Hard Choices: Ask AMD Part 2, The Answers

Here it is, folks. The answers to all your AMD questions. Well, not all of them. Dropped some, reworded others, added a few twists of my own. The usual. The senior AMD suit in question is Roy Taylor. His official title is Corporate Vice President, Global Channel Sales. That’s right, Corporate Vice President, Global Channel Sales. Soak up the seniority. He’s been at AMD for 12 weeks having spent the previous 12 years at arch enemy Nvidia. So let’s just say he’s got plenty of insight into graphics, CPUs and gaming. Did I mention he is indeed quite senior?

Little bit of house keeping first. Roy was not pre briefed, the interview scenario was highly informal and the device of humour, gasp, was used in some of his answers. So read what follows with a sense of bonhomie and don’t take all of it completely seriously. Or complain about the bloody Poll Tax (that’ll make sense later).

Oh and I’ve flagged up who asked the questions. Apologies if you asked the same question and didn’t get a mention. The editor’s decision is final and all that jazz.

Why does AMD have such bizarre and confusing product branding? (Vandelay)
We used to have great names. Then we sent people to Harvard Business school and they came back and messed them all up.

Do you expect APUs to outperform and replace dedicated GPUs? (Keyrock)
If you look at the history of our business and the trend of greater integration, it’s inevitable. Take audio. Audiophiles will kill me for saying this, but you could argue that [in terms of digital audio fidelity] the job is done. Once the GPU is capable of full photorealism, that job is done, too. And at that point it automatically starts to become integrated. But this is a long way off.

When are you going to ship a PS4-style APU in something like a SteamBox and without the Sony baggage? (james.hancox)
Well, we can’t ship a SteamBox, we’re humble pedallers of chips. We enable technology, we don’t make the boxes. I know it’s very annoying, but as for the PS4 chip, until it comes out, it’s not something I can talk about.

Why is your driver support so shoddy? (PatrickSwayze)
This is my most often asked question on Twitter. Somebody once said he thought our drivers are so bad, they must be written by a moose. I said the problem is that without thumbs, we can’t hit control-enter. In all seriousness, we have an absolute obligation to keep improving drivers. However, to keep every single user happy among the millions of different configurations is a very difficult task. I can say I worked at Nvidia for 12 years and we had all the same questions.

Specifically with regard to games that are sponsored by AMD. People who buy those games have a right to expect them to run not only faster and better on AMD hardware, but also to run faultlessly.

What does AMD think about the optimisation race among triple-A titles. Is it healthy or does it fracture the gaming community? (DigitalSignalX)
In general it’s good, because the competition to support the developer results in better, higher quality games. Whatever money we give them or work we do with them, it results in a better game. However, if it results in shader replacement which is only designed to make the performance better [in benchmarks] and not in interests of the gamer or the developer, that’s wrong. So it can be done right. Or it can be done wrong.

Why does the Mac edition of the 7950 have such a ridiculous markup over the PC version? (FuzzyPuffin)
I don’t know, I wasn’t aware that it did! My bad, I’ll go take a look!

What are your plans to support VR in general and Oculus Rift specifically? (snorkel)
That’s an interesting question. Brendan Iribe [OR’s CEO] is a very good friend of mine. I haven’t had chance to try it yet. And before I comment professionally and personally, I want to try it. Brendan keeps inviting me down, but I just haven’t had a chance.

Will AMD solder chips onto motherboards? (analydilatedcorporatestyle)
Permanent attach and no socket? We have SoC parts that are ‘soldered’ in that parlance. But we’re committed to sockets. In the channel business and the enthusiast space, we’re committed to sockets.

Has AMD given up on the desktop CPU horse race with Intel? It would be terrible if Intel became the only game in town. (Snargelfargen)
Absolutely it would be terrible and good Lord, no. We hope to have some news for you very shortly on that [apologies, legalese disclaimer alert: AMD is in what’s known as a ‘quiet period’ and due to financial markets strictures is limited in what it can say about future product plans, but I’m thinking Steamroller].

