Comments Locked

56 Comments

Back to Article

  • Mugur - Monday, January 14, 2013 - link

    It's wrong whether there is a possibility, however remote, to manage to squeeze the chip to more than 7W and blow any thermal solution calculated for this...

    But I wonder if this is actually possible, because of the lowered frequency and voltage.
  • R0H1T - Monday, January 14, 2013 - link

    It seems to me you're trying to cover that epic fail over here ~
    http://www.anandtech.com/show/6529/busting-the-x86...

    I dunno but when it comes to the chipzilla everyone, especially on AT, feels that they(Intel) are so far ahead vis-à-vis the rest of the pack that they might as well stomp over the competition if the antitrust laws would allow'em to do so in the first place aye ? This isn't the case clearly proven by the 32nm SOI Vishera beating i5 at 22nm trigate not to mention the exynos 5x beating anything the Atom has to offer, IIRC the former is half a node back as well, which shows the bias or fanboyism synonymous with the iSheep !
  • scottjames_12 - Monday, January 14, 2013 - link

    On which planet does Vishera beat IVB, especially when it comes to power efficiency? /trollfeeding
  • R0H1T - Monday, January 14, 2013 - link

    Who said anything about power efficiency ? Check performance numbers & benchmarks on any of the sites out there including AT, just FYI single threaded performance doesn't count (cause frankly its a sick joke nowadays) as multicore CPU's are supposed to run stuff on more than one core !

    /not impressed, shrugs off !
  • scottjames_12 - Monday, January 14, 2013 - link

    Oh my apologies, I though seeing as the article is mainly concerning TDP's and power efficiency, that your comment might be on topic.. my bad.

    It is good to see AMD being competitive with heavily threaded workloads. It's a shame that heavily threaded workloads are still so specialised, explaining why Vishera falls behind in the regular benchmarks which are more relevant to just about everyone. Single threaded performance doesn't count? LOL. Ok...
  • R0H1T - Monday, January 14, 2013 - link

    I think you're confused here, or maybe that's just me, anyways here is what I said ~

    1) AT refuted the fact, rightly so mind you, that Intel won't be getting IVB or Haswell in the 7W power envelope cause that's the best case scenario taking mostly light(er) workloads in mind which really isn't how their usual TDP is measured !

    2) The fact is Intel isn't so much ahead of the competition that they can do whatever they'd want to, as most of the AT readers would like to imagine, so incase of ARM they're lagging on the performance front vis-à-vis top of the line Exynos or even Apple's custom made Swift SoC all other things being equal. For AMD they have a full node advantage on the x86 front so really its a no brainer as to why they're more efficient & certainly this Bulldozer architecture has more potential for refinement than the nearly decade old core microarchitecture that has shown signs of aging as proven by the minimal gains of IVB over SNB !

    Lastly if you're so concerned about single threaded performance then why not get a "ye olde" celeron & overclock it to ~8Ghz or so for getting more out of those inefficiently coded programs ? I'll reiterate this ~ just because the software front is lagging doesn't mean that AMD is to be held accountable for their deficiencies !
  • scottjames_12 - Monday, January 14, 2013 - link

    1) Um, does it really matter? If the part has a lower heat output that enables usage in thinner and lighter devices (than the U series parts) does it really matter how the rating is calculated? These parts will be able to be used in devices that are designed to dissipate 7W. Sure, you'll be able to make it exceed 7W with a power virus, but I'm sure they will have something along the lines of configurable TDP that would keep the heat under control.

    2) Who is saying that they are? Although certainly if you look at market share and shipments, it is easy to draw the conclusion that Intel are the ones to beat. I personally don't care or play favourites. I buy whatever is best when I've got money to spend. And regarding IVB vs SNB, you know that IVB was a tick.. right?

    Your last paragraph is pretty childish, really (which is funny, considering your post title). Most people aren't going to buy a processor that is slower 95% of the time, based on the hope that software more suited to it's architecture is 'coming soon'.
  • R0H1T - Monday, January 14, 2013 - link

    IVB was a tick indeed(no pun intended) however it failed to deliver the desired performance gains expected by the majority of users(enthusiasts mainly) which sorta points that the headroom is limited in this case for Intel so AMD will indeed close that inevitable gap(fingers crossed) as & when they move onto 22nm or smaller process nodes.

