10 ms·

The End of x86? An Update

104 points by jhund 14y ago

luu 14y ago
Working in microprocessors, I hear this a lot, but, in the long run, Intel has a fundamental advantage over ARM, and ARM doesn't seem to have a fundamental advantage over Intel [1]. People talk about RISC vs. CISC, and how ARM can be lower power because RISC instructions are easier to decode, but I don't hear that from anyone who's actually implemented both an ARM and an x86 front-end [2]. Yes, it's a PITA to decode x86 instructions, but the ARM instruction set isn't very nice, either (e.g., look at how they ran out of opcode space, and overlayed some of their "new" NEON instructions on top of existing instructions by using unused condition codes for existing opcodes). If you want to decode ARM instructions, you'll have to deal with having register fields in different places for different opcodes (which uses extra logic, increasing size and power), decoding deprecated instructions which no one actually uses anymore (e.g., the "DSP" instructions which have mostly been superseded by NEON), etc. x86 is actually more consistent (although decoding variable length instructions isn't easy, either, and you're also stuck with a lot of legacy instructions) [X]. On the other hand, Intel has had a process (manufacturing) advantage since I was in high school (in the late 90s), and that advantage has only increased. Given a comparable design, historically, Intel has had much better performance on a process that's actually cheaper and more reliable [3]. Since Intel has started taking power seriously, they've made huge advances in their low power process. In a generation or two, if Intel turns out a design that's even in the same league as ARM, it's going to be much lower power. This reminds me of when people thought Intel was too slow moving, and was going to be killed by AMD. In reality, they're huge and have many teams working a large variety of different projects. One of those projects paid off and now AMD is doomed. ULV Haswell is supposed to have a TDP ~10W with superior performance to the current Core iX line [4]. Arm's A15 allegedly has a TDP of ~4W, but if you actually benchmark the parts, you'll find that the TDPs aren't measured the same way. A15 uses a ton of power under load, just like Haswell will [5]. When idle, it won't use much power, and will likely have worse leakage, because Intel's process is so good. And then there's Intel's real low power line, which keeps getting better with every generation. Will a ULV version of a high-end Intel part provide much better performance than ARM at the same power in a couple generations, or will a high performance version of a low-power low-cost Intel part provide lower power at the same level of performance and half the price? I don't know, but I bet either one of those two things will happen, or that new project will be unveiled that does something similar. Intel has a ton of resources, and a history of being resilient against the threat of disruption. I'm not saying Intel is infallible, but unlike many big companies, they're agile. This is a company that was a dominant player in the DRAM and SRAM industry that made the conscious decision to drop out the DRAM industry and concentrate on SRAMs when DRAM became less profitable, and then did the same for SRAMs in order to concentrate on microprocessors. And, by the way, they created the first commercially available microprocessor. They're not a Kodak or Polaroid; they're not going to stand idle while their market is disrupted. When Toshiba invented flash memory, Intel actually realized the advantage and quickly became the leading player in flash, leaving Toshiba with the unprofitable DRAM market. If you're going to claim that someone is going to disrupt Intel, you not only have to show that there's an existing advantage, you have to explain why, unlike in other instances, Intel isn't going to respond and use their superior resources to pull ahead. [1] I'm downplaying the advantage of ARM's licensing model, which may be significant. We'll see. Due to economies of scale, there doesn't seem to be room for more than one high performance microprocessor company [6], and yet, there are four companies with ARM architecture licences that design their own processors rather than just licensing IP. TI recently dropped out, and it remains to be seen if it's sustainable for everyone else (or anyone at all). [2] Ex-Transmeta folks, who mostly when to Nvidia, and some other people whose project is not yet public. [3] Remember when IBM was bragging about SOI? Intel's bulk process had comparable power and better performance, not to mention much lower cost and defect rates. [4] http://www.anandtech.com/show/6355/intels-haswell-architecture http://www.anandtech.com/show/6355/intels-haswell-architectu... [5] Haswell hasn't been released yet, but Intel parts that I've looked at have much more conservative TDP estimates than ARM parts, and I don't see any reason to believe that's changed. [6] IBM seems to be losing more money on processors every year, and the people I know at IBM have their resumes polished, because they don't expect POWER development to continue seriously (at least in the U.S.) for more than another generation or two, if that. Oracle is pouring money into SPARC, but it's not clear why, because SPARC has been basically dead for years. MIPS recently disappeared. AMD is in serious trouble. Every other major vendor was wiped out ages ago. The economies of scale are unbelievably large. [X] Sorry, I'm editing this and not renumbering my footnotes. ARMv8 is supposed to address some of this, by creating a large, compatibility breaking, change to the ISA, and having the processor switch modes to maintain compatibility. It's a good idea, but it's not without disadvantages. The good news is, you don't have to deal with all this baggage in the new mode. The bad news is, you still have the legacy decoder sitting there taking up space. And space = speed. Wires are slow, and now you're making everything else travel farther.
- protomyth 14y ago
  Intel's problem is the technology, it is their business practices. ARM is all about customization, and Intel is about commodity motherboards. Intel has restricted every item that might give one PC manufacturer an advantage (e.g. no custom chipsets).
  - ajross 14y ago
    This is missing the point. The third party chipset market died off for exactly the same reason that you posit: all the features that were in the "chipset" are now in the SoC. So what you seem to be saying is that ARM only makes CPU cores (and now a GPU, I guess), so there's a wide market available for NVIDIA and Qualcomm to enter to provide a complete SoC. Intel, on the other hand, makes CPU cores, and GPUs, and display controllers, and DRAM controllers, and USB controllers, and PCI bridges, and audio hardware. And they put all that stuff on a single chip for their customers. ... and somehow you're spinning this as an advantage for ARM Ltd.?
    - protomyth 14y ago
      No, the third party chip chipset market died because Intel killed it. NVIDIA had a superior chipset with a better GPU and Intel stopped them. Intel wants commodity production of a very limited number of designs. Intel doesn't want customizations that can be offered by other companies. Intel wants Intel's GPU not a choice of NVIDIA, S3, ARM, and soon AMD/ATI. Intel wants to ship "Phone Motherboard 1", etc. Its an advantage because ARM lets other companies play and Intel won't.