What are AMD’s priorities with regard to single-thread performance versus multi-thread performance? How’s Steamroller looking regards single-thread performance? (Parge, my good self)
Both are important. That sounds like an easy-out answer. But both are important. Unfortunately, prior benchmarks favoured one side. A new generation of benchmarks are coming that will better show that both are important.
Apart from the corporate-correct standard answer that we can’t comment on unannounced products, the truth is both have their place, so you can expect us to address both with new products. But you can also read into all this that we put an emphasis on one aspect more than the other [I think Roy’s talking about multi-thread here, folks].

Why isn’t AMD making a more serious push into the mobile market. Intel has thrown itself headlong into the market with Atom. Nvidia has pursued mobile with Tegra. What is AMD doing? (Brun)
In terms of mobile, we certainly have not done a good enough job communicating our products and I’ve been hired to fix that.
We’ve actually done a pretty well in mobile, but we haven’t communicated that. As for tablets, that’s a separate question. Watch this space, I’m very excited about tablets, especially Windows tablets.

Can we have an update on inconsistent frame timing / micro stuttering that some benchmark outfits have been harping on about recently? (particlese)
What happened to my former employer that I loved for so long [Nvidia]? So much effort into such a negative issue! We do take it very seriously. But I believe it’s intellectual rather than experiential. This is something that requires you to dig very deep to expose it, that requires you to run your game at 75 per cent below normal speeds to even see it. It’s a bit like the Poll Tax. I don’t recall anyone calling up and complaining about it!

What are AMD’s wins in the new consoles going to do for PC gaming? And will they help AMD take the fight to Intel and push game engines to support more efficient threading? (Parge and my good self)
It will make a big impact on PC gaming. We’re going to be able to take the benefits of development work and scale. Any rational and logical examination will say that if games developers are developing for our hardware, there will be a natural benefit. Having said that, it’s important to be clear that we do not support proprietary standards or features. We will continue to introduce new technologies and improve gaming, but we won’t do anything proprietary.

As for competing with Intel, if you look where Intel is going in terms of GPU investment and the selection of an APU by Sony [for the Playstation 4], all that reflects the fact that heterogeneous system architectures [google it, folks] are the future. I believe that so sincerely that I voted with my own wallet and my family’s future by joining AMD. The future is HSA.

Are you going to make CPUs that aren’t a steaming pile of shit anytime soon? (wuwul)
That’s a little harsh! They’re not. Come and ask me face to face!

No senior suits were harmed in the course of this interview and Roy Taylor has never threatened to overrule anyone. Well, not much, anyway.


  1. Jason Moyer says:

    But I believe it’s intellectual rather than experiential. This is something that requires you to dig very deep to expose it, that requires you to run your game at 75 per cent below normal speeds to even see it.

    Or you could just play basically any DX11 game on a 7000 series card and notice that regular benchmarking software shows the framerate holding steady while there is obvious, horrid stuttering happening. In most games that have the option, I get higher framerates with my 7870 using DX10 or DX11 renderers but I nearly always end up playing with the DX9 renderer because of the obvious microstuttering.

    • The Dark One says:

      My driver complaint is how a DirectX update update ( link to support.microsoft.com ) broke font rendering in Firefox. It’s apparently tied to AMD drivers, but my card is off the monthly release cycle now. :/

      • weary_scientist says:

        Awesome! I noticed this too, but thought the Magic of the Internet was just removing letters to annoy me.

    • PoulWrist says:

      Sounds like you’re experiencing something different than “microstutter”.

    • Bremze says:

      link to pcper.com

      “From a single GPU perspective though, the Radeon HD 7870 is indeed faster than the GeForce GTX 660 in nearly all of our tests, whether that be single monitor or multi-monitor configurations”

      Either you’re running Crossfire (get Radeon Pro, set a frame limiter and Adaptive V-Sync), are not on the newest drivers, or there is something else in your setup that’s acting up.