    Its hard to say whether you're being serious or sarcastic on that single threaded point ! Anyone who's looking for a better single threaded performer(CPU) against one that can perform better at multithreading is clearly going backwards & I think you that too right ?
  • scottjames_12 - Monday, January 14, 2013 - link

    Unfortunately for AMD, they seem to always be a step behind Intel when it comes to process nodes. Sucks for them but that is life. Perhaps Piledriver on a 22nm process would be faster than IVB in all areas, but Intel will have Haswell by then, which should see a bigger performance increase than IVB over SNB. Time will tell.

    I was definitely serious about the threaded point. Just forget 'single vs multithreading' for a moment. The facts are, IVB is faster than Vishera in most usage cases. And it's not like IVB is bad at heavily threaded workloads - when Vishera wins, it only just wins. On the flip side, when Vishera loses, it tends to lose badly. Only a fool would think that single threaded performance doesn't matter, especially when you consider certain types of workloads can't be multi-threaded.
  • R0H1T - Monday, January 14, 2013 - link

    I'd say quite alot of consumer level apps are multicore optimized these days also there's only so much an app can do on a single core. If an app is not scaling well on 2 or more cores then you can pretty much bet that there are better alternatives out there, unless its one of a kind, so yes whilst single threaded performance is an important metric in most benchmarks I certainly wouldn't give it no more than a 10% weightage at any given time !
  • ImSpartacus - Monday, January 14, 2013 - link

    Who said anything about performance? You just said that Vishera was "beating" Ivy Bridge. This article is about power efficiency, so Intel completely "beats" AMD in the context of this article.

    Stay on topic, Troll.
  • R0H1T - Monday, January 14, 2013 - link

    Nice way to budge in, with an insult not to mention going completely bonkers for no reason !

    The article is about Intel advertising its 10~17W TDP parts with another moniker now "SDP" which quite frankly is a special case scenario akin to certain products running under a given set of conditions ! It has nothing to do with AMD also my reference was in context to something that was explained clearly !

    If you're an Intel employee/shareholder then what can I say "learn to read" aye, if not try to respect others & be civil for once at the very least to me !
  • ImSpartacus - Monday, January 14, 2013 - link

    Your thread title is "Seriously grow up !" and I'm the one going bonkers? Troll harder, I'm not mad yet.
  • R0H1T - Monday, January 14, 2013 - link

    I'm pretty sure that was done to wind up folks like you, guess it worked better than I thought |-:
  • Wolfpup - Monday, January 14, 2013 - link

    You keep talking about what Intel's claiming regarding TDP, which as others have pointed out, is 100% irrelevant as these ARM solutions don't even GIVE a TDP (and if they did, would probably be calculated differently too).
  • R0H1T - Monday, January 14, 2013 - link

    My point of contention was just one i.e. Intel was deliberately trying to sell their stuff emphasizing on that 7W SDP marketing B$

    As far as ARM is concerned AT did test the exynos 5x with an estimated 4W SDP, do remember that its an SoC unlike Intel where Broadwell would be the first of its kind, also most ARM SoC's are well within that number. Now the other part of the story is that smartphones/tablets don't need a 300$ chip especially considering the fact that their desktop counterparts aren't running even moderately optimized software that can truly utilize 2 or more cores even after ~10yrs of multicore revolution !
  • silverblue - Monday, January 14, 2013 - link

    Only per-core, but the performance gap is at times quite severe.
  • jjj - Monday, January 14, 2013 - link

    "how far Intel is away from being able to fit Core into an iPad or Nexus 10 style device"
    That's not relevant in any way with Intel's prices. A chip that costs 100$ more than the ARM SoC ads 300$ to the ipad's retail price, good luck selling that.
    The Tegra comparison is with T3 and that one is somewhat crippled by the memory subsystem. Early 20nm A15 could be half of what Intel will have then in the low clock realm ( by then ARM folks might enable using all 8 cores in bigLITTLE so the math gets more complicated). Second half of 2014 we could see A5x series although i wouldn't bet on it and maybe finFET .But core vs core perf has little relevance, you got die size and plenty of other compute units to factor in.
    Intel is stuck with huge margins ,if they give that up with nothing that adds a lot of revenue they wipe out 50B of market cap and that's money lost by someone. We'll see what the new CEO does , i hope he'll be more willing to take risks , Intel is no fun lately.
  • wsw1982 - Monday, January 14, 2013 - link

