AMD reveals K12: New ARM and x86 cores are coming

Since Rory Read took the helm at AMD, he and his management team have crafted and pursued a new strategy for the company’s future. Somewhat unusually for a major CPU maker, the “new” AMD has been pretty reserved about communicating its plans. In fact, I think the only truly important revelations from the new AMD have come in the context of just two events, an Analyst Day in 2012 and a “Core Innovation Update” this morning.

Fortunately, after this morning, we know a whole heck of a lot more about AMD’s strategy going forward, about what Read’s “ambidextrous” approach to development means, and about specific products coming in the next two years. We now know that AMD intends to compete where it has done well historically—in desktops, servers, and laptops, as well as embedded systems and mobile devices—by fielding CPUs compatible with the two major instruction set architectures: ARM and x86.

The biggest news is about chips coming in 2016, but let’s step through these things in order, since they’re all part of a progression.

The first bit of news worth noting today was a live demo of AMD’s Seattle chip. This system-on-a-chip is intended for use in servers, and it incorporates eight ARM Cortex-A57 CPU cores. AMD demoed this SoC running Linux and the LAMP web services stack, and the firm said it intends for the 28-nm Seattle to ship in volume by the end of this year.

AMD’s presence lends instant credibility to the nascent ARM server market. The pull from big consumers of server-class hardware like Google and Facebook has already set the stage for ARM CPUs to succeed here, but the hardware options for 64-bit ARM processors have so far been limited.

AMD has also been shipping server versions of consumer APUs like Kabini, so it has low-power, server-class SoCs for both ARM and x86 in its quiver for 2014.

What happens in 2015 is even more interesting. AMD has updated SoCs in the works based on a 20-nm fabrication process, and those chips will adopt a common “design framework,” in the words of Dr. Lisa Su, that goes by the code-name “SkyBridge.” The idea with SkyBridge is for ARM and x86 versions of AMD’s products to share infrastructure, so they’ll be pin-compatible with the same motherboards and systems, regardless of which ISA they can execute.

Dr. Su said she expects the first example of SkyBridge products to target embedded systems and “some client markets,” presumably low-power devices like tablets or small laptops. These products will use either the “Puma+” CPU cores that AMD just shipped in its Mullins APU or a low-power-optimized version of the Cortex-A57 core licensed from ARM. One can imagine a tablet based on this first low-power SkyBridge platform that could ship with an ARM-compatible SoC for Android and another variant, equipped with the x86-compatible chip instead, running Windows. The hardware could be identical save for the SoC itself.

Both of these chips will incorporate key AMD technologies, including the GCN graphics architecture that powers all current Radeon GPUs and support for HSA, AMD’s model for shared execution of code across CPU and GPU cores in the same system. HSA provides a layer of abstraction above the execution hardware, so making use of it on both x86 and ARM ought to prove reasonably seamless. True HSA support in hardware requires a coherent path to memory for the GPU, like AMD built into Kaveri, so its inclusion in an ARM-based SoC will require some honest work.

According to Su, some of these SkyBridge processors and motherboards could have a socket-based infrastructure, probably much like what’s out there now for desktop Kabini. That development could give ARM an opportunity to make inroads on the desktop, particularly in emerging markets where the current desktop Kabini may already be getting some traction.

SkyBridge will start with these low-power SoCs, but Dr. Su made clear that this dual-track development framework will eventually encompass a whole family of products across different markets. We’ll likely see different pinouts and sockets for chips of different sizes, with an ARM and x86 version of the silicon in each case.

SkyBridge sets the stage for new products coming in 2016, and that’s where things truly become intriguing.

AMD revealed that it is working on not one, but two brand-new, built-from-scratch CPU architectures. It has licensed the ARMv8 ISA and, in Dr. Su’s words, “we are already well on our way to developing custom ARM cores.” At the same time, AMD is building a brand-new, x86-compatible CPU core that will serve to replace Bulldozer and its lineage.

I’m not quite sure whether the biggest news here is the announcement of a new high-end x86 architecture from AMD or the fact that ARMv8 is getting the same treatment.

Since we’re talking about a replacement for Bulldozer, this is an entirely different class of beast from the “Puma+” and Cortex-A57 cores in the first SkyBridge parts. AMD’s execs noted that these are high-frequency, high-performance CPU cores that will span the range from laptops to desktops to servers—and not just “microservers.” We don’t know exactly how big they’ll be compared to Bulldozer or, say, Haswell. The smartest play might be to aim for something a little smaller than those cores, but that’s the basic class of performance they’re undoubtedly meant to achieve. AMD is returning to its roots, aiming to produce a best-in-class big core.

Well, two of them, this time.

AMD used the code-name “K12” to refer to the ARM core. I’m not clear yet whether that name also applies to the x86 core.

These two CPU microarchitectures will be, in the words of CTO Mark Papermaster, “sister cores.” (Papermaster came to AMD from Apple, and he managed to lure CPU architect Jim Keller from Apple, too, shortly after Keller led the development of Apple’s own 64-bit ARM core.) Keller explained during this morning’s Q&A session that the new cores will share more than just pin compatibility. He said they will be “compatible at the pin level and inside.” That likely means that the ARM and x86 SoCs based on these new cores will share the same internal plumbing—things like the I/O ring around the edges of the chip and the last-level cache. AMD’s design teams will then be able to fit, say, four ARM cores or four x86 cores into the space on the interior section of the chip.