      • GettCouped says:

        becareful with current frame timing benchmarks. Although, I like pcper, their benchmark is largely an invalid one.

        link to anandtech.com

        great article on why and the future of da bench!

        • pmh says:

          You seem to be complaining that PCPer are not using FCAT, but they very much are (they and TechReport have been on the forefront of this for a long while now). The methodology is linked to in the article he provided.

    • tetracycloide says:

      That answer is a complete dodge. What’s so odd about the “the problem is all in your head dodge” is that AMD gave an honest answer to the question in a recent article on anandtech. The answer? They didn’t really test for it before so things weren’t optimized to reduce it.

      • Premium User Badge

        particlese says:

        What’s also odd is that his other answers were pretty honest, within the legal limits. However, upon re-reading the answer, I think it does have some non-dodgy elements to it:

        -It is a problem that requires digging deep to find subtle problems. (I’ll say “apparently” since it’s really beyond my expertise, despite reading the Anandtech article.)
        -They obviously do take it seriously since they’ve made noticeable improvements.
        -The former employer that he loved for so long didn’t have as much of a problem I think because they already checked for it and worked on it. Maybe that’s what he was trying to convey, but I suspect he’s just had to put up with the question a billion times a day recently.

        • czerro says:

          That’s a weird way to interpret his comment. He says they got all the same sort of questions regarding drivers when he worked at NVIDIA. Also, FCAT doesn’t really show what people seem to think it shows, and it’s hard to say if it’s even related to microstutter. Combine the fact that people don’t understand what FCAT reports, the low tech savvy of the average user, and the average person being confused about what microstutter phenemona is verse traditional frame drops due to performance dips. So instead of addressing the fact that NVIDIA’s FCAT, while interesting, is just as partial a picture as traditional benchmarking tools, and we really need to see whats going on inside the render chain. He instead succinctly stated that the problem needs further investigation by all parties.

    • Premium User Badge

      particlese says:

      I do like this guy (based entirely on how he answered these questions), but I must say the stuttering is definitely experiential. I only found out it was a Thing after several months of intermittent googling.

      I’m running a single 5850, so I would hope the drivers are more mature and stable in my case (as opposed to the 7xxx’s), but the general performance is also lower there, obviously. Nevertheless, in places where Skyrim runs at 60fps, it’s really smooth with nice, regular, easily-noticeable stuttering. (Try turning around smoothly if it’s not obvious while running.) Where the average frame rate is lower, it’s much harder to notice, which makes sense given that the variation between frame intervals is then much smaller compared to the average interval (one might refer to the standard deviation of the frame delivery period).

      Unfortunately, the only Nvidia GPUs I have for comparison are a Geforce 3 Ti 200 (overclockable to the 500’s speeds, oh yeah!) and a GeForce GT 630M*, so I haven’t been inspired to test those. Benchmarkers indicate that Nvidia has the problem under reasonable control, though, and I think I read it’s because their driver people already put lots of effort into it.

      Anyway, as I said in the previous article, I noticed a big improvement since one of AMD’s beta drivers (I think it was pre-13.1, which I see is out now, but I’m on a different comp), so obviously some folks at AMD have been doing some good work on it.** Señor Taylor would do well to know this!

      *Extremely similar to the 540M and 440 if we go by the code name and almost all other specs, depending on which versions you’re talking about…how’s that for annoying-to-the-enthusiast, business-smart branding?
      **See also: A link posted in response to my question in the previous article’s comments.

      • Jason Moyer says:

        I’m not sure what the exact cause is, but I found a recommendation on Steam to use a frame-limiter with Skyrim to get rid of the camera jitteriness/irregular movement. For some reason if the game’s framerate is higher than your monitor’s refresh rate, it starts acting up. Using something like Dxtory to lock my framerate to 59 keeps everything smooth as silk. Supposedly this is a problem with other Bethesbryo games but I’ve never experienced it outside of Skyrim.