    You should compare product in same category. A15 only has Atom level performance so far, and is beaten by the upcoming ATOM 32nm z2580 with two 5-years-old architecture.(checkout the Antutu benchmark on Lenovo K900, and compare it with Nexus 10, and keep in mind, the z2580 is likely a less than 2w part but Exynos 5 4w part). And if you compare the price of the Atom with the ARM parts, you can see they are quite competitive. The core series could scale it's power consumption down to ARM level doesn't necessary means it's a ARM level processor. You see how they outperform the ARM level atom
  • ypsylon - Monday, January 14, 2013 - link

    At least from my point of view. All fine with low voltage. Would love to build that kind of system myself. Useless because I can't build such system as it is SoC. I'm one of those who will always (or as long as possible with all rumors flying around) buy motherboards, CPUs and stuff. Detest laptops/netbooks/tablets and everything connected with that family in any way.
  • MonkeyPaw - Monday, January 14, 2013 - link

    Actually, you bring up a point that has been totally overlooked thus far. These are not SOCs, just lower leakage parts and a suspect marketing approach to reach 7W*. Still need to factor in a southbridge. That's more power and more cost. If Intel has its way, tablet prices will either go up $200, or they will start getting the terrible screens that plague notebooks.
  • fkanker - Monday, January 14, 2013 - link

    Anand how can you cover Intel on such a thing? SDP is misleading. What is "a typical use"? Will today's typical use be the same in one year? Intel should rate 7W as a TDP without any violation. The same applies to the Exynos 5250 you took as an example to justify the SDP, it just shouldn't be used in a tablet.
  • wsw1982 - Monday, January 14, 2013 - link

    Yet both Samsung and ARM declare the A15 consume less power than A9...
  • fkanker - Monday, January 14, 2013 - link

    Well, it is your right to believe...
  • Klimax - Monday, January 14, 2013 - link

    What belief?
  • fkanker - Monday, January 14, 2013 - link

    That's his right to believe both Samsung and ARM declaration ;)
  • wsw1982 - Monday, January 14, 2013 - link

    I guess you mean "integrity" :P
  • fkanker - Monday, January 14, 2013 - link

    If you want so. But what's your point?
  • wsw1982 - Monday, January 14, 2013 - link

    My point is any company bluffing
  • fkanker - Monday, January 14, 2013 - link

    And because everyone's bluffing it should be covered like Anand is covering Intel's SDP here?
  • wsw1982 - Tuesday, January 15, 2013 - link

    I think SDP is actually a more useful number when you use it to compare to ARM parts. Because as you said, the TABLET from ARM don't use TDP or EVEN SDP, and Intel 's chips here is compete with those chips that mostly only publish their lowest possible power consumption during no-load. And you can only wandering the SDP of it by each individual products, but at least it's measurable for both ARM and X86.

    The consumer do need a cross platform consistence comparison matrices, do they?

    When I first heard Nvidia comparing tegra 2 with ATOM, they use the lowest power consumption of tegra and TDP of ATOM (2.7w) back then, I was impressed...
  • Wolfpup - Monday, January 14, 2013 - link

    I'd be surprised if they claimed such a thing, but at a given process node/design that's obviously false.
  • jwcalla - Monday, January 14, 2013 - link

    They probably claim that the A15 is more power-efficient (i.e., GFLOPS / W). The A15 should be significantly more power-efficient than the A9, putting aside the GPU of course.
  • wsw1982 - Tuesday, January 15, 2013 - link

    ARM/ARM partners actually measure/publish the best performance during full load, and lowest idle power during no load. They get at least consistence...
  • MrSpadge - Monday, January 14, 2013 - link

    Capping power limits hard is inefficient. Look at Intels Turbo mode and how chips are allow to shortly spike above TDP, as long as temperatures and cooling are fine. This improves responsiveness, i.e. how fast a system actually feels. Thermals only enter the equation under sustained loads.
  • fkanker - Monday, January 14, 2013 - link

    And that's the problem with SDP. Under load, depending on the cooling efficiency, you might end up with a given CPU reference performing differently across different devices.
    Under load the base clock might also not be respected, i. e. the CPU will throttle to keep 7W/Tj80C. The performance becomes unpredictable.