AMD’s Mark Papermaster (left) and Jim Keller (right)

Presumably, these sister x86 and ARM cores will perform about the same, but they evidently are not just two variants of the same microarchitecture adapted to different ISAs. Keller was very complimentary about the ARMv8 ISA in his talk, saying it has more registers and “a proper three-operand instruction set.” He noted that ARMv8 doesn’t require the same instruction decoding hardware as an x86 processor, leaving more room to concentrate on performance. Keller even outright said that “the way we built ARM is a little different from x86” because it “has a bigger engine.” I take that to mean AMD’s ARM-compatible microarchitecture is somewhat wider than its sister, x86-compatible core. We’ll have to see how that difference translates into performance in the long run.

Deep into the Q&A, some analysts were asking how AMD would differentiate itself from other contenders in the burgeoning ARM server space. Given what Keller and Papermaster shared, the answer seems quite obvious. If AMD were only offering A57-based parts like Seattle, it might have trouble standing out, but we’re talking now about a full-fledged, next-generation processor worthy of the Opteron name. The other players in the ARM server space have, at least so far, concentrated almost exclusively on small, lower-power cores. Few have even made the transition to 64-bit addressing, and nothing in ARM’s own portfolio of licensed cores is anywhere near this potent. The ARM-compatible K12 may be one of a kind if it arrives on schedule in 2016.

If it enables the folks building data centers to get the same sort of per-thread performance from ARM-based servers that they can from x86 processors, the K12 could obviate the need for x86 CPUs almost entirely. It could help key a dramatic transition from a single dominant ISA to two competing options—or it might even spark a longer transition away from the Intel-dominated x86 world to one ruled by the more open and expansive ARM ecosystem.

With this “ambidextrous” plan, AMD can win regardless of how the ISA battle plays out. That concept sounded promising when Rory Read first floated it a couple of years ago, but now that we can see more details, it seems even better than most of us probably expected.

As always, everything depends on the specifics. AMD has to deliver these products, and they have to perform well in order to matter. This company can’t afford another Bulldozer-style fiasco. But in some senses, AMD now looks to be even better positioned than Intel for the next few years. We haven’t been able to say that for a long, long time.

Comments closed
    • ronch
    • 7 years ago

    According to this [url=http://www.kitguru.net/components/cpu/anton-shilov/amd-to-roll-out-all-new-high-performance-x86-micro-architecture-in-2015/<]Kitguru article[/url<]: [quote<]The next-generation high-performance x86 micro-architecture from AMD will scrap the CMT approach in favour of more traditional SMT[/quote<] Better late than too late.

      • maxxcool
      • 7 years ago

      🙂 no more $%^&ing modules and disparate fpu’s ? Yup that’s a win. To bad it will be jaguar+…

        • sschaem
        • 7 years ago

        Would you take 4 core over 4 module ?

        So the other way to look at it.. now FPU have half the amount of branching logic.

    • ronch
    • 7 years ago

    Don’t count AMD out just yet, guys. I have a feeling we will see a resurgent AMD in the next few years.

    • blastdoor
    • 7 years ago

    “Few have even made the transition to 64-bit addressing, and nothing in ARM’s own portfolio of licensed cores is anywhere near this potent. The ARM-compatible K12 may be one of a kind if it arrives on schedule in 2016.”

    Denver?
    [url<]https://techreport.com/news/20201/nvidia-unveils-project-denver-an-arm-compatible-cpu-for-desktops-and-servers[/url<]

      • Damage
      • 7 years ago

      …isn’t shipping yet! Also, contrary to Jen-Hsun’s initial statements about Project Denver way back when, it doesn’t appear to be going into servers any time soon.

        • blastdoor
        • 7 years ago

        And K12 is shipping?

        I guess it depends on how you define “server”. maybe if you exclude HPC then Denver won’t count, but I bet Denver ends up in HPC.

          • Damage
          • 7 years ago

          I wouldn’t be surprised if Denver ended up in HPC, particularly onboard a future GPU. But we don’t yet know a lot about Denver’s performance and power draw, nor do we have a schedule for it making any sort of appearance in a more traditional sever-class SoC. Seems to me that Denver will probably be quite some distance from competing with Haswell and its successors.

          Anyhow, I simply made a true statement (few have made the 64-bit transition) and followed it with a possibility (K12 may be one of a kind if it arrives in 2016). I’m happy to be proven wrong, but nothing definite says Denver will be the core onboard a product that makes that happen.

            • the
            • 7 years ago

            K12 may not be alone. Both Broadcom and Qualcomm are planning server oriented SoC’s with Samsung being a possible wildcard jumping into this market. In addition, these companies have ARM ISA licenses that they’re using to develop new ARM core designs.

            The one thing that is expected of K12 that isn’t implicated from other designs is that it is going to be both a high clock and wide design. So far high end ARM core designs have had the dynamic power range to slip into mobile devices as the primary CPU (Cortex A15 on 28 nm and A57 are examples). AMD hasn’t said how low power K12 will go.

            • ronch
            • 7 years ago

            Of the many ARM licensees planning to put out custom cores, only AMD has actual, extensive experience building real server chips.

            • the
            • 7 years ago

            Broadcom has a history of server products. Especially if you consider there acquisition of Serverworks years ago.

      • sschaem
      • 7 years ago

      But I dont think its showing on any nvidia recent roadmap. Some speculated , because new roadmap (that use to show denver) have now removed the mention of it, that this project has been put on the backburner. No roadmap, no date, no mention of it.