        • Snargelfargen says:

          Yeah Skyrim’s micro-stuttering is a unique issue caused by a quirk in the way Gamebryo’s internal clock works. Trouble-shooting it was hell, especially since the forums are full of folks with helpful “advice” about micro-stuttering caused by crossfired GPU’s, slow cpus and assorted other similiar but actually completely unrelated problems.
          DXTORY did the job for me as well, although I’m a couple patches behind, and supposedly the issue has been fixed for some.

          E: also only encountered the problem with Skyrim, although I remembr lots of complaints about the same issue in Fallout 3.

          • Premium User Badge

            particlese says:

            Ah, that sounds about right. Before the driver-based fix, I actually used ENB Series for its frame limiter to work around the problem, but it took me a while to find a preset that looked like vanilla Skyrim and could be easily configured to do almost nothing else if I wanted to play at a nice frame rate. My other posts have been in reference to non-ENB Skyrim, though, so I didn’t want to confuse an already long post.

            Thanks for the tip on Dxtory!

    • Continuity says:

      I have the same card and no problem with micro stutter, my understanding was that the stutter is something that affects crossfire. If I were you i’d look to some other problem in your system to try and resolve.

      • Premium User Badge

        particlese says:

        It’s certainly an issue for some single-card users (here‘s a press-based example, although they’re using a 7950), but yeah, there’s always the possibility of weird interactions with the rest of the system. I know for sure that the driver update solved the only problem bothering me, though, so my free time is safe…for now — bum Bum BUMMMMMM!!

  2. Gap Gen says:

    “Are you going to make CPUs that aren’t a steaming pile of shit anytime soon? (wuwul)
    That’s a little harsh! They’re not. Come and ask me face to face!”

    I for one would put money on the RPS Fight Series.

    • MacTheGeek says:

      I’m surprised we haven’t seen a campaign for it on Kickboxstarter.

  3. SanguineAngel says:

    Hey there’s some lovely responses here

    • meloncrab says:

      Yes, somebody really knows how to answer questions nicely and honestly.

      • MattM says:

        The way EA answers questions is so crazy. Anyone interested enough to read an interview with an EA representative is probably knowledgeable enough to see through their marketing lies. If an EA PR guy just said “We made SimCity always online because we wanted to combat piracy and to create a online treadmill that keeps players engaged and spending money on dlc” then I would be far less enraged then when they try and claim it’s for the user’s own good.

        • SanguineAngel says:

          I wouldn’t say i would be far less enraged overall but I would not be additionally enraged about their blatant lies :)

  4. Curry the Great says:

    “Why does AMD have such bizarre and confusing product branding? (Vandelay)
    We used to have great names. Then we sent people to Harvard Business school and they came back and messed them all up.”

    Would be nice if he didn’t just deflect this one with a joke and actually explained why they do this. But it probably comes down to confusing the consumer into accidentally spending more money or buying a lower quality product. Why doesn’t he want to really tackle this?

    • Sakkura says:

      Nvidia does it, the OEMs push for it, AMD eventually gave in.

    • TheApologist says:

      It probably does come down to that, yes. But at least the joke was funny :)

    • tobecooper says:

      He tackled it though, didn’t he? The confusing naming is business – the stuff the marketing people were taught in their uber schools – and they do indeed earn more money because of it.

      • Curry the Great says:

        Well, why doesn’t he say he wants to do something about it or that he’s powerless to stop it? He jokes about corporate suits being corporate suits, then he acts like one not stating his opinion on it or how it actually works. Stuff like that makes me pretty cynical about these companies. What’s so bad about actually saying: “it’s regrettable but we think we make more money like this”?

        • tobecooper says:

          I don’t believe that in such a large corporation honesty can overrule money. And openly admitting what you want him to admit wouldn’t be the best PR move.

          But I’m cynical.

          • Curry the Great says:

            I know that’s common in corporate-land, but if I really think about, why the hell wouldn’t you be honest about your desire to make money to your customers? Companies do everything to make money, not to increase consumer trust or whatever (as long as they think the money saved on getting th consumer trust is more than spending on it), I appreciate them way more if they’re honest about this. Bleh, I’m probably idealistic and I guess there’s a fear of breaking the status quo and being branded as an evil company that outright admits it.