      - ajross 14y ago
        This is still spinning. What you're saying is still isomorphic to: ARM has an "advantage" because it's a smaller company with fewer product offerings and no ability to fab its own chips. That's just insane. ARM Ltd. (the company) is successful in spite of these facts, not because of them. Maybe what you're really trying to talk about is the "ARM ecosystem", where the big mix of players has a market incentive to try new stuff. And there you might have a point. But it's certainly no disadvantage to Intel specifically -- every one of those players wants to be doing what Intel already is (Apple is very close already, with their own CPU core and SoC design).
mtgx 14y ago
Ever since I've read the innovator's dilemma around 2006 or so, I've tried to watch for other examples of disruptions happening in the tech industry, including laptops (disrupting PC's), iPhone/Android phones (disrupting Nokia/RIM smartphones), iOS/Android (disrupting Windows/Mac OS) and a few others. But while you could still find something to argue about in some of those case, especially when the "fall off a cliff" hasn't happened yet for those companies (disruption takes a few years before it's obvious to everyone, including the company being disrupted), I think the ARM vs Intel/x86 one has been by far the most obvious one, and what I'd consider a "by-the-book" disruption. It's one of the most classical disruption cases I've seen. If Clayton Christensen decides to rewrite the book again in 2020, he'll probably include the ARM vs Intel case study. What will kill Intel is probably not a technical advantage that ARM has and will have. But the pricing advantage. It's irrelevant if Intel can make a $20 chip that is just as good as an ARM one. Intel made good ARM chips a decade ago, too. But the problem is they couldn't live off that. And they wouldn't be able to survive off $20 Atom chips. The "cost structure" of the company is built to support much higher margin chips. They sell 120 mm2 Core chips for $200. But as the articles says, very soon any type of "Core" chip will overshoot most consumers. It has already overshot plenty, because look at how many people are using iPads and Android tablets or smartphones, and they think the performance is more than enough. In fact, as we've seen with some of the comments for Tegra 4 here, they think even these ARM chips are "more than enough" performance wise. That means Intel is destined to compete more and more not against other $200 chips, but against other $20 chips, in the consumer market. So even if they are actually able to compete at that level from a technical point of view, they are fighting a game they can't win. They are fighting by ARM's rules. Just like Innovator's Dilemma says, they will predictably move "up-market" in servers and supercomputers, trying to chase higher-profits as ARM is forcing them to fight with cheaper chips in the consumer market. But as we know ARM is already very serious about the server market, and we'll see what Nvidia intends to do in the supercomputer market eventually with ARM (Project Denver/Boulder). As for Microsoft, which is directly affected by Intel/x86's fate, Apple and Google would be smart to accelerate ARM's takeover of Intel's markets. Because if Microsoft can't use their legacy apps as an advantage against iOS and Android, that means they'll have to start from scratch on the ARM ecosystem, way behind both of them. Apple could do it by using future generations of their own custom-designed ARM CPU in Macbooks, and Google by focusing more on ARM-based Chromebooks, Google TV's, and by ignoring Intel in the mobile market. Linux could take advantage of this, too, because most legacy apps work on ARM by default.
krschultz 14y ago
Whenever I read about modern desktops, I think about the graphs in the Innovator's Dilemma. [1] The incumbent players (Intel, Microsoft, Dell, HP) are all competing on the established metrics of performance & price, but those are no longer the metrics that matter. ARM is pushing the power efficiency angle. Apple is winning on industrial design. The entire computer industry (excluding phones, which has obviously already been disrupted) is right on the verge of being flipped on its head. There were hints of that with the netbook wave, but they weren't quite good enough. The iPad and subsequent high end Android tablets are close, but not 100% there. But we are just about at the point where ARM vs x86 is equivalent for the mass market, and that really is going to shake things up. [1] http://upload.wikimedia.org/wikipedia/commons/thumb/8/8e/Disruptivetechnology.gif/450px-Disruptivetechnology.gif http://upload.wikimedia.org/wikipedia/commons/thumb/8/8e/Dis...
sbov 14y ago
I thought one of the major advantages Intel held was that it owned its own manufacturing, allowing it to iterate more quickly. However, this article seems to claim it needs to shed itself of that. Is it not really an advantage in x86 then? If it is an advantage in x86, why isn't it here? Or is this just a case of blindly copying ARM's business model?
- marshray 14y ago
  Thinking purely objectively, it's always an advantage to have the most money in the bank, the largest customer installed base. Owning the best semiconductor foundaries is a big advantage for the forseeable future. But so once was also owning the world's largest battleship. Things change. How often do the most disruptive changes come from or favor those with the largest physical plant?
- yxhuvud 14y ago
  I think it depends on how it is structured. Owning your own manufacturing is likely more easy to iterate than hiring other to other companies to manufacture for you. What ARM is doing is the opposite - they are the ones hired by the foundries to provide an architecture, and then it is up to the foundries to make the architecture run. This largely decouples the architecture design from manufacturing constraints giving them free reign to innovate. If a foundry can't cope, then that will not hurt the bottom line for ARM since other foundries can, whereas if you own the manufacturing, you actually have to pay to avoid obsoleteness.
programminggeek 14y ago
The problem is 2-fold ARM is being supported by a lot of companies - Apple, Samsung, Microsoft, etc.... vs Intel by itself on x86 basically. Second, ARM chips are too cheap. Intel's biz is built on $100+ chips. ARM chips are like $10. If intel's chip prices drop to say $25, they don't have nearly the money for R&D. X86 won't die, but it can't grow and over time that's going to hamstring Intel. Call it peak x86.
- eliben 14y ago
  Why can't it grow? Server builders still want the fastest CPUs. For them, x86 still gives the best performance per watt and dollar. Yes, ARM servers are starting to appear but I don't think they're up to par with Intel yet. And the cloud needs servers. Lots and lots of servers. True, Intel faces stiff competition here. But folks are sometimes forgetting Intel wasn't always a monopoly in its field. It had competition, lots of it over the years. I wouldn't bury them just yet.
  - st0p 14y ago
    But compared to notebooks, laptops, tablets and smartphones server will always remain a small market.
    - ShirtlessRod 14y ago
      All that data needs to be stored somewhere...
- twotwotwo 14y ago
