discussed in this thread. For instance, the 1.79% figure is, by Steam Survey definition, low as it is not fully inclusive (you defined current gen cards). However, to your point, the missing GPUs make up a fairly insignificant portion of the population that in the best case would bring the total to ~3%. Not a large difference in the grand scheme of things and it would take some massive error margins to change this. Many in this thread are hesitant to put as much faith in Steams numbers as you are. The reason has little to do with the qualifications of Valve's software engineers or the appropriateness of their methods. The issue is simple observation. Valve has identified issues with its data collection that have significantly altered the presented data multiple times in the past. While this is a good thing in the sense that it shows that Valve is constantly evaluating and adjusting their methods to get more accurate results, it also leads people to believe that this historical precedence will continue. It doesn't help the perception when their anecdotal evidence (and also mine coincidentally) is at odds with the results. However, I have not seen a better data source presented here. I have yet to see any alternate sources at all suggested in this thread. [b<]So, for the time being, I'll accept the Steam survey as the most representative data currently available.[/b<][/i<] Discussion Rather than argue percentages here and there, lets discuss your talking points. If I understand you correctly, the point you are making is that AMD's current generation cards don't have much of a footprint in the gaming scene and, consequently, are no longer targets for optimization or at least at risk of loosing this status. While I do believe AMD has a problem here, I believe their position is not yet as dire as you portray. Of course it is not nearly as rosy as some others would like to believe either. Some things AMD has working for them: 1) Similar architecture across generations - This is probably their greatest asset working for them. While it doesn't necessarily guarantee the best optimizations for the latest generation GPUs, it does grant the newest generation a certain amount of optimization due to support for previous generations. 2) Console ports - Optimizations for consoles require hardware usage, frame analysis, and timing analysis tools that are in many ways very specific to the console it is running on. Some of these consoles have resources not common to PCs. Consequently, the optimizations don't directly port to PC either. However, there are commonalities in what the architecture is geared for and developers will avoid techniques that the architecture is simply poorly suited to. This does benefit AMD GPUs to some extent. 3) Low level APIs - AMD invested early in low level APIs and have had a rather large influence on DX12 and Vulkan. This seems to have paid off with early DX12/Vulkan titles being measurably better optimized for AMD hardware. However, while AMD had a head start on nVidia in regards to low level APIs, that lead has all but disappear and the two can be considered at parity. While it is likely that low level APIs will remain relatively strong for AMD, they can no longer push it as much of a discriminator. Some things AMD has working against them: 1) Similar architecture across generations - While this can be considered a good thing for now, it get developers in the habit of ignoring the finer details of their hardware. As soon as a significant departure from the existing architecture occurs, optimizations start from zero. 2) Market/Mindshare - Having low census in the gaming crowd may be workable in the short term, but a continuously low census will, as you fear, result in your platform no longer being considered a target for optimization. Given the industry's tendency to optimize to the lowest common denominator, I don't think we can discount 2xx/3xx series cards yet, but if AMD doesn't get some traction with a newer generation card before they loose relevance, things will get tough. 3) Developer Relations - AMD has never had as much success getting developers on board as nVidia. This situation is improving due to console wins, but they need to keep working these relations if they want their platform to remain targets for optimization.

        • Voldenuit
        • 1 year ago

        AMD’s driver team also needs to get their act together.

        Currently, there’s a Radeon bug that makes load times in Destiny 2 insanely log if anyone in the game you’re matched with has that driver. It’s been there for at least a month.

        Meanwhile, nvidia’s 397.31 driver had an uninstallation bug that was fixed within a week with the 397.55.

        • uni-mitation
        • 1 year ago

        0- I respectfully object to having the Steam survey passed as an accurate & precise representation of the population or any statistical inferences. This is an issue of math, and the underlying mathematical proofs that support such inferences. This is a matter of what is able to be disproven, and not a matter of feeling or opinion.