          • tobecooper says:

            As you mentioned in your first comment, this sort of strange branding that AMD has, is potentially hurtful to customers who don’t research their purchases enough. So, I think candidness isn’t an option.

    • Vandelay says:

      Although chuffed that my question appears at the very top of the article (am I allowed to shout FIRST! just this once?) I am disappointed in the dodge. An amusing dodge it might have been, but a dodge nevertheless.

      I’m certain it is a marketing decision, where they confuse people into buying the wrong thing, but won’t the poor performance they get from their cheap card, despite it having bigger numbers on the front of the box, just put them off buying AMD again? I can’t see how it doesn’t cause more people to be careful what they spend their money on, in the long run.

      • Baines says:

        If you don’t buy AMD, then what are you going to buy? Nvidia also uses naming schemes designed to caused confusion.

        • Sakkura says:

          IMO Nvidia are worse and were the ones to pioneer this sort of thing. Not that it absolves AMD, but it does make it a bit hard to do anything about it. Sure, Intel has a logical naming scheme for their GPU. But even their top-end HD Graphics 4000 sucks.

    • Deadly Sinner says:

      The naming conventions make sense and takes 5 minutes to learn. How would you name them, considering AMD comes out with 8 or more cards a year? Switching the first two numbers around is the only possible improvement I can think of, only because the second number is a better indicator of power.

  5. Sakkura says:

    I’m sad that you didn’t ask about Linux driver support. It seems like a relevant issue in this Steampenguin era.

    • Nasarius says:

      If you asked now, you’d just get a sugar-coated version of “we don’t give a shit about Linux”.

      The Steam Box is totally unproven in terms of price and market. The first ones will have to use NVIDIA chips. If they’re successful, then maybe AMD will start taking Linux seriously. But not before then.

      • Sakkura says:

        It’s not just about the Steam Box, in fact that’s a tiny fraction of the issue. The fact is Steam is going Linux for ALL PCs.

        • uh20 says:


          right now, if your running linux, its nvidia, or nothing (heh)
          mac essentially has to put up with bad drivers no matter what, but hey, they (collectively) are either rich or gullible, so you wont hear about it from them.

    • Jeremy Laird says:

      I did ask, but felt the answer was repetitious regards the general driver quality answer. So I dropped it.

    • Low Life says:

      I love how (based on the very limited sample size of this comment thread) Nvidia seems to be the more consumer-friendly option for gaming on Linux, while Mr. Linux publicly gives the finger to Nvidia because they’re so difficult to work with. I guess Nvidia’s set their priorities on making customers rather than (open source) friends.

      • Premium User Badge

        particlese says:

        With their driver code dumps (new, shiny, and theoretically useful!), I’d say AMD is quite Linux-friendly, but in a very different way, I guess, based on popular opinion. I haven’t used anything but Intel graphics under Linux for years (it’s usually sufficient for what I do), and I haven’t been following reports on Nvidia’s or AMD’s driver quality, but I’m fairly certain that Nvidia is still just a blob factory, albeit a well-oiled one, given the comments that abound.

        • Sakkura says:

          My impression is that AMD does very little work on Linux support, but are relatively forthcoming with information and resources for the Linux community to work with. Nvidia seem to do a lot more work on their Linux drivers, but they keep everything locked down.

  6. Thermal Ions says:

    “Roy was not pre briefed”

    No, but he was pre prep’ed by his marketing department, and they likely expected almost all those questions. They also would have gone through the possible comedic answers to fend off difficult questions.

    Of course his position title could mean “Head of Marketing” in actuality, so he may have prep’ed himself.

    /jaded off

    (sorry don’t feel like his responses gave me anything more than I expected after reading each of the questions – except maybe the moose and lack of thumbs image, that caught me off guard)

    • Jeremy Laird says:

      No, he really wasn’t. The first he knew of the interview was precisely 10 seconds before it began when I said, “hey Roy, I’ve got some questions for you from RPS.”