    Moreover, we don't have any exact information on the way SDP is defined. The considered workloads might as well be heavier in one year, maybe mix more CPU and GPU loads, who knows?
  • rigel84 - Monday, January 14, 2013 - link

    Hi, can someone explain why ARM are so far ahead of Intel and AMD in the tablet and smartphone area? And why don't we see ARM CPUs for the desktop if they use so little power, wouldn't ARM just have to overclock the lot until they hit a TDP of 77 watt like the i3570, and then beat it?
  • Klimax - Monday, January 14, 2013 - link

    I think they loose efficiency too fast and still too weak performance for devices from 17W and higher. (There's a reason why Intel's engineers don't use too high frequencies, which would be probably required for ARM)
  • MrSpadge - Monday, January 14, 2013 - link

    Simple: they are not. By now their newest and greatest, the A15 manages to slightly outperform the old Atom design, at about 2x the power consumption. That's still rather slow for desktops. And "simply overclocking" won't change this significantly: the entire chip is made for moderate power consumption at modest frequencies. Designing for higher frequencies automatically reduces power efficiency, let alone higher voltages to reach higher clock speeds. And using more cores won't help percieved performance, since this is still laregely dominated by single thread performance (that's why Brazos feels much snappier than Atom, despite comparable multi-thread performance).
  • fkanker - Monday, January 14, 2013 - link

    I would rather say Exynos 5250 instead of "A15". The A15 is just a design that has to be implemented, and can even be modified. The SoC designer must also add its own memory controller, which can lead to big performance differences. Like AMD implementing Ivy Bridge with a memory controller of its own would result in a "slighty" different product than a Core i7.

    With the Exynos 5250 I have the feeling that Samsung made a quick and crappy implementation just to have the first A15-based SoC on the market. Its power consumption just doesn't seem logical compared to A9-based SoCs (Tegra 3, OMAP4,...). So I'd rather wait Tegra 4 and Exynos 54xx reviews before concluding on power consumption of the A15-based SoCs.
  • Kevin G - Monday, January 14, 2013 - link

    The main reason why ARM are so far ahead of Intel and AMD in terms of market share in the mobile area is due to ARM's historically low power consumption compared to desktop centric processors and ARM's wide use in the embedded market. Remember that feature phones spawned from the embedded market and smart phones diverged from feature phones. The other factor for ARM's popularity is that they'll license their core designs or the entire architecture.

    As to why we don't see ARM in the desktop, much of it has to do with the dominance of Windows and it being x86 focused. Outside of Itanium support in the server world (which Windows 2008R2 was the last), MS hasn't had a version of Windows desktop running on a processor other than an x86 chip for nearly a decade. Windows RT last year changed that and the possibility of a Windows based ARM desktop now exists. However, due to MS vision that Windows RT is to be used in mobile devices, the chances of any manufacturer releasing a Windows RT desktop are slim to none. Of course Windows isn't the only desktop OS as there is Mac OS X and Linux. OS X and iOS are predicted to merge to a common hardware platform. With Apple designing their own ARM based cores, seeing a Mac Mini or MacBook based around an ARM core is more likely but years away. Linux on the desktop using ARM exists but the hardware is more for the DIY hobbyist than a mainstream consumer product. Linux on ARM desktops can happen but ARM manufacturers have to expand their SoC capabilities to really compete. How many ARM SoC's have support for PCI-E 16x for example?

    Increasing the core voltage and clocking an ARM core higher is a distinct possibility for increasing performance on the desktop. The problem is that current ARM designs are no geared toward high clock speeds due to their mobile, low power roots. A high performance ARM chip is certainly possible and ARM themselves is seemingly eager to design such a core. As such I'd only expect a Cortex A9 design to scale upward of 2.5 Ghz if given TDP leniency of a desktop. Cortex A15's are a bit more high performance but I'd be surprised if they could get past 3 Ghz. With IPC on the same level of Atom, even with such a clock speed boost their performance compared to desktop Ivy Bridge or Pile Driver would be lack luster. ARM does have a desire to move into the low end server segment so they likely have a real high performance design that'll only be used in desktops/servers in the works. Again, such a design would be several years away.
  • Wolfpup - Monday, January 14, 2013 - link

    Because ARM's focus is on very low end chips, and Intel's has traditionally been on high end chips. It's only recently that ARM's designs have started to get into the realm of...well...comparatively ancient x86 chips, and Intel's started scaling things down to hit lower price points and power usage.

    "And why don't we see ARM CPUs for the desktop if they use so little power"

    They use little power because they're very low performance. Atom and AMD's equivalents already have that covered in CPUs that are at least as efficient, but of course also compatible.
  • jemima puddle-duck - Monday, January 14, 2013 - link

    Price is all that matters. The only people that care about speed are nerds, and there aren't enough of them to bother with making it.
  • pSupaNova - Monday, January 14, 2013 - link

    True, especially when the Nexus 10 retails at $399 has a screen with more pixels than a 1080p TV.