        1- “The goal of a statistical investigation is the improvement of a process. That point has been reiterated often through this text, and it is the main point to keep in mind as we review and extend the ideas involved in the design of experiments. ” Paragraph 1, page 609 of Probability and Statistics for Engineers, Edition 4, by Richard l. Scheaffer/James T. McClave

        It further goes to state and use an appropriate example:

        “Computer chips (semiconductors) are manufactured in a round wafer (about 3 inches in diameter) that holds many chips. The wafers are scored and broken apart to retrieve the chips for actual use in circuits. This process is rather crude and causes much waste. The question for investigation, then, is ‘How can the breaking process be improved to gain a higher yield of usable chips?’ Suppose an engineer suggests a new method, call it method B, to compete with the standard, method A. These two methods form the treatments to be compared in an experiment. ( After all, we cannot [b<]take the word of the developer that the new method is better!).[/b<] " Paragraph 6, page ID, outlining a concrete example of two treatments, and how must we be very meticulous that the design of such experiment follows the scientific method. Emphasis added by me. 2- Experiments are conducted in all types of research to determine whether certain "treatments" are better than other. Math, specifically statistics, and statistical inference is one of the tools that help us make inferences about whole populations from some sample size. That is how we are able to have some sort of forecast of presidential elections by having a sampling distribution serve as a better approximation meanwhile the sampling has maximum amount of randomness. "The gold standard of statistical experiments is the simple random sample. In such a sample of size n individuals, every member of the population has the same likelihood of being selected for the sample, and every group of n individuals has the same likelihood of being selected." [url=https://www.thoughtco.com/what-is-statistical-sampling-3126366<] Courtney Taylor, PH.D in Mathematics [/url<] But Courtney also warns that: "Good simple random samples require some work to obtain. If our data has been obtained haphazardly and in a cavalier manner, then no matter how sophisticated our analysis, statistical techniques will not give us any worthwhile conclusions." 3- Steam's methodology of sampling is not clearly structured, we are not privy to the algorithm that they use to count the machines and henceforth we are unable to judge by ourselves and not take their word of their methodology's fitness as a true representative sample of the population. Futhermore, [url=https://store.steampowered.com/hwsurvey/?<] Steam [/url<] states that "Participation in the survey is optional, and anonymous." We are unable to judge what figures they have, how is the sampling able to overcome and take away this obvious bias except by having a totally randomized sampling of the population? 4- In conclusion, I have cited authoritative and persuasive evidence & reasoning as to why I categorically object to the usage of this survey to draw any meaningful statistical inferences that are able to withstand statistical rigorous scrutiny. 5- I welcome anyone that is willing to take his/her time to prove me wrong with likewise authoritative reasoning & evidence to the contrary. I will gladly accept that my position is incorrect, and learn something new! If you have feelings or opinions, then feel free to share them, but I feel no compulsion to answer to such vacuous statements. uni-mitation

          • Pancake
          • 1 year ago

          So, you dusted off some old text on your bookshelf and typed out paragraphs from it. I think you’d be better off starting with the “understanding working with percentages” post on Math Forum you linked to earlier and studying my Johnny AMD at Gametown fruits example.

          The only thing you have against the world renowned Steam Hardware Survey is that you don’t know how it works. Hey, neither do I. As you say, we’re not privy to it. In some sense our critical analysis stops here. There’s no need to impugn bias. No need to fling FUD. Do you somehow think that NVidia fans are more likely to bias the survey than AMD fans. Judging by the outrage in the comments here I would have said the opposite was more likely… But, none of this is objective.

          What I would like to see is some investigative journalism, perhaps a reach out by – oh I dunno, perhaps a TR staff member – to Steam to do a deep dive into how the survey works. It would be very illuminating and be of benefit to the tech audience beyond just TR. It could be big. Bigly big. Yuge.

            • Chrispy_
            • 1 year ago

            Oh cool, This particular line of threads is all text walls!
            I’m on a phone, can someone TLDR everything for me? Thanks.
            FYI I have an attention span of eight seconds, so no more than 50 words please.

            • Voldenuit
            • 1 year ago

            Chrispy. Pancake. BurntMyBacon.

            I don’t know what uni-mitation is on about, but suddenly I want breakfast.

            • Chrispy_
            • 1 year ago

            Yes – Steamed crispy bacon pancakes. [i<]Sooo[/i<] much better than multiple walls of text!