  7. RProxyOnly says:

    You know what.. at least the guy has enough knowledge to acknowledge that that there is a difference between digital and analogue music.. AND knows his audience enough to make sure he differentiates between them.

    In case you don’t understand what he is saying…. as far as digital music goes.. it can’t get any better, it’s as ‘hi-fi’ as it’ll get, (you can to a certain degree increase the ‘hertz’.. but no matter what you do you still have the absolutely massive drop off at both ends, top and bottom, of the waveform) the job is done, which is opposed to analogue music, where the job isn’t done, because we aren’t given superior analogue any more… even vinyl records pressed today are digitally processed, so there is no point of putting it on a vinyl as opposed to sticking it on cd.

    I’m surprised anyone from any industry would admit that these days.

    • The Dark One says:

      Well, the need for Soundblasters certainly died when Vista cut the functionality that let EAX work, but discrete sound cards still offer cleaner output than integrated Realtek solutions.

      • RProxyOnly says:

        That’s actually more about implementation rather than differentiating quality… garbage in garbage out. Sound out of any computer is still digital regardless of it’s source… Eax and alchemy are all just digital sound processing.

      • Nucleus says:

        The difference between onboard and discreet audio is as massive as SD vs HDTV; of course if the other components in the chain are up to it. A shiny new soundblaster won’t matter if you’re streaming 128kbs mp3 and playing it through 50£ speakers.

        • RProxyOnly says:

          It’s NOT SD vs HD.. (edit) ‘original’ SD was analogue.. onboard and discreet are still both digital.

          • Low Life says:

            I think they’re referring to the quality of the analog outputs (and D/A conversion) done on integrated audio compared to an external device. Of course if you just take digital audio out, it doesn’t matter if you take it straight from the motherboard or from a $5000 device, it’s going to be the same stuff.

            Though techically it’s not really a case of onboard vs. external, but $5 vs. $100 (or “insert price of your external DAC here”). Of course a dedicated audio device that costs as much as a motherboard is going to have higher quality DACs (and ADCs) than the one found on a motherboard.

            Just like an APU is (for the foreseeable future) going to be worse than a CPU + separate GPU, because an APU can’t/won’t cost $700. But comparing a $200 CPU/GPU combo vs. $200 APU is not so clear, becaue you get some benefits for having things closer together.

          • RProxyOnly says:

            Low life.. I’m not sure that is the point that’s being made (in the interview… I think that’s what you mean?).. I’m pretty sure he’s just drawing the difference between analogue and digital music as ‘things’.. rather than conversion methods/equipment.

            That was my take anyway..

          • Durkonkell says:

            Standard Definition has been digitally sourced and – in many cases – digitally delivered for years. Sure, if you watch an analogue broadcast it’s analogue (strangely enough), but anything on a DVD or via a satellite or cable service is going to be fully digital all the way through. Digital Betacam (a professional digital tape format for SD) launched in 1993. The first terrestrial digital HD broadcasts occurred some time after that.

            Yeah, SD was analogue only for years, but for pretty much the whole history of digital HD broadcast, SD has been digital too. The comparison is valid, at least in that respect. Whether it’s actually valid in terms of discrete audio cards giving you four times more… clarity or something is another matter. I’d suspect that the difference is not as pronounced, but I’m not an audiophile.

          • SuicideKing says:

            I think he was simply saying that integrated eventually becomes as good as discreet.

        • Vandelay says:

          Certainly true that there are other factors apart from the sound card, but there is definitely an improvement over onboard and dedicated. I am using ~£50 5.1 speakers and the difference is certainly noticeable and I am far from an audiophile. Even music through Spotfiy (at least the high quality tracks,) play with far greater depth and the surround sound effect is much more detailed.

          I was skeptical at first, but I would recommend a sound card to anyone , particularly of they listen to a lot of music.