      In contrast, at the moment AMD seem to have a roadmap for the K12.

      Personally, I think Denver is alive. And if nvidia wanted,they could beat AMD to market by 18month or more.

      But then again, Apple already got a custom 64bit ARM SoC , in products, for a long time now.

    • Shouefref
    • 7 years ago

    I’ve just discovered something odd. There is a small country in Europe, called Belgium (there are a lot of countries like that in Europe, with 10 million inhabitants). In that country there seems to be no official agent for AMD anymore. That means that all Belgians HAVE TO buy Intel chips! No wonder AMD has bad figures.

    I wonder how the situation is in those other little countries, like Portugal, Greece, the Netherlands… If they lost their agents there too, they might have lost half a continent.
    And how is the situation in the smaller American states?

      • Saribro
      • 7 years ago

      You can buy AMD chips in Belgium just fine. What the hell are you talking about?

        • Shouefref
        • 7 years ago

        Maybe you can buy them via the internet, but it’s very difficult to get one if you enter a brick&mortar shop and asks for them, whether boxed or in a pc.

        I also don’t understand why I get downvoted for stating a fact.

    • itachi
    • 7 years ago

    I can only hope AMD will make a nice come back in the CPU scene, but in 2 years.. wow, wonder how many new Intel will be out by then, then the FX 83xx will be old news ! hha, well with that said I’m still very happy with my FX 8320 that I just bought, beside the power efficiency, it’s performance are pretty badass for the price.

      • sschaem
      • 7 years ago

      Have you tried undervolting the chip?

      I was able to lower the voltage by about 15% and still raise the clock speed on my 8320 to 4.2ghz.
      (Not many reviewer did this, even so the chip is fully unlocked…)

      Its a 125w chip, and undervloting will drop that to ~95w.
      Frankly 95w, under load (and almost nothing at idle), for a 4ghz+ 8 core CPU / 8MB cache is nothing to cry about, specially when its often under $150.

        • itachi
        • 7 years ago

        I’m at 1.4v for 4.3 right now from what I’ve seen on forums it seem my CPU is a power vampire, but that’s not really the problem, though, my problem is more the heat all this power generates, had to decrease my OC from 4.5 to 4.3 because my socket temps were going far beyond 70° c I have an Asus m5a99fx pro rev 2.0 , I’ll have to buy a 12cm fan to try to fix it by putting it on the backside of the motherboard.

          • sschaem
          • 7 years ago

          Those chips can take a beating (as in lots of voltage) but the reward in pushing it to the limit is not worth it in my view
          (I think the FX line got a bad reputation in heat/power because they can be pushed so hard)

          What clock speed did the chip settle at (max stable clock) with 1.2v?

    • CBHvi7t
    • 7 years ago

    [quote<]If it enables the folks building data centers to get the same sort of per-thread performance from ARM-based servers that they can from x86 processors, the K12 could obviate the need for x86 CPUs almost entirely. It could help key a dramatic transition from a single dominant ISA to two competing options—or it might even spark a longer transition away from the Intel-dominated x86 world to one ruled by the more open and expansive ARM ecosystem.[/quote<] That would be inverse history. We had many architectures and instruction sets. Why should Server and PC coders suddenly start writing software for another ISA just because it is dominant on mobile phones? Maybe it is time to reverse history but I am not buying AMD stock.

    • Kretschmer
    • 7 years ago

    Chance of AMD executing anything interesting to the desktop/laptop enthusiast in x86: 0%.

    • Mat3
    • 7 years ago

    AMD will never have a chance at building a high end CPU until they embrace simultaneous multi-threading. It’s the only way to have both strong single and multi threaded performance. But this is AMD, they don’t believe in SMT, so performance-wise, we’ll just more of the same with a different name.

      • ronch
      • 7 years ago

      The wider the core is, the more it makes sense to implement SMT. I suspect their next x86 core will FINALLY have it.

    • balanarahul
    • 7 years ago

    Excited, but skeptical.

    • deruberhanyok
    • 7 years ago

    This gives me the warm fuzzies. Kabini and Kaveri look great (Kaveri especially so, if the lower-power variants ever make it to market), and the idea of AMD executing this roadmap and kicking out competitive products makes me smile. It worked really well for “us” (enthusiasts) last time… and it can again.

    Fingers crossed for you, AMD.

    • ronch
    • 7 years ago

    AMD must be one of the most amazing tech companies ever. Sure, their products haven’t been the best out there in recent years but tell me, which other company is doing x86 processors for servers, supercomputers, desktops, notebooks, tablets, and embedded apps, high end workstation graphics, graphics for serious gaming, supplies the CPU and/or GPU for all three current-gen consoles, does semi-custom chips, and is plotting to enter the ARM microserver market (and later also plans to sell big ARM cores)… [u<]all on a meager budget and the uncertainty of having GF and TSMC as their manufacturing partners[/u<]? Not Intel, not Nvidia, not Apple, not Samsung, not IBM. Nobody. They may be spreading themselves a little too thin but heck, it's all still absolutely amazing. And the kicker? We all thought they had to let go of big x86 cores to be able to concentrate on all these other things but boy, this news about a new x86 core in the works is the news of the century! I hope it'll be really good and will encourage AMD to keep competing in the x86 space for as long as it's relevant.

    • Chrispy_
    • 7 years ago

    Given the schizophrenia displayed by Win8/WinRT – having a dual ISA option would have solved the entire mess nicely, and given AMD a nice advantage over Intel.