  I don't know that X86 can't grow over time (servers) but I think you're spot on that dirt cheap chips, and big companies that want to keep them dirt cheap, are in some sense Intel's deeper problem. Intel could make up any given technical gap eventually, but it's hard for them to shut down competition when other, very well-funded players want to maintain it. But on the flipside, ARM right now is very, very low-end -- the fancy new Cortex-A15s only match up against Atoms in single-thread CPU performance, and Atom < ULV < LV < desktop/server. You get away with it in mobile because of lower user expectations and heavy use of the GPU. When you look a little further out (Cortex-A53/57 on newer processes) you can picture ARM in actual client computers, or at least in super-zippy mobile gadgets that some people will happily replace their computers with. (Consumer software will probably adapt to an ARMy world too--use the GPU well, adapt to slower cores, use UI tricks to hide some CPU-caused delays.) But I can't see an ARM chip that acts exactly like a Xeon within the next few years. I bet ARM finds some niches in the datacenter, where servers today have far more CPU than they need, or applications adapt well to a sea of tiny slow cores, or both. (For instance, Facebook uses AMD memcached boxes; they could just as well use ARM, and are looking at it. Intel will make cheap slow cores for those use cases, too.) And I bet ARM will put some price pressure on Intel. But all the things a top-of-the-line Intel chip does to maximize instruction-level parallelism will be really hard for anyone to copy for a very long time.
eliben 14y ago
Oh the horror. What a bunch of random, clueless and non-technical crap. I especially like the part where he compares the operating costs of Intel (a company owning a number of multi-billion-dollar fabs that are far above the competition in capabilities) and ARM holdings (a comparatively tiny intellectual property shop). Unfortunately, it is exactly by the advice of such "strategically thinking" MBAs our industry is often run :-(
- revelation 14y ago
  I like how he quotes the 7.6 billion primitive chips (embedded processors just a notch above a bunch of random logic gates) "shipped" by ARM to the high-margin (for Intel) PC market.
  - makomk 14y ago
    Even ARM's "embedded processors just a notch above a bunch of random logic gates" are fully-fledged 32-bit microprocessors with hardware support for preemptive multitasking, running at fairly respectable clock speeds. And that's the sub-dollar embedded stuff!
- marshray 14y ago
  Yes, it's of course a grapes-to-watermelons comparison. But it's still worth considering. This economic concept of 'dimishing returns' (small competitors can operate more efficiently than large ones) is to a big degree what enables and inspires the HN tech startup scene.
- sgt101 14y ago
  I guess that factoring in the rest of the vertical for ARM is rather difficult to do? Who ships the most ARM chips? Is it Samsung?
  - njyx 14y ago
    It's a lot of white-label Fabs, so manufacturers can take the blueprints, tweak and then do their own runs. I think the author's comparison is spot on - the ARM model is actually a killer advantage. It will actually shrink the value of the market while pushing the volumes through the roof. That sounds like a crazy thing to do - unless you're the one in control of the show.
OldSchool 14y ago
After reading about great performance of newer ARM-based offerings I was surprised when I compared real-world performance at the same clock speed recently: ARM doesn't even come close to any recent-generation x86. This is certainly one very important measure of architecture. A quick sunspider test with a US Samsung Galaxy S3 1.5GHZ snapdragon on Jelly Bean's likely highly-optimized browser shows performance very comparable to a first generation intel 1.66GHZ Atom 230 single core on the latest Firefox. Granted it's a mostly single-threaded test anyway but the ARM has both cores available and the test is pretty cpu-bound after it starts. I'd estimate the latest i7 is at least 3x faster per-GHZ on this lightweight but fairly general (cpu-wise) test. For heavy lifting, a recent i7 with it's cache size, memory bandwidth and accompanying i/o would probably compare to an ARM that is running at about 5x the clock speed. I don't think that ARM can be suddenly declared the best at anything other than maybe performance-per-TDP. Performance-per-cycle is the more difficult problem to solve... ask AMD how hard that's been since the original Intel core series appeared on the scene in 2006. Prior to this and after it wasn't just a chip clone maker, AMD dominated this metric.
- berkut 14y ago
  Exactly - The latest A15 benchmarks show the ARM chips being competitive speed-wise (in terms of computation power) to the first Core-2 Duo from 2006. Intel are 6 years ahead. Yeah, the ARM cores have the power advantage generally at the moment, but under load there won't be much difference looking at what Haswell's ULV seems to be promising compared to the A15, and you can bet that the Intel chips will leak at idle much less and will probably have better sleep states.
- rayiner 14y ago
  But performance per TDP is what matters. Battery technology is getting better much slower than CPU technology. We're already at the point where CPU speed is "good enough," but we're not at the point where battery life is "good enough." I've downsized from a Core 2 MBP to an iPad. It has 1/4 the RAM and runs at half the clock speed. Do I care? No! Web browsing is fast and fluid, photos load plenty fast, editing documents in pages is plenty fast. And it lasts through my whole 12+ hour workday, letting me leave the charger at home and often not even bothering to charge it every night. That's huge and much more important to me, and I'd imagine most people, than whether it can be imperceptibly faster.
bascule 14y ago
Missing from this post is any sort of discussion about how modern x86 CPUs are poorly designed for the types of programs most people are developing these days. Managed language runtimes represent the bulk of programs people are running on servers (think: Java/Scala/Clojure, PHP, Python, Ruby). These environments not only lack "mechanical sympathy" but also have requirements above and beyond what x86 can do. To take Cliff Click's word for it, managed language runtimes consume 1/3 of their memory bandwidth on average zeroing out memory before handing objects to people. If x86 supported an instruction for doing just-in-time zeroing into L1 cache, this penalty could be eliminated, and that 1/3rd of memory bandwidth could be used for actual memory accesses instead of just zeroing out newly allocated objects. In an age where RAM is the new disk, this would be huge. Unfortunately the amount of time it takes to get a feature like this into an Intel CPU is a bit mind boggling. Azul started talking to Intel about hardware transactional memory early last decade, and Intel is finally shipping hardware transactional memory in the Haswell architecture in the form of transactional synchronization extensions.
- brigade 14y ago
  PPC had that (zero one cacheline) and this caused a good amount of complaining when the G5 came out and increased the cacheline size, making dcba/dcbz worthless and dcbzl have different behavour between G4 and G5.
- zvrba 14y ago