    • MattM says:

      The analogue record format may have a large theoretical fidelity, but practically speaking the machines that pressed the vinyl had limited precision and the needles that read records were only so sensitive. Similarly, no matter what input signal you put into a set of speakers there is a limit to the sound quality they can produce.

      • RProxyOnly says:

        I’m talking comparison of source quality, of course output can vary dramatically depending on equipment, so that being my point ‘Theoretical’ has nothing to do with it.. The massive differences in sound harmonics is absolutely demonstrable .

        • MattM says:

          Records also have a TON of inaccuracies. For my mom’s birthday, I converted all of her old albums to mp3s. Their may have been additional fidelity in those records that I didn’t notice, but they definitely had a ton of pops, plenty of hiss and many infinitely looping sections. If CD digital audio isn’t good enough, then I think we should move on to better digital formats rather than trying to preserve vinyl. There are just too many issues with with it.

          • RProxyOnly says:

            The pops and hisses are simply a by product of materials used, processing method and equipment failure.. but I’ll be honest with you, I’ll take pops and hisses AND the additional wave frequencies over sterile half music any day of the week… Digital music just sounds ‘dead’ to me.

          • Sparkasaurusmex says:

            The difference between digital and analog music isn’t really that big compared to the difference between live and recorded (digital or analog) music. Basically you get awesome sound live, but once it’s recorded it’s been recorded…not a big deal if it’s analog or digital, the big deal is the instruments and musicians and style and composition.

            Same’s with graphics: Who cares, just get good artists to paint good pictures and everything that comes after will be fine.

  8. MattM says:

    I find I don’t really care about single threaded performance anymore. It seems that just about every game these days that is cpu intensive at least duo threaded and most are quad threaded. If a game is single threaded, then I find that it is undemanding enough that any recent cpu is sufficient and more cpu demanding games are all multithreaded.
    Not trying to claim that there aren’t scenarios where single threaded performance is important, just that I don’t encounter them in my computer usage.

    • Nasarius says:

      Dwarf Fortress is the big one. In fairness, any world simulation should be built multi-threaded from the ground up, but DF wasn’t, and it remains single-threaded and limited by the performance of one CPU core.

      • MattM says:

        After thinking for a bit, I realized that there is one application I use where single thread performance is important, PCSX2. I guess I still care a little about single thread performance.

    • Asurmen says:

      By duo and quad threaded do you honestly mean 100% CPU usage with equal load between each thread? Because I can’t say I see that happening in any game I play. Yes, they’ll be multiple threads working, but a single thread will be doing 90% of the work while the rest piss around doing occasional bits of work, meaning single threaded performance is very important and the reason why AMD CPUs are inferior to Intel for gaming performance.. Games are crap for using multi threads, and they have been forever.

      • MattM says:

        It seems that most modern games perform much better on multicore cpus. Single cores have very low max fps. If I were recommending Cpus for a new buyer, I would say that a quad core will perform better in most newer games than a duo core with a slightly higher performance per core.

        • Asurmen says:

          They do yes, but with games the vast majority of work is still done on a single thread. AMDs single thread performance is so poor compared to Intel while being comparable or better on multi threads is the reason they’ve been the underdog for gaming for quite some time.

        • Snargelfargen says:

          There, uh, aren’t really any single-core cpus being made any more so that’s hardly a fair comparison. They stopped making Pentium 4 processors five years ago.

    • jalf says:

      Singlethreaded performance doesn’t mean “performance of a single-core CPU”, or even “performance of singlethreaded application”, really. It means “the performance of each individual core.

      So yes, while hopefully your game can take advantage of 2 or even 4 cores, it will still run better on 4 *fast* cores than 4 *slow* cores. That is singlethreaded performance. (And that is why it still matters, too). Each core can run a single thread, and you want each of those to be fast, no matter how many other cores you have.

  9. DodgyG33za says:

    “It’s a bit like the Poll Tax. I don’t recall anyone calling up and complaining about it!”

    Bad analogy. The poll tax lead to massive refusal to pay, a million person march, rioting on the streets of London and the fall of Thatch.