    Sadly, AMD always seems to take too long to get these ideal solutions to market. It often seems like Intel can pull a superior solution out of nowhere when AMD have been working on it for years.

    I guess that’s the difference between two companies where the R&D budget of one is larger than the gross turnover of another (I’m sure that’s not actually true, but that’s what it feels like to me sometimes).

    • Rza79
    • 7 years ago

    [quote<]AMD's presence lends instant credibility to the nascent ARM server market. The pull from big consumers of server-class hardware like Google and Facebook has already set the stage for ARM CPUs to succeed here, but the hardware options for 64-bit ARM processors have so far been limited.[/quote<] I think you can rule out big players like Google and Amazon already. Google's going with Power8: [url<]http://www.wired.com/2014/04/openpower/[/url<] Amazon wants to develop it's own cores: [url<]http://gigaom.com/2014/04/28/amazon-joins-other-web-giants-trying-to-design-its-own-chips/[/url<]

    • ronch
    • 7 years ago

    By the way guys, AMD DID state that they won’t be competing in the high end x86 space anymore, right? So which of the three possibilities is right?

    1. AMD was just bluffing to make Intel feel secure and languish.

    2. AMD decided that selling cheap-ass Kabini chips is worse than selling lemonade and decided they really need to go back to the high end.

    3. The upcoming new x86 core really isn’t gonna blow Intel’s socks off. But then, why bother when they could [u<]probably[/u<] tweak Jaguar/Puma to reach new heights (as some folks over at Extremetech are suggesting)? Maybe we should start a poll. I vote #1.

      • kalelovil
      • 7 years ago

      You can have a high-IPC chip without trying to sell them to the traditional PC enthusiast market.
      For instance, Apple’s A7 CPU cores and Nvidia’s upcoming Denver architecture.

        • auxy
        • 7 years ago

        Neither of which are high-IPC compared to “big” desktop cores.

          • the
          • 7 years ago

          Apple’s Cyclone core is very wide whose IPC is on the level of Intel’s Nehalem/Westmere architecture. It is a similar 6 wide architecture. In fact, Apple’s Cyclone core bares a resemblance to the old DEC Alpha EV6 architecture that ruled the server world 15 years ago. It also helps that some of the old Alpha engineers were employed at Apple to help design the Cyclone core.

          • kalelovil
          • 7 years ago

          Only if you apply your ‘big desktop core’ term so restrictively than it only includes Haswell.

      • Flatland_Spider
      • 7 years ago

      4. The realized that’s the road to obsolescence, and they have to flex their muscle if they want to still be a viable company.

        • ronch
        • 7 years ago

        That’s analogous to #2.

      • Ninjitsu
      • 7 years ago

      Well, you may be right about 3, because Keller’s basically saying that he expects the ARM core to have better performance than the x86 core.

    • ronch
    • 7 years ago

    New x86 core (as well as new ARM core). It’s been a long time since I felt great excitement about a new x86 core from AMD. I almost never felt excitement about Intel cores and have always felt excitement about what Intel’s competitors are doing and planning to release (that includes Cyrix, but…). This is how I felt when AMD was nearing the release of Bulldozer. Let’s hope that AMD [u<][b<]REALLY[/u<][/b<] delivers this time.

    • ronch
    • 7 years ago

    A few things:

    1. If AMD will simply offer pin-compatibility between their x86 and ARM chips, what would spur companies deeply entangled with x86 to hop over to ARM? They can’t just ditch their x86 legacy. This will only make them retain their motherboards and other existing hardware but they’d have to look for new ARM apps or port their stuff over to ARM. That would make transitioning (which would give meaning to the ‘bridge’ in SkyBridge’) still difficult.

    2. I guess these projects have been chugging along for a few years now. So that’s where all of AMD’s money is going! I kinda suspected it.

    3. 2016 is just in time for my next CPU upgrade. See, I got an FX-8350 PC back in Dec. 2012 and I’ve been planning to hold on to it for 5 years because I totally love the fundamental Bulldozer architecture, its idiosyncrasies notwithstanding. By Dec. 2017 or thereabouts, assuming these new cores will not be delayed (crosses fingers), the platform for these new chips will have been available and hopefully mature enough

    4. If AMD can deliver these new cores and blow all our heads off, then I JUST HOPE GF and TSMC will have fab processes that won’t sabotage these cores!

    5. This news totally made my day! I love it! GO AMD!!!

      • sschaem
      • 7 years ago

      What make you think those core will be used in a ‘workstation’ class CPU?
      Most likely will be low power soc…

      I dont see any chance AMD will make anything competitive to skylake-e, and most likely will just approach desktop sandybridge, at best.

      I got an fx 8320 last year, best PC I ever built at all levels. im also set for a few years,
      But even with this news I have no beleif that my next workstation will be AMD powered.

        • ronch
        • 7 years ago

        [quote<]What make you think those core will be used in a 'workstation' class CPU?[/quote<] Because they're saying that it will replace Bulldozer.

          • Ninjitsu
          • 7 years ago

          Well, considering both Piledriver and Steamroller replaced Bulldozer, and Steamroller is only in APUs so far, it’s not exactly a given. Though of course, Steamroller is still fundamentally Bulldozer.

      • sweatshopking
      • 7 years ago

      Y U LOVE BULLDOZER?! IT’S SLOW AND HOT!

        • ronch
        • 7 years ago

        Oh but that’s the way I like it! Slow and hot!

          • sweatshopking
          • 7 years ago

          I’m more of a fast and tepid guy.