  > Unfortunately the amount of time it takes to get a feature like this into an Intel CPU is a bit mind boggling. I think this is an unfair feature comparison. Zeroing L1 cache is way simpler operation than TM which has been designed to support two modes of operation [legacy -- which only speeds up traditional LOCK-based synchronization -- and true TM], must support transaction aborts and restarts, etc. Also, 10 years ago, TM was still a very active research area -- i.e., people had no clue about which ideas were performant and scalable and, not the least, feasible to implement in HW.
dharma1 14y ago
I wonder when we'll start seeing apps running on ARM capable of matching current x86 based content creation apps? 2 years? 3? Intel will catch up with power consumption. The biggest thing going for ARM is price, and because of price their user base is blowing up much faster than Intel, on more types of devices, and in more parts of the world. Most of the developing world's contact with computing is/will be ARM phones and tablets, and the number of people developing software for ARM will skyrocket
Tichy 14y ago
I was disappointed when I tried a fractal simulator on my Nexus 7 recently, and it couldn't zoom smoothly. Perhaps not the most common task in the world, but I think there is still demand for more computing power out there...
- lmm 14y ago
  I used a tablet for a while; now I use an ultrabook. I'm hoping tablets qua tablets are a fad, and going forwards new systems will combine the good parts of tablets (small, light, touchscreen, long battery life) with the good of traditional machines (keyboard, wintel backwards compatibility, performance)
- cageface 14y ago
  There's one on iOS that uses a gpu shader to render and it's plenty smooth.
  - Tichy 14y ago
    Interesting - so maybe there is still something to develop for Android.
jmentz 14y ago
Companies are different from instruction sets, and the disruptor is the ARM instruction set... on Qualcomm. QCOM is already called "the Intel of the mobile world" and, as the world is going mobile, thar be the disruptor.
bhauer 14y ago
Analyses that repeat the “post PC” mantra in its various forms may be correct in doing so, but the mantra is getting threadbare. Since I’ve read this sort of thinking so many times (desktops are dead, Intel is doomed, ARM is the new hotness, etc.), I don’t find it terribly interesting to hear the same restated. Don’t get me wrong, I appreciate the detailed analysis provided by the author, but the thesis is unsurprising. Here’s what I would like to read if a technology journalist could dig it up: What kind of strategic planning is going on within the halls of Intel, Dell, HP, Lenovo, et al with respect to keeping the desktop PC relevant? Put another way: I find it astonishing that several years have been allowed to pass since desktop performance became “good enough.” The key is disrupting what people think is good enough. The average consumer desktop and business desktop user does consider their desktop’s performance to be good enough. But this is an artifact of the manufacturers failing to give consumers anything to lust for. Opinions may vary, but I strongly believe that the major failure for desktop PCs in the past five years has been the display. I use three monitors--two 30” and one 24”--and I want more. I want a 60” desktop display with 200dpi resolution. I would pay dearly for such a display. I want Avatar/Minority Report style UIs (well, a realistic and practical gesture-based UI, but these science-fiction films provided a vision that most people will relate to). I can’t even conceive of how frustrating it is to use a desktop PC with a single monitor, especially something small and low-resolution like a 24 inch 1920x1080 monitor. And yet, most users would consider 24” 1920x1080 to be large and “high definition,” or in other words, “good enough.” That’s the problem, though. As long as users continue to conceive of the desktop in such constrained ways, it seems like a dead-end. You only need so much CPU and GPU horsepower to display 2D Office documents at such a low resolution. There was a great picture CNet had in one of their reports (and I grabbed a copy at my blog [1]) showing a user holding and using a tablet while sitting at a desktop PC. In the photo, the PC has two small monitors and is probably considered good enough to get work done. But the user finds the tablet more productive. This user should be excused for the seemingly inefficient use of resources because it’s probably not actually inefficient at all. The tablet is probably easier to read (crisper, brighter display) and faster, or at least feels faster than the PC simply because it’s newer. Had desktop displays innovated for the past decade, the PC would need to be upgraded. Its CPU, GPU, memory, and most likely disk capacity and network would need to be beefier to drive a large, high-resolution display. So again, what are the PC manufacturers doing to disrupt users’ notions of “good enough,” to make users WANT to upgrade their desktops? I say the display is the key. [1] http://tiamat.tsotech.com/i-see-the-problem http://tiamat.tsotech.com/i-see-the-problem
martinced 14y ago
PC sales, even with the release of Windows 8, did drop 21% compared to one year ago... OK. But would it even be remotely possible to consider that year-to-year China is in recession, a lot of european countries are in recession and U.S. is not in a great position (e.g. the manufacturing sector is firing people left and right), Japan is in a terrible situation, etc. and that this may be playing a role on the number of PCs sold? Year-to-year sales of cars in France has gone down by 20%. When people enter a recession they tend to try to save money: cars and PCs are expensive things. Smartphones not so much (especially with all the "plans" luring people who cannot count in). I think that smartphones and tablet did play a role in the "minus 21%" that TFA mentions but I'm also certain that the worldwide recession is playing a role too. People don't afford what they see as "expensive" that easily. $100 + five-years-unlimited-plan-and-i-can-rape-your-children, they don't pay that much attention and so smartphones tend to be more "recession proof".
tluyben2 14y ago
Not sure about the end of x86 but he is right about one point I've been shouting about for years; mainstream computers are far too powerful for the average user. My mother reads mail and watches pictures of kids/grandkids with her computer; what is the i7/8gb/500gb with a crapload of gpu cores for? Why pay for that kind of power while a cheap Android laptop would easily suffice? My parents, grandparents, uncles or even my siblings and cousins have 0 need for that power nor for Windows. None of them. They notice the difference when they have/touch an iPad or Android pad/computer; they find it easier to wield; they use a handful of apps anyway. So because it has manufacturing advantages, Intel, in my eyes, doesn't have to strive for power or compatibility for the future chips; they just need to use almost no battery. Only thing I hear non computer savvy people talk about is battery life and 'clear screen'. So high res (nexus 10) screens, screens you can view without squeezing your eyes in bright sunlight, solar cells invisibly built in and a few days battery life for a <= $500 price and you'll be selling until silicon runs out. Even for coding you don't really need all that power most of the time; if you are compiling big source trees, sure, but why not just do that in the cloud at EC2 or a dedi server? So you can freely work on your laptop. Game playing and very heavy graphical or music work I can see you need a fast computer in front of your nose for, but further?