            • uni-mitation
            • 1 year ago

            ^_^

            Thank you and Voldenuit for making this the cherry on top.

            uni-mitation

        • Pancake
        • 1 year ago

        What an excellent, well considered and even-balanced post. Way better than my antagonistic “stop the presses!” hyperbole. But I see you have no upvotes. What does that say about TR readership?

          • Redocbew
          • 1 year ago

          It means we care more about breakfast than walls of text? It’s the most important meal of the day after all.

          This morning I must have waffles.

          [url<]https://www.youtube.com/watch?v=9jhrA-DdiOk[/url<]

      • Cuhulin
      • 1 year ago

      There are two fundamental problems with your post, even assuming that the Steam survey data is to be believed: (1) The methodology you use to say “1.79%” is simply bad math; and (2) you jump from a portion of the AMD cards’ market share in the past to claiming that no one will optimize for them in the future, which is missing the point that optimizations will be done in the future based on relevant market shares then.

      The methodology problem is particularly troublesome, because it fits in the category of “lies, damn lies, and statistics”. Steam’s numbers break down a population that adds to 100% just like they should, and AMD has a market share in those numbers of 14.89%. To get the low number you want to cite, you write off the supposedly old cards’ 13.1% market shares and then say that the small group of more recent cards you cite have only a 1.79% market share. However, that 1.79% is no longer out of a population that adds to 100%. This would not, in itself, result in a huge increase, but when you make the subtraction you make for AMD’s older cards, it is only reasonable to make a similar subtraction for the older Nvidia and Intel cards, of which there are a large number, some with substantial percentages. Put simply, you have not applied consistent statistical analysis to the newer cards from all the companies involved to compare cards from 1 to 2 years old across all the companies involved.

      I am not saying that there was not a decline in AMD gamers’ market shares in that time period – no doubt there was, at least in part due to the fact that AMD did not ship a card comparable to the 1070 until the Vega cards, but every company has dips and rises in market shares at times.

      Those changes over time are most important when one projects into the future, as those are what will be involved with optimizations to be done. Consider the increasing movement to variable refresh rates. AMD supports monitors and, essentially starting this year, televisions with VRR standards that don’t require significant cost increases for the monitor and television manufacturers. That is likely to result rather quickly in a tangible benefit for AMD cards compared to Nvidia cards at any given price point – you know, what leads to increased market shares. If I’m developing a game now, I expect that the market for people actually buying my games for performance is going to be substantially correlated with people buying hardware for performance – i.e., the people who reasonably would be buying AMD more in the future due to better monitor performance.

        • Pancake
        • 1 year ago

        Thanks for your considered post. Adds to the discussion.

        1) It isn’t about maths, it’s just adding up simple arithmetic. Nothing controversial whatsoever (unless you’re like uni-mitation who doesn’t understand addition). What you can conclude from that is what is up for debate. If you look at the Steam Hardware Survey, of the GPUs listed 1.79% is current gen AMD cards. In contrast 32% is current gen NVidia.

        I mean, sure, if you want to consider all GCN population is targeted for optimisation. Fine, whatever. That really isn’t the interesting point. It’s about the “demographic” trend. It’s a bit like in the Handmaid’s Tale where few children are being born. The overall total population might seem fine but there’s a massive hole in recruitment of new generation.

        1.79%, 2%, 5%. Put in a margin that suits your comfort zone. Whatever. Then look at the NVidia 1060 that garners a total GPU figure of 12+% in Steam all by itself. Scream denial all you like but that’s a figure that should rightly scare the hell out of AMD.

        Tech Report editors – here’s a challenge. Investigate the state of gaming GPUs. What’s being used. Investigate but maybe go beyond Steam. The gun is smoking. Go dig up the most interesting story of the decade in PC gaming.

          • Cuhulin
          • 1 year ago

          I don’t mind comparing the 1.79 to 32 (taking your word on the additions). At least that is apples to apples, and it is a heck of a lot better than 1.79% compared to most the remaining 88.21%.

          But then one needs to reflect on what makes up the 32%.