            • NeelyCam
            • 7 years ago

            You should join the Minute Men

      • smilingcrow
      • 7 years ago

      “I totally love the fundamental Bulldozer architecture”

      Really!

        • ronch
        • 7 years ago

        It’s called unconditional love.

          • smilingcrow
          • 7 years ago

          That’s such a strange phrase as love is unconditional anyway so it’s a completely unnecessary adjective.

            • ronch
            • 7 years ago

            Love isn’t automatically unconditional. What about selfish love? We’ve got that one too, right?

            • smilingcrow
            • 7 years ago

            That’s not love that’s something else entirely.

            • ronch
            • 7 years ago

            Really, there are more than just one kind of love. Take BROTHERLY LOVE, for example. Right, SSK?

            • smilingcrow
            • 7 years ago

            There is an infinite number of ways in which love expresses itself and they are all unconditional.
            The object in question is irrelevant be it a male friend, lover, family member etc.
            Don’t confuse the form with the essence. Love has no concern for form as it is truly unconditional.

            The term Tough Love is commonly used and in many cases it has nothing or little to do with love. Love is formless so it can be ‘tough’ or anything else that is required in the moment. But true tough love is very different from a tough love that is conditional.

            • sschaem
            • 7 years ago

            ? love is conditional, thats what define love, conditions. Without conditions love has no meaning.

            Unconditional love is a form on mental illness, thats why people have to put that adjective in front not to confuse it with real love.

            • smilingcrow
            • 7 years ago

            I don’t want to get into a semantic discussion on what love is but just to be clear I’m not talking about anything to do with romance or relationships.

            • Ninjitsu
            • 7 years ago

            bool love=true;

            IT’S UNCONDITIONAL.

    • ronch
    • 7 years ago

    This bit of news literally brightened my day. I am so happy AMD is planning a big comeback. I just sincerely hope they can really pull it off this time. Nonetheless, I am a proud AMD fan!

    I also kinda predicted not long ago in one of my posts here on TR that it would make sense for AMD to build a CPU that contains both x86 and ARM compatibility. Guess nobody really believed me back then, and it’s not like AMD will consider building such a device although I don’t think it’s entirely out of the question given how AMD seems to want everybody to hop on to ARM. Interesting days ahead, for sure.

    • UnfriendlyFire
    • 7 years ago

    “high-frequency, high-performance”

    GF: “Hey AMD. We finally got the FinFET working. Too bad it only scales up to 2 GHz before its power consumption shoots through the roof. You okay with that bro? Kthxbai.”

    • Bensam123
    • 7 years ago

    So… I get my ‘ahem’, I told you so out of this.

    [url<]https://techreport.com/news/23817/amd-arm-core-will-be-licensed-not-amd-own-creation?post=682346[/url<] [url<]https://techreport.com/news/23811/amd-to-build-arm-compatible-64-bit-processors?post=681958[/url<] Oh god, the I told you so is so good, *starts fapping fervorously*. (Keep in mind the above was all happening when everyone was like 'OMG AMDS GOING UNDER, tHE SKY IS FALLING!')

      • NeelyCam
      • 7 years ago

      In three years everyone will be saying SKYBRIDGE IS FALLING

        • ronch
        • 7 years ago

        Why the hate, Neely?

          • NeelyCam
          • 7 years ago

          Not hate; just boredom. I’ve realized that I don’t have much to be excited about in the CPU world.. my two desktops are perfectly fine for a few years to come. This ambidextrous stuff doesn’t matter to me since I’m not going to be buying a server chip anytime soon – micro or otherwise. (And personally I don’t know why AMD is going the ARM route. Why not leverage the x86 ecosystem since they have the x86 license? Why try to compete with a whole bunch of other companies who are better funded?)

          I’d like to buy an ultrabook later this year, but jdaven says Broadwell won’t be out by then, and none of these other ones seem to offer the kind of performance and efficiency I “need’ to justify the upgrade.

          I’m eligible for a phone upgrade in November, but 22nm Intel phones are nowhere to be found, and everything else is still 28nm Snapdragons, just like what I have now.

      • derFunkenstein
      • 7 years ago

      Say anything vaguely enough and you can be “right”. I clicked those links expecting you to make the bold proclamation that their second ARM design would be in-house or whatever, but no.

        • Bensam123
        • 7 years ago

        Mmm… making a ‘normal’ arm chip first, to test the waters then making their own spin on it. Hybrid chips… Improvements on efficiency after they work with ARM.

          • chuckula
          • 7 years ago

          So your insane powers of prognostication indicate that AMD is copying what Nvidia did with Tegra only doing it about 5 years later?

      • maxxcool
      • 7 years ago

      Ah, so you are saying AMD is dead… I agree.

        • ronch
        • 7 years ago

        It ain’t over until the fat lady sings.

          • maxxcool
          • 7 years ago

          MMM >:) fat ladies 🙂

      • sweatshopking
      • 7 years ago

      YOU don’t need to share that?! WTF?!! WHY WOULD YOU SAY THAT ON HERE?!?!!

        • ronch
        • 7 years ago

        Just wait till the Captain sees your post…

    • anotherengineer
    • 7 years ago

    So I wonder if the bulldozer replacement will be using Samsung/GF 14nm?

      • ronch
      • 7 years ago

      Prolly.