    Or is there another poll tax?

    • Chris D says:

      I was assuming there must be another one, perhaps in America or wherever Roy Taylor’s native country is. Otherwise that would seem to be the most ill-chosen simile in the history of the world.

    • RProxyOnly says:

      He should have had his toys taken away for that comment. /end.

    • MattM says:

      Yeah my monocle fell off when I read this.
      link to vgcats.com
      Poll taxes were a pretty shitty part of American history.

      • RProxyOnly says:

        Actually I think most of us here have the UK poll tax issues on our mind rather than the American version, (didn’t know there was one).

        • InternetBatman says:

          It was used to keep black people (and occasionally poor people) from voting. It’s not a great thing.

          • Premium User Badge

            Joshua says:

            It is also back with a vengeance with requiring an ID to vote (in the states, that is), specifically for voting.

          • Baines says:

            Joshua, that’s true and not true.

            Most of the states that require an ID to vote will also give you an ID for free for exactly that reason.

            The exception is North Carolina, which came up with a “solution” that makes the whole issue *worse*. North Carolina’s solution was a law where, if you are too poor to afford an ID, then you can swear an oath that you are too poor and the government will give you one for free. Except if the government decides that you weren’t too poor, you will be charged with felony perjury, which can bring up to a five year jail sentence. Bad on its own, but even worse there is no concrete definition of “too poor” in the law. So anyone who swears that they can’t afford to pay for an ID would be opening themselves to the risk of a five year jail term.

            North Carolina’s voter ID law is currently being challenges for those reasons.

            EDIT: I should note that voter ID laws find support due to how easy it is to perform voter fraud, and not because it pushes out poorer, minority voters. Though there is some of the latter. In North Carolina, there is an ongoing issue of illegal immigrants in the form of migrant workers, which has affected various laws as well as voter opinions on both sides of various issues.

  10. RProxyOnly says:

    “As for competing with Intel, if you look where Intel is going in terms of GPU investment and the selection of an APU by Sony [for the Playstation 4], all that reflects the fact that heterogeneous system architectures [google it, folks] are the future. I believe that so sincerely that I voted with my own wallet and my family’s future by joining AMD. The future is HSA.”

    So basically what he is saying (granted it’s VERY between the lines..).. why buy a console when you can have a full pc instead of a cut down one.

    • Guvornator says:

      I’m not sure he’s saying that, but if he were that would be a very interesting answer for someone whose company is powering 2 big next gen consoles…

    • Saldek says:

      After following Jeremy’s advice (“google it, folks”), I believe Taylor is simply saying that he believes the combination of a CPU and a GPGPU on a single die is where development is heading. The PS4 APU is mentioned as a strong indicator for this development. And further down the line he sees these APUs being based on a heterogeneous system architecture.

      HSA APUs can apparently challenge the discrete CPU/GPGPU model of our desktops by:

      – allowing both CPU and GPGPU to creat work items
      – allowing both to use the same memory (obviating memory copies/cache flushes)
      – reducing dispatch latency
      – reducing power consumption

      I don’t think he’s addressing the performance the PS4 APU might have compared to a desktop setup. It seems to be all about the viability of the architecture AMD is commited to.

      link to developer.amd.com

      link to en.wikipedia.org

  11. adonf says:

    Does anybody understand that poll tax analogy?

  12. uh20 says:

    i dont know what to say about this guy, he’s definitely softening up his answers for the company.

    it seems that AMD has a few problems:mainly undeveloped drivers (especially linux), then unstructured products, and they have close to no idea what is going on in the VR/Tablet world.

    from a company standpoint they are also not communicating well their upsides, and instead relying on distributing their hardware to things like upcoming consoles, if they cant get their boring products to sell directly to customers, then they are going to go down almost the exact same way other electronics distributers and manufacturers went out.

    • TechnicalBen says:

      VR seems unrelated. It’s down to the VR provider, not he graphics card. If the graphics card can do 3D, it can do VR. Wait, does AMD do 3D vision?