104 points by jhund 14y ago

luu 14y ago

Working in microprocessors, I hear this a lot, but, in the long run, Intel has a fundamental advantage over ARM, and ARM doesn't seem to have a fundamental advantage over Intel [1]. People talk about RISC vs. CISC, and how ARM can be lower power because RISC instructions are easier to decode, but I don't hear that from anyone who's actually implemented both an ARM and an x86 front-end [2]. Yes, it's a PITA to decode x86 instructions, but the ARM instruction set isn't very nice, either (e.g., look at how they ran out of opcode space, and overlayed some of their "new" NEON instructions on top of existing instructions by using unused condition codes for existing opcodes). If you want to decode ARM instructions, you'll have to deal with having register fields in different places for different opcodes (which uses extra logic, increasing size and power), decoding deprecated instructions which no one actually uses anymore (e.g., the "DSP" instructions which have mostly been superseded by NEON), etc. x86 is actually more consistent (although decoding variable length instructions isn't easy, either, and you're also stuck with a lot of legacy instructions) [X]. On the other hand, Intel has had a process (manufacturing) advantage since I was in high school (in the late 90s), and that advantage has only increased. Given a comparable design, historically, Intel has had much better performance on a process that's actually cheaper and more reliable [3]. Since Intel has started taking power seriously, they've made huge advances in their low power process. In a generation or two, if Intel turns out a design that's even in the same league as ARM, it's going to be much lower power. This reminds me of when people thought Intel was too slow moving, and was going to be killed by AMD. In reality, they're huge and have many teams working a large variety of different projects. One of those projects paid off and now AMD is doomed. ULV Haswell is supposed to have a TDP ~10W with superior performance to the current Core iX line [4]. Arm's A15 allegedly has a TDP of ~4W, but if you actually benchmark the parts, you'll find that the TDPs aren't measured the same way. A15 uses a ton of power under load, just like Haswell will [5]. When idle, it won't use much power, and will likely have worse leakage, because Intel's process is so good. And then there's Intel's real low power line, which keeps getting better with every generation. Will a ULV version of a high-end Intel part provide much better performance than ARM at the same power in a couple generations, or will a high performance version of a low-power low-cost Intel part provide lower power at the same level of performance and half the price? I don't know, but I bet either one of those two things will happen, or that new project will be unveiled that does something similar. Intel has a ton of resources, and a history of being resilient against the threat of disruption. I'm not saying Intel is infallible, but unlike many big companies, they're agile. This is a company that was a dominant player in the DRAM and SRAM industry that made the conscious decision to drop out the DRAM industry and concentrate on SRAMs when DRAM became less profitable, and then did the same for SRAMs in order to concentrate on microprocessors. And, by the way, they created the first commercially available microprocessor. They're not a Kodak or Polaroid; they're not going to stand idle while their market is disrupted. When Toshiba invented flash memory, Intel actually realized the advantage and quickly became the leading player in flash, leaving Toshiba with the unprofitable DRAM market. If you're going to claim that someone is going to disrupt Intel, you not only have to show that there's an existing advantage, you have to explain why, unlike in other instances, Intel isn't going to respond and use their superior resources to pull ahead. [1] I'm downplaying the advantage of ARM's licensing model, which may be significant. We'll see. Due to economies of scale, there doesn't seem to be room for more than one high performance microprocessor company [6], and yet, there are four companies with ARM architecture licences that design their own processors rather than just licensing IP. TI recently dropped out, and it remains to be seen if it's sustainable for everyone else (or anyone at all). [2] Ex-Transmeta folks, who mostly when to Nvidia, and some other people whose project is not yet public. [3] Remember when IBM was bragging about SOI? Intel's bulk process had comparable power and better performance, not to mention much lower cost and defect rates. [4] http://www.anandtech.com/show/6355/intels-haswell-architecture http://www.anandtech.com/show/6355/intels-haswell-architectu... [5] Haswell hasn't been released yet, but Intel parts that I've looked at have much more conservative TDP estimates than ARM parts, and I don't see any reason to believe that's changed. [6] IBM seems to be losing more money on processors every year, and the people I know at IBM have their resumes polished, because they don't expect POWER development to continue seriously (at least in the U.S.) for more than another generation or two, if that. Oracle is pouring money into SPARC, but it's not clear why, because SPARC has been basically dead for years. MIPS recently disappeared. AMD is in serious trouble. Every other major vendor was wiped out ages ago. The economies of scale are unbelievably large. [X] Sorry, I'm editing this and not renumbering my footnotes. ARMv8 is supposed to address some of this, by creating a large, compatibility breaking, change to the ISA, and having the processor switch modes to maintain compatibility. It's a good idea, but it's not without disadvantages. The good news is, you don't have to deal with all this baggage in the new mode. The bad news is, you still have the legacy decoder sitting there taking up space. And space = speed. Wires are slow, and now you're making everything else travel farther.

protomyth 14y ago
Intel's problem is the technology, it is their business practices. ARM is all about customization, and Intel is about commodity motherboards. Intel has restricted every item that might give one PC manufacturer an advantage (e.g. no custom chipsets).
- ajross 14y ago
  This is missing the point. The third party chipset market died off for exactly the same reason that you posit: all the features that were in the "chipset" are now in the SoC. So what you seem to be saying is that ARM only makes CPU cores (and now a GPU, I guess), so there's a wide market available for NVIDIA and Qualcomm to enter to provide a complete SoC. Intel, on the other hand, makes CPU cores, and GPUs, and display controllers, and DRAM controllers, and USB controllers, and PCI bridges, and audio hardware. And they put all that stuff on a single chip for their customers. ... and somehow you're spinning this as an advantage for ARM Ltd.?
  - protomyth 14y ago
    No, the third party chip chipset market died because Intel killed it. NVIDIA had a superior chipset with a better GPU and Intel stopped them. Intel wants commodity production of a very limited number of designs. Intel doesn't want customizations that can be offered by other companies. Intel wants Intel's GPU not a choice of NVIDIA, S3, ARM, and soon AMD/ATI. Intel wants to ship "Phone Motherboard 1", etc. Its an advantage because ARM lets other companies play and Intel won't.
    - ajross 14y ago
      This is still spinning. What you're saying is still isomorphic to: ARM has an "advantage" because it's a smaller company with fewer product offerings and no ability to fab its own chips. That's just insane. ARM Ltd. (the company) is successful in spite of these facts, not because of them. Maybe what you're really trying to talk about is the "ARM ecosystem", where the big mix of players has a market incentive to try new stuff. And there you might have a point. But it's certainly no disadvantage to Intel specifically -- every one of those players wants to be doing what Intel already is (Apple is very close already, with their own CPU core and SoC design).