    • NovusBogus
    • 7 years ago

    A new, potentially non-sucky AMD architecture? HALLELUJAH, MY BROTHERS, THE PRAYERS OF THE FAITHFUL HAVE BEEN ANSWERED!!!1!1!11eleven

    AMD needs to focus on making a quad-core CPU that’s “good enough”, low power and very very cheap. Let Intel tire itself out chasing the enthusiast market with diminishing-returns R&D.

      • ronch
      • 7 years ago

      A CPU that’s good enough, has 4 cores, sips power and is very cheap? We have that today and it’s called Kabini.

      • BaronMatrix
      • 7 years ago

      Why do fanbois act like there are 30 other CPU makers out there and AMD is at the bottom.. There are two and one of them is a Cheating monopolist…

        • NeelyCam
        • 7 years ago

        Welcome back.

        Curious how BaronMatrix and dpaus both came back for the same article. Coincidence?

          • chuckula
          • 7 years ago

          Coincidence?

          I THINK SO!!!

    • Mat3
    • 7 years ago

    What a colossal waste of resources Bulldozer was. Even in its third iteration it’s pretty mediocre and now it’s being flushed. They could have just continued using the Phenom 2 design with the small incremental improvements they typically made to it every couple years and it would have turned out at least as good at a tiny fraction of the cost.

    But it could have been even better; Phenom 2 cores were pretty small compared to Intel’s stuff so there was more room to increase performance. Unfortunately, it was really stagnate for the last 4 or 5 years of its life. They had hardly done anything with the core itself since what, 2006?

      • NovusBogus
      • 7 years ago

      AMD severely misjudged the direction the PC world would take; Bulldozer was designed for a massively parallelized future that never happened.

        • UnfriendlyFire
        • 7 years ago

        That, and GF/TMSC started shifting towards low-power consumption silicon dies.

        And AMD also underestimated the tablet market in its initial stages, until it was a bit too late.

      • ronch
      • 7 years ago

      Well, AMD’s been flogging K10 for so long that it would have required major modifications to bring the K10 architecture more up-to-date. What I can’t believe is how AMD claimed that the third AGU is actually just sitting there doing nothing when Bulldozer came out. WHAT?? You’ve been spreading word that K7/K8/K10 had THREE ALUs and THREE AGUs for the LONGEST TIME! And now you say that the third AGU is just there to make laying out the chip easier? Crazy. Next thing AMD says Bulldozer is actually an in-order execution core!

    • Binglewood
    • 7 years ago

    Could AMD come up with something more interesting as a code name than “SkyBridge”? It seems they are trying to confuse people into thinking this is an Intel product. Intel has SandyBridge, IvyBridge, and SkyLake. I guess their marketing department had fun with this one.

      • anotherengineer
      • 7 years ago

      Sky is the word this year, just as this was the other year.

      [url<]http://www.youtube.com/watch?v=2WNrx2jq184[/url<]

      • Peldor
      • 7 years ago

      AMD has been selling bridges for a long time now.

        • MadManOriginal
        • 7 years ago

        Bridges to nowhere!

      • NeelyCam
      • 7 years ago

      Sort of like Thunderbolt -> Lightningbolt

      It’s much cheaper to hire a highschool intern for a marketing strategy position than one of those fancy Ivy League grads

    • jjj
    • 7 years ago

    You seem too optimistic. AMD does have few resources,not much cash and adding a 3rd core is not that easy (not saying they shouldn’t have , ofc a custom ARM core was a given).
    My biggest concern is that they can’t go for phones all that soon. They would need a lot more assets , connectivity, location and so on and they can’t afford to buy much.They could license some parts or license their IP to others ofc but it would be nice to see them have some plans for that market.With flexible and stretchable screens likely to arrive soon enough, tablets and laptops will fade away so not targeting phones and glasses seems a very hard choice. In phones they are behind a bunch of people that are likely to have custom cores soon and many are bigger than them.. Sure in some ways not being in phones can be a blessing too, they can focus less on low power and more on high perf and solid perf/watt but with no server or desktop ARM ecosystem it’s riskier..

    Other than that ,it’s nice to see so many custom cores ,should be lots of fun, right now it’s just boring with A15 on 28nm not being all that great and Snapdragon as the only other option in the high end.
    i do wonder if Mediatek is making it’s own. By revenue they are bigger than AMD, Nvidia and many others and they ship very high volume so it would be reasonable enough to go there.

    • TheCollective
    • 7 years ago

    It will be interesting to see how people take advantage of the ARM processor in a traditional datacenter. Right now, virtualization drives decisions like these for all but the biggest companies. If you get Microsoft to run Windows on ARM (not just tablets), you’ll get my attention. As it stands, the x86 platform is still by far the best platform for virutalization.

      • cmrcmk
      • 7 years ago

      x86 is generally the only platform for virtualization (IBM’s power virtualization is practically non-deployed from what I see). Until VMware or HyperV run on ARM, x86 will definitely control the datacenter in most companies for the foreseeable future.

        • willmore
        • 7 years ago

        You two seem to be confusing datacenters with small departmental server setups. Real, large scale, datacenters don’t tend to use a lot of Windows. The real work being done is in the LAMP stack or other software. None of that needs Windows.

        Once you’re in the OSS world, moving to a new arch really isn’t all that big of a deal. Sure, there’s plenty of poorly written junk that doesn’t port well, but the last few decades have helped a great deal. The constant tug between OSX, Linux, the BSDs, and the other commercial Unix systems has shaken out a lot of the bad practices. At least in the important pieces of software this is true.