          As you have noted, looking at this message and others, two cards – the 1060 and 1070 – make up more than 75% of the Nvidia gaming users in the current Steam survey. Those cards have been on the market for rather a long time for graphics cards, and they’re pretty good cards, so they should have a lot of users. (Note: I use a 1070 in the machine on which I am typing this, for example, because it pushes the pixels to my multiple 4k monitors well. I probably even show up in the Steam survey, because I sometimes play Heroes of Might and Magic III HD on this machine, even it is supposed to be a work machine.)

          Does that scare AMD? I don’t know. I am certain they wish it was theirs, but, on the other hand, they are selling all the Vega cards they can make right now, and probably at higher margins than they would receive if they were only selling to gamers (no need for spiffs, for discounts from their normal prices to retailers, and so on.) It isn’t a bad place to be, for the moment. So one cannot tell from the outside whether they are scared or simply working to do better in the future.

          Your point about optimizations is an effort to say “the current position in the race is what will remain in the future”, however, and as to that, I think you are sorely missing on evidence. Maybe my questioning results from being old. I remember when the dominant business software in the office was WordStar, dBase II, and Lotus 1-2-3. All of that changed.

          Any good gaming company is going to consider those changes when deciding where to target its software. Moreover, very few gaming companies are going to target their code at the 32% of the market anyway – no one wants to leave behind too much of the remaining 68%. Optimization for the fastest cards likely will continue as a kind of marketing – “look what we can do with this program for those who can afford the hardware” – but the next generation of cards won’t be 1060’s and 1070’s – so current numbers will not apply.

          This is why a snapshot of current market shares is not “the most interesting story of the decade in PC gaming” – TR already covers the fact that one cannot buy the most interesting cards – Vegas and 1080 ti’s – for anything short of a huge mark-up. He also covers Nvidia’s apparent attempts, through its recent exclusive branding agreements, for example, to lock in its current lead. (Are your posts a new approach to that? I would hope not, since that would raise serious anti-trust issues.)

          Let’s see where the market stands in a couple of years, ok? Hopefully, by then, we will be able to buy 1270’s or Vega 3’s and see where the market stands.

    • JustAnEngineer
    • 1 year ago

    I expect that many of the two-core systems are notebooks.

    The two graphics cards that took a large step down this month are the same two that mysteriously skyrocketed up the charts when Valve started overweighting internet cafe machines by two orders of magnitude: GeForce GTX960 and GTX750Ti.

      • jihadjoe
      • 1 year ago

      Desktops too. i3s are [url=http://indianretailsector.com/news/computer-market-analysis-laptop-and-desktop/<]incredibly popular[/url<] in other places.

    • chuckula
    • 1 year ago

    It’s a dark secret, but Gabe has never really been able to count higher than two.

    Half life 1
    Half life 2
    Half life 2: Episode 1
    Half life 2: Episode 2

    So basically, Valve was just unable to count more than two AMD systems.

      • Eggrenade
      • 1 year ago

      Don’t forget about:

      DotA/DotA 2
      CS/CS:GO
      Portal/Portal 2
      Team Fortress/TF2
      Left 4 Dead/Left 4 Dead 2

      I think that’s it.

      • Neutronbeam
      • 1 year ago

      Half life 3 confirmed!

        • Star Brood
        • 1 year ago

        RIP. Dead before announcement. Thanks microtransactions.

          • K-L-Waster
          • 1 year ago

          I think it was dead a looooonnnnnggggggggg time before anyone even thought of microtransactions.

      • Welch
      • 1 year ago

      Damn your cheap and good comment warranting my up vote

        • blastdoor
        • 1 year ago

        Ditto

          • jihadjoe
          • 1 year ago

          I see Gabe upvoted you

      • Generic
      • 1 year ago

      I thought it was common knowledge that machines think in base 2…

      • NovusBogus
      • 1 year ago

      Gabe was very strongly opposed to the launch of the Playstation 3. Coincidence? I think not.

        • moose17145
        • 1 year ago

        But he was very much in favor of the Playstation 4, as it is a multiple of 2.

          • K-L-Waster
          • 1 year ago

          Although he still insisted on calling it the Playstation100…

Pin It on Pinterest

Share This