mtgx 14y ago

Ever since I've read the innovator's dilemma around 2006 or so, I've tried to watch for other examples of disruptions happening in the tech industry, including laptops (disrupting PC's), iPhone/Android phones (disrupting Nokia/RIM smartphones), iOS/Android (disrupting Windows/Mac OS) and a few others. But while you could still find something to argue about in some of those case, especially when the "fall off a cliff" hasn't happened yet for those companies (disruption takes a few years before it's obvious to everyone, including the company being disrupted), I think the ARM vs Intel/x86 one has been by far the most obvious one, and what I'd consider a "by-the-book" disruption. It's one of the most classical disruption cases I've seen. If Clayton Christensen decides to rewrite the book again in 2020, he'll probably include the ARM vs Intel case study. What will kill Intel is probably not a technical advantage that ARM has and will have. But the pricing advantage. It's irrelevant if Intel can make a $20 chip that is just as good as an ARM one. Intel made good ARM chips a decade ago, too. But the problem is they couldn't live off that. And they wouldn't be able to survive off $20 Atom chips. The "cost structure" of the company is built to support much higher margin chips. They sell 120 mm2 Core chips for $200. But as the articles says, very soon any type of "Core" chip will overshoot most consumers. It has already overshot plenty, because look at how many people are using iPads and Android tablets or smartphones, and they think the performance is more than enough. In fact, as we've seen with some of the comments for Tegra 4 here, they think even these ARM chips are "more than enough" performance wise. That means Intel is destined to compete more and more not against other $200 chips, but against other $20 chips, in the consumer market. So even if they are actually able to compete at that level from a technical point of view, they are fighting a game they can't win. They are fighting by ARM's rules. Just like Innovator's Dilemma says, they will predictably move "up-market" in servers and supercomputers, trying to chase higher-profits as ARM is forcing them to fight with cheaper chips in the consumer market. But as we know ARM is already very serious about the server market, and we'll see what Nvidia intends to do in the supercomputer market eventually with ARM (Project Denver/Boulder). As for Microsoft, which is directly affected by Intel/x86's fate, Apple and Google would be smart to accelerate ARM's takeover of Intel's markets. Because if Microsoft can't use their legacy apps as an advantage against iOS and Android, that means they'll have to start from scratch on the ARM ecosystem, way behind both of them. Apple could do it by using future generations of their own custom-designed ARM CPU in Macbooks, and Google by focusing more on ARM-based Chromebooks, Google TV's, and by ignoring Intel in the mobile market. Linux could take advantage of this, too, because most legacy apps work on ARM by default.

krschultz 14y ago

Whenever I read about modern desktops, I think about the graphs in the Innovator's Dilemma. [1] The incumbent players (Intel, Microsoft, Dell, HP) are all competing on the established metrics of performance & price, but those are no longer the metrics that matter. ARM is pushing the power efficiency angle. Apple is winning on industrial design. The entire computer industry (excluding phones, which has obviously already been disrupted) is right on the verge of being flipped on its head. There were hints of that with the netbook wave, but they weren't quite good enough. The iPad and subsequent high end Android tablets are close, but not 100% there. But we are just about at the point where ARM vs x86 is equivalent for the mass market, and that really is going to shake things up. [1] http://upload.wikimedia.org/wikipedia/commons/thumb/8/8e/Disruptivetechnology.gif/450px-Disruptivetechnology.gif http://upload.wikimedia.org/wikipedia/commons/thumb/8/8e/Dis...

sbov 14y ago

I thought one of the major advantages Intel held was that it owned its own manufacturing, allowing it to iterate more quickly. However, this article seems to claim it needs to shed itself of that. Is it not really an advantage in x86 then? If it is an advantage in x86, why isn't it here? Or is this just a case of blindly copying ARM's business model?

marshray 14y ago
Thinking purely objectively, it's always an advantage to have the most money in the bank, the largest customer installed base. Owning the best semiconductor foundaries is a big advantage for the forseeable future. But so once was also owning the world's largest battleship. Things change. How often do the most disruptive changes come from or favor those with the largest physical plant?
yxhuvud 14y ago
I think it depends on how it is structured. Owning your own manufacturing is likely more easy to iterate than hiring other to other companies to manufacture for you. What ARM is doing is the opposite - they are the ones hired by the foundries to provide an architecture, and then it is up to the foundries to make the architecture run. This largely decouples the architecture design from manufacturing constraints giving them free reign to innovate. If a foundry can't cope, then that will not hurt the bottom line for ARM since other foundries can, whereas if you own the manufacturing, you actually have to pay to avoid obsoleteness.

programminggeek 14y ago

The problem is 2-fold ARM is being supported by a lot of companies - Apple, Samsung, Microsoft, etc.... vs Intel by itself on x86 basically. Second, ARM chips are too cheap. Intel's biz is built on $100+ chips. ARM chips are like $10. If intel's chip prices drop to say $25, they don't have nearly the money for R&D. X86 won't die, but it can't grow and over time that's going to hamstring Intel. Call it peak x86.