        • stdRaichu
        • 7 years ago

        We use both pSeries and zSeries at work, and I don’t think I’ve ever seen IBM iron where virtualisation wasn’t used to at least some degree. The [url=http://en.wikipedia.org/wiki/Logical_partition_%28virtual_computing_platform%29<]LPAR[/url<] system is stupidly flexible.

    • dpaus
    • 7 years ago

    First new AMD architecture in a Long Time – and it looks veeeeerrrryyyy interesting.

      • auxy
      • 7 years ago

      No joke! I’m pretty hype! As a big AMD fangirl, this is super exciting! I can’t wait for more news. AMD’s disappointed me over and over, but somehow I feel like this might be worth getting excited over! (Or possibly I’m just a relentless optimist. Nah, couldn’t be that…) (≧◇≦)ノシ

      • the
      • 7 years ago

      Actually AMD’s Bobcat and Bulldozer line were the last new architectures in the x86 space since the Pentium 4. In light of that, it is genuinely interesting that AMD will have been able to pull three from scratch x86 architectures in a decades time.

      These from the ground up designs really don’t happen that often. Typically we just get one architecture that has slowly evolved over time. Haswell can trace its roots all the way back to the Pentium Pro. The last from scratch design Intel has built is Silvermont for usage in newer Atom chips.

        • Damage
        • 7 years ago

        New x86 architectures? Since Pentium 4?

        “New” is subjective, but much changed in:

        P4 Prescott/Netburst II (30 pipeline stages; this is new vs. Northwood/Willamette)

        Banias? (P6 revised for power)
        Conroe? (Banias massively revised for IPC)
        Nehalem? (Conroe revised, integrated memory controller and SMT)

        Sandy Bridge (Clean sheet)
        Haswell (From 6 to 8 wide, 256-bit vectors and datapaths for them)

        K8 (64-bit, integrated memory controller, multi-core)
        Other AMD stuff (eh)

        Bulldozer (Clean sheet)

        Atom Silverthorne (Clean sheet)

        Atom Silvermont (Clean sheet)

        Via Isaiah (Clean sheet)

        Bobcat (Clean sheet)
        Jaguar (15% IPC gain, two added pipeline stages)

        Seems like a lot of newness, including six clean-sheet designs between P4 and this new AMD core.

        Edit: David Kanter says I missed the Knights Corner core, which is also new, not really P54C. 🙂

          • Alexko
          • 7 years ago

          I don’t know why you have Sandy Bridge as “clean sheet”. Very much of it is inherited from Nehalem; at least as much as Conroe inherits from Banias.

            • Damage
            • 7 years ago

            That seems to be a widely held belief, but it is not how Intel architects explained it. Here’s my article on the subject:

            [url<]https://techreport.com/review/19670/a-quick-primer-on-sandy-bridge[/url<] The key bits: [quote<]That mystery, it turns out, is pretty juicy, because Sandy Bridge is part of the unprecedented wave of brand-new x86 microprocessor architectures hitting the market. Just weeks after AMD disclosed the outlines of its Bulldozer and Bobcat cores, Intel has offered us an answer in the form of its own substantially new microarchitecture. Now, making a claim like I just did is fraught with peril, since new chip designs almost inevitably build on older ones, especially when you're talking about Intel CPUs. That's the thing about Sandy Bridge, though: one of its architects proclaimed at IDF that it was essentially a from-the-ground-up rebuild of the out-of-order and floating-point execution engines. Such changes were necessary to accommodate the doubled vector width of the AVX instruction set, and it means something fairly momentous. As my friend David Kanter observed, this is, at long last, the breaking point where one can finally say virtually nothing remains of the P6 (Pentium Pro) roots that have undergirded everything from the Conroe/Merom Core 2 to the Nehalem/Westmere Core i-series processors. Not only has the execution engine changed, but nearly everything around it has been replaced with new logic, as well, from the front-end and branch predictor to the memory execution unit. Outside of Sandy Bridge's CPU cores, the "glue" logic on the chip is all new, too. The inter-core connections, memory controller, and power management microcontroller have been tailored to accommodate the presence of a graphics processor. Even the integrated graphics engine bears little resemblance to what has come before. If you're looking for a golden age of CPU design, we're living in it, folks.[/quote<] Not sure what more you want to classify something as new.

            • chuckula
            • 7 years ago

            Damage is quite right about SB being a major departure from previous designs.

            Let’s forget about many of the new microarchitectural changes in the cores themselves (of which there are plenty). Instead, let’s just look at the ring bus and data paths not only between CPU cores but with the GPU… even if the cores were practically copies of Nehalem, the cache architecture and data path changes alone are massive vs. what was present in Nehalem.

            • the
            • 7 years ago

            My interpretation of that is that nothing remains of the Pentium Pro which is rather uncontroversial considering the numerous changes over time. But I still see an evolution as parts of Nehalem/Westermere were still inherited which still had some P6 roots.

            The on-die bus was also an evolution form Nehalem/Westmere. The ring bus topology was found in Lynnfield for consumer systems (though all the socket 1156 chips with integrated graphics don’t have that internal ring bus). The HD 3000 and later GPU’s just occupy a slot on the ring bus just like a processor would. For what its worth, I do see Intel dropping this ring bus in favor of a cross bar in the Skylake generation.

            • ronch
            • 7 years ago

            But that’s like saying the K8 is a clean-sheet design, isn’t it? When K8 was released, AMD claimed that they practically took out all the parts, redesigned them, and placed them back together. And then it also had 64-bit goodness and an IMC. We can also look at the K8 as a clean sheet design if we go by how Intel claimed that SB is a clean sheet design, can’t we? Block diagrams of Nehalem (and even Core 2) look very similar to Sandy/Ivy, and even die shots will tell you that they aren’t totally different from each other that Sandy could be classified as a clean-sheet design. Clean sheet is going from K10 to Bulldozer. Now that IS clean sheet!

            • sweatshopking
            • 7 years ago

            MY SHEET WAS CLEAN BEFORE READING YOUR POST, BUT IT MADE ME PEE MY BED TO TEACH YOU A LESSON ABOUT ARGUING WITH MY BFF SCOTTY WASSON.

            • ronch
            • 7 years ago

            Still going after that Platinum Subscription, eh? 😉

            • sweatshopking
            • 7 years ago

            Dang skippy

          • the
          • 7 years ago

          Sandy bridge was not a clean start but just an evolution of the Nehalem/Westmere core design. Thus Intel’s primary macro architecture has its roots going back nearly a two decades. You’d never think that Haswell is related to the Pentium Pro without the numerous steps in between.

          Pentium Pro -> Pentium 2/3 -> Banias/Dothan -> Yonah -> Conroe -> Nehalem/Westmere -> Sandy/Ivy Bridge -> Haswell/Broadwell

          I’ve been under the impression that the Bonnell core used in the early Atoms were based upon the original Pentium. Also Larrabee has its roots in the original Pentium as does 48 core Intel’s Single Chip Cloud Computer concept design.
          P54c -> Bonnell (Silverthone) -> Saltwell
          P54c -> Larrabee (no public release) -> Knights Corner
          P54c -> SCC

          Oh, and because Intel has a thing for blast from the past, they dug up the 486 to reappear in a new design:
          486 -> Quark

          The Netburst had several revisions including one that reached tape out but never was released.
          Williamette -> Northwood -> Prescott -> Tejas (unreleased)

          David Kanter is of the belief that architecture in Knights Landing is a threaded version of Silvermont meaning that it has an evolutionary base and not a fresh design.
          Silvermont (new) -> Knights Landing

          So on the Intel side of thing, the only that really appears to be a full clean slate design is Silvermont over the past decade.

          And for reference, here is a bit of AMD’s evolution:
          K7 -> K8 -> K8L (dual core) -> K10 (Phenom) -> K10.5 (Llano)
          Bulldozer (new) -> Piledriver -> Steamroller -> Excavator

          And of course I forgot about Via because well… they’re easy to forget about.
          C3 -> C7 -> Eden
          Isaiah (new)

          So over the past decade or so there have been plenty of evolution but relatively little in terms of new microarchitecture from a clean slate.

            • ronch
            • 7 years ago

            You forgot to mention Bobcat as one of the few clean-sheet designs in recent years. And I believe Silverthorne was also a clean-sheet design.

            • dragosmp
            • 7 years ago

            …also:

            P54c => K5
            Pentium Pro => K6 => K6-2 => K6-3 => K7

            If we see the lineage of these CPU from a design philosophy point of view I think you’re right. On the other hand, if one looks at the actual change between two consecutive cores in the same lineage it is arguable they’re “new” – case in point SB vs Nehalem/Lynfield; they are so different we could say like Intel that SB is new.

            Warning car analogy (I work in this industry): a few years back we (company) had to design a V8, because the V10 we had was banned. Aside from the obvious -2 cylinders we had to redesign everything from the airbox, oil circuit, pistons, camshafts, all the balancing and reduction gears, gearbox; the new V8 was 95% new, but we had applied the same principles for designing the V8 as for the older V10. This means the engines are related, so I agree with your assessment that SB is an evolution of Lynnfield (and further of P-Pro). Company PR was Intel-like “the all new, more efficient, smaller displacement, more power/kilo, lower consumption” and it was in a way, but having moved to real all-new V6 a few years later it seems the old evolution wasn’t quite as revolutionary as the PR made it seem.

      • Tristan
      • 7 years ago

      What looks interesting ? There are not any details. This may be the same crap like phenoms and bulldozers.

        • ronch
        • 7 years ago

        Oh, the pessimist…

          • Klimax
          • 7 years ago

          How many times AMD PR promised and really delivered?

            • ronch
            • 7 years ago

            Don’t mistake AMD’s CTO and lead chip architect for one of those sleezy marketing guys who work for AMD.

      • echo_seven
      • 7 years ago

      I am reallllllly curious how the “Skybridge” concept will play out — to what extent the ARM and x86 infrastructures are common. There’s probably a reason why no one’s tried it before, but you could say the same about the SpaceX legs-on-a-rocket so yeah, really curious.

      And I actually like the name. At least it’s a better code name than Beema and Kabini (reminds me of “beanie”).

        • ronch
        • 7 years ago

        SkyBridge… sounds like a play on Sandy Bridge/Ivy Bridge and Skylake/Skymont (if those are indeed the code names… they seem to have changed or something).

      • ronch
      • 7 years ago

      Yeah. I wonder how wide and fast it would be. And let’s hope TSMC and GF don’t sabotage it!!!

        • Flatland_Spider
        • 7 years ago

        Let’s hope they can talk Intel into letting them fab some chips.

      • NeelyCam
      • 7 years ago

      Wow – I thought you were dead, choked on wings..

      Welcome back!!

      • sweatshopking
      • 7 years ago

      WHERE HAVE YOU BEEN?!?!?!

Pin It on Pinterest

Share This