eliben 14y ago
Why can't it grow? Server builders still want the fastest CPUs. For them, x86 still gives the best performance per watt and dollar. Yes, ARM servers are starting to appear but I don't think they're up to par with Intel yet. And the cloud needs servers. Lots and lots of servers. True, Intel faces stiff competition here. But folks are sometimes forgetting Intel wasn't always a monopoly in its field. It had competition, lots of it over the years. I wouldn't bury them just yet.
- st0p 14y ago
  But compared to notebooks, laptops, tablets and smartphones server will always remain a small market.
  - ShirtlessRod 14y ago
    All that data needs to be stored somewhere...
twotwotwo 14y ago
I don't know that X86 can't grow over time (servers) but I think you're spot on that dirt cheap chips, and big companies that want to keep them dirt cheap, are in some sense Intel's deeper problem. Intel could make up any given technical gap eventually, but it's hard for them to shut down competition when other, very well-funded players want to maintain it. But on the flipside, ARM right now is very, very low-end -- the fancy new Cortex-A15s only match up against Atoms in single-thread CPU performance, and Atom < ULV < LV < desktop/server. You get away with it in mobile because of lower user expectations and heavy use of the GPU. When you look a little further out (Cortex-A53/57 on newer processes) you can picture ARM in actual client computers, or at least in super-zippy mobile gadgets that some people will happily replace their computers with. (Consumer software will probably adapt to an ARMy world too--use the GPU well, adapt to slower cores, use UI tricks to hide some CPU-caused delays.) But I can't see an ARM chip that acts exactly like a Xeon within the next few years. I bet ARM finds some niches in the datacenter, where servers today have far more CPU than they need, or applications adapt well to a sea of tiny slow cores, or both. (For instance, Facebook uses AMD memcached boxes; they could just as well use ARM, and are looking at it. Intel will make cheap slow cores for those use cases, too.) And I bet ARM will put some price pressure on Intel. But all the things a top-of-the-line Intel chip does to maximize instruction-level parallelism will be really hard for anyone to copy for a very long time.

eliben 14y ago

Oh the horror. What a bunch of random, clueless and non-technical crap. I especially like the part where he compares the operating costs of Intel (a company owning a number of multi-billion-dollar fabs that are far above the competition in capabilities) and ARM holdings (a comparatively tiny intellectual property shop). Unfortunately, it is exactly by the advice of such "strategically thinking" MBAs our industry is often run :-(

revelation 14y ago
I like how he quotes the 7.6 billion primitive chips (embedded processors just a notch above a bunch of random logic gates) "shipped" by ARM to the high-margin (for Intel) PC market.
- makomk 14y ago
  Even ARM's "embedded processors just a notch above a bunch of random logic gates" are fully-fledged 32-bit microprocessors with hardware support for preemptive multitasking, running at fairly respectable clock speeds. And that's the sub-dollar embedded stuff!
marshray 14y ago
Yes, it's of course a grapes-to-watermelons comparison. But it's still worth considering. This economic concept of 'dimishing returns' (small competitors can operate more efficiently than large ones) is to a big degree what enables and inspires the HN tech startup scene.
sgt101 14y ago
I guess that factoring in the rest of the vertical for ARM is rather difficult to do? Who ships the most ARM chips? Is it Samsung?
- njyx 14y ago
  It's a lot of white-label Fabs, so manufacturers can take the blueprints, tweak and then do their own runs. I think the author's comparison is spot on - the ARM model is actually a killer advantage. It will actually shrink the value of the market while pushing the volumes through the roof. That sounds like a crazy thing to do - unless you're the one in control of the show.

OldSchool 14y ago

After reading about great performance of newer ARM-based offerings I was surprised when I compared real-world performance at the same clock speed recently: ARM doesn't even come close to any recent-generation x86. This is certainly one very important measure of architecture. A quick sunspider test with a US Samsung Galaxy S3 1.5GHZ snapdragon on Jelly Bean's likely highly-optimized browser shows performance very comparable to a first generation intel 1.66GHZ Atom 230 single core on the latest Firefox. Granted it's a mostly single-threaded test anyway but the ARM has both cores available and the test is pretty cpu-bound after it starts. I'd estimate the latest i7 is at least 3x faster per-GHZ on this lightweight but fairly general (cpu-wise) test. For heavy lifting, a recent i7 with it's cache size, memory bandwidth and accompanying i/o would probably compare to an ARM that is running at about 5x the clock speed. I don't think that ARM can be suddenly declared the best at anything other than maybe performance-per-TDP. Performance-per-cycle is the more difficult problem to solve... ask AMD how hard that's been since the original Intel core series appeared on the scene in 2006. Prior to this and after it wasn't just a chip clone maker, AMD dominated this metric.

berkut 14y ago
Exactly - The latest A15 benchmarks show the ARM chips being competitive speed-wise (in terms of computation power) to the first Core-2 Duo from 2006. Intel are 6 years ahead. Yeah, the ARM cores have the power advantage generally at the moment, but under load there won't be much difference looking at what Haswell's ULV seems to be promising compared to the A15, and you can bet that the Intel chips will leak at idle much less and will probably have better sleep states.
rayiner 14y ago
But performance per TDP is what matters. Battery technology is getting better much slower than CPU technology. We're already at the point where CPU speed is "good enough," but we're not at the point where battery life is "good enough." I've downsized from a Core 2 MBP to an iPad. It has 1/4 the RAM and runs at half the clock speed. Do I care? No! Web browsing is fast and fluid, photos load plenty fast, editing documents in pages is plenty fast. And it lasts through my whole 12+ hour workday, letting me leave the charger at home and often not even bothering to charge it every night. That's huge and much more important to me, and I'd imagine most people, than whether it can be imperceptibly faster.

bascule 14y ago

Missing from this post is any sort of discussion about how modern x86 CPUs are poorly designed for the types of programs most people are developing these days. Managed language runtimes represent the bulk of programs people are running on servers (think: Java/Scala/Clojure, PHP, Python, Ruby). These environments not only lack "mechanical sympathy" but also have requirements above and beyond what x86 can do. To take Cliff Click's word for it, managed language runtimes consume 1/3 of their memory bandwidth on average zeroing out memory before handing objects to people. If x86 supported an instruction for doing just-in-time zeroing into L1 cache, this penalty could be eliminated, and that 1/3rd of memory bandwidth could be used for actual memory accesses instead of just zeroing out newly allocated objects. In an age where RAM is the new disk, this would be huge. Unfortunately the amount of time it takes to get a feature like this into an Intel CPU is a bit mind boggling. Azul started talking to Intel about hardware transactional memory early last decade, and Intel is finally shipping hardware transactional memory in the Haswell architecture in the form of transactional synchronization extensions.

brigade 14y ago
PPC had that (zero one cacheline) and this caused a good amount of complaining when the G5 came out and increased the cacheline size, making dcba/dcbz worthless and dcbzl have different behavour between G4 and G5.
zvrba 14y ago
> Unfortunately the amount of time it takes to get a feature like this into an Intel CPU is a bit mind boggling. I think this is an unfair feature comparison. Zeroing L1 cache is way simpler operation than TM which has been designed to support two modes of operation [legacy -- which only speeds up traditional LOCK-based synchronization -- and true TM], must support transaction aborts and restarts, etc. Also, 10 years ago, TM was still a very active research area -- i.e., people had no clue about which ideas were performant and scalable and, not the least, feasible to implement in HW.

dharma1 14y ago

I wonder when we'll start seeing apps running on ARM capable of matching current x86 based content creation apps? 2 years? 3? Intel will catch up with power consumption. The biggest thing going for ARM is price, and because of price their user base is blowing up much faster than Intel, on more types of devices, and in more parts of the world. Most of the developing world's contact with computing is/will be ARM phones and tablets, and the number of people developing software for ARM will skyrocket

Tichy 14y ago

I was disappointed when I tried a fractal simulator on my Nexus 7 recently, and it couldn't zoom smoothly. Perhaps not the most common task in the world, but I think there is still demand for more computing power out there...

lmm 14y ago
I used a tablet for a while; now I use an ultrabook. I'm hoping tablets qua tablets are a fad, and going forwards new systems will combine the good parts of tablets (small, light, touchscreen, long battery life) with the good of traditional machines (keyboard, wintel backwards compatibility, performance)
cageface 14y ago
There's one on iOS that uses a gpu shader to render and it's plenty smooth.
- Tichy 14y ago
  Interesting - so maybe there is still something to develop for Android.

jmentz 14y ago

Companies are different from instruction sets, and the disruptor is the ARM instruction set... on Qualcomm. QCOM is already called "the Intel of the mobile world" and, as the world is going mobile, thar be the disruptor.

bhauer 14y ago

Analyses that repeat the “post PC” mantra in its various forms may be correct in doing so, but the mantra is getting threadbare. Since I’ve read this sort of thinking so many times (desktops are dead, Intel is doomed, ARM is the new hotness, etc.), I don’t find it terribly interesting to hear the same restated. Don’t get me wrong, I appreciate the detailed analysis provided by the author, but the thesis is unsurprising. Here’s what I would like to read if a technology journalist could dig it up: What kind of strategic planning is going on within the halls of Intel, Dell, HP, Lenovo, et al with respect to keeping the desktop PC relevant? Put another way: I find it astonishing that several years have been allowed to pass since desktop performance became “good enough.” The key is disrupting what people think is good enough. The average consumer desktop and business desktop user does consider their desktop’s performance to be good enough. But this is an artifact of the manufacturers failing to give consumers anything to lust for. Opinions may vary, but I strongly believe that the major failure for desktop PCs in the past five years has been the display. I use three monitors--two 30” and one 24”--and I want more. I want a 60” desktop display with 200dpi resolution. I would pay dearly for such a display. I want Avatar/Minority Report style UIs (well, a realistic and practical gesture-based UI, but these science-fiction films provided a vision that most people will relate to). I can’t even conceive of how frustrating it is to use a desktop PC with a single monitor, especially something small and low-resolution like a 24 inch 1920x1080 monitor. And yet, most users would consider 24” 1920x1080 to be large and “high definition,” or in other words, “good enough.” That’s the problem, though. As long as users continue to conceive of the desktop in such constrained ways, it seems like a dead-end. You only need so much CPU and GPU horsepower to display 2D Office documents at such a low resolution. There was a great picture CNet had in one of their reports (and I grabbed a copy at my blog [1]) showing a user holding and using a tablet while sitting at a desktop PC. In the photo, the PC has two small monitors and is probably considered good enough to get work done. But the user finds the tablet more productive. This user should be excused for the seemingly inefficient use of resources because it’s probably not actually inefficient at all. The tablet is probably easier to read (crisper, brighter display) and faster, or at least feels faster than the PC simply because it’s newer. Had desktop displays innovated for the past decade, the PC would need to be upgraded. Its CPU, GPU, memory, and most likely disk capacity and network would need to be beefier to drive a large, high-resolution display. So again, what are the PC manufacturers doing to disrupt users’ notions of “good enough,” to make users WANT to upgrade their desktops? I say the display is the key. [1] http://tiamat.tsotech.com/i-see-the-problem http://tiamat.tsotech.com/i-see-the-problem

martinced 14y ago

PC sales, even with the release of Windows 8, did drop 21% compared to one year ago... OK. But would it even be remotely possible to consider that year-to-year China is in recession, a lot of european countries are in recession and U.S. is not in a great position (e.g. the manufacturing sector is firing people left and right), Japan is in a terrible situation, etc. and that this may be playing a role on the number of PCs sold? Year-to-year sales of cars in France has gone down by 20%. When people enter a recession they tend to try to save money: cars and PCs are expensive things. Smartphones not so much (especially with all the "plans" luring people who cannot count in). I think that smartphones and tablet did play a role in the "minus 21%" that TFA mentions but I'm also certain that the worldwide recession is playing a role too. People don't afford what they see as "expensive" that easily. $100 + five-years-unlimited-plan-and-i-can-rape-your-children, they don't pay that much attention and so smartphones tend to be more "recession proof".

tluyben2 14y ago

Not sure about the end of x86 but he is right about one point I've been shouting about for years; mainstream computers are far too powerful for the average user. My mother reads mail and watches pictures of kids/grandkids with her computer; what is the i7/8gb/500gb with a crapload of gpu cores for? Why pay for that kind of power while a cheap Android laptop would easily suffice? My parents, grandparents, uncles or even my siblings and cousins have 0 need for that power nor for Windows. None of them. They notice the difference when they have/touch an iPad or Android pad/computer; they find it easier to wield; they use a handful of apps anyway. So because it has manufacturing advantages, Intel, in my eyes, doesn't have to strive for power or compatibility for the future chips; they just need to use almost no battery. Only thing I hear non computer savvy people talk about is battery life and 'clear screen'. So high res (nexus 10) screens, screens you can view without squeezing your eyes in bright sunlight, solar cells invisibly built in and a few days battery life for a <= $500 price and you'll be selling until silicon runs out. Even for coding you don't really need all that power most of the time; if you are compiling big source trees, sure, but why not just do that in the cloud at EC2 or a dedi server? So you can freely work on your laptop. Game playing and very heavy graphical or music work I can see you need a fast computer in front of your nose for, but further?