    Intel can't enter this game with out shaving of huge margins.

    When ARM starts putting out 64bit parts say good by to the ultra book segment.
  • StormyParis - Monday, January 14, 2013 - link

    Hey, TDP is important, and too high. And Intel's 7W SDP is a joke, those guys are lamerz. I've got CPUs with a 3.5W PFOOMADP, sorry, that acronym while truer doesn't ring similar enough to TDP. What TLA ending is DP is still available people ? No, not GDP. Not PDP, sounds musty for some reason. Hey ZDP it is !

    My 3.5W ZDP chips beat your 7W SDP, Intel. Eat that !

    PR: "Revolutionary low-ZDP CPUs are available right now from ZDP inc,, at a low multiple over old-school TDP CPUs. Low-ZDP CPUs. ZFP is a new thermal dissipation figure, that more truely reflects a CPU's actual real-life workload over a period of time. "

    Most of you spend a lot of your day idling and sleeping. So do our CPUs, and they too should get credit for that !
  • Wolfpup - Monday, January 14, 2013 - link

    You're obsessing over a number Intel gave and ignoring that these parts are more powerful than ARM's stuff, at power levels that either the same or not much worse.

    I really don't understand why you and others are obsessing on this, completely missing the actual point.
  • mikk - Monday, January 14, 2013 - link

    What about this: http://www.theverge.com/2013/1/9/3856050/intel-can...

    It is true, i7-3689Y runs at only 800MHz by default?
  • Anand Lal Shimpi - Monday, January 14, 2013 - link

    I believe that is true for all Core processors. When idle everything will step down to 800MHz, Y-series or not.

    Take care,
    Anand
  • mrdude - Monday, January 14, 2013 - link

    Did they mention how much throttling we're going to get with these chips? It's all fine and dandy to declare something a 7W SDP (with an unknown workload...?) but if getting to that low SDP is going to be achieved by persistent throttling - which is a pretty safe bet - then what's the point?

    For example, what about the *SDP* while gaming or other dual load scenarios? Would it be closer to that TDP, thus that SDP would be achieved by hammering down via the C/P-states? In essence, all you'd be paying for is a costlier chip that can't work at advertised speeds for more than a fraction of a second.

    The pricing just goes to show just how out of touch Intel is with reality. Their chipset alone costs as much as a modern day ARM SoC. Intel's selling premium performance at skyrocket prices to an audience who doesn't care for it.

    Good luck to them.
  • mikk - Monday, January 14, 2013 - link

    Hmm afaik Idle frequency on the bigger Ivy Bridge CPUs are 1,6 Ghz. Base clock is a different thing. Base clock is applied under load. (it mostly runs with turbo frequency) Intel ark says 1,1+ Ghz base clock for Y-Series. theverge says 800 Mhz. Someone is wrong.
  • wsw1982 - Monday, January 14, 2013 - link

    My Lenovo W520 running at 800Mhz when idle/under light loading. And it has a ?2.6GHz q2820 with TDP of 45w
  • mikk - Monday, January 14, 2013 - link

    Mobile CPUs might have a lower idle state. Maybe theverge mixed up Idle state with base frequency...
  • Quizzical - Tuesday, January 15, 2013 - link

    In other words, sometimes the chip won't use more than 7 W. If you build a tablet around it that can only dissipate 7 W, then sometimes it won't fry. Sometimes.

    Stupid acronyms are just begging for some crank to turn it into a sarcastic backronym.
  • jamannetje - Tuesday, January 15, 2013 - link

    With these 13W parts being rather expensive ($250) I wonder how low could one go with a standard part, let's say a Intel Core i3-2130, which is a comparable product at half the price. At the same frequencies as the Y-parts and a lowered voltage what would be de TDP?
  • MarcHFR - Wednesday, January 16, 2013 - link

    "At the end of the day, 7W SDP Ivy Bridge (and future parts) are good for the industry"

    But bad for customers, who will not know the performance of what they buy. Isn't that more important, Anand ?
  • x86 user - Thursday, May 9, 2013 - link

    We (university outfit) develop a Windows 8 device (prototyped as of now) using an Intel i5u (1.7 GHz @17W Total Power Dissipation) and planning on going into production (India/China Region) with about 100K units a year for our users. The total estimated build cost of this device is around $500/- at the 100K/year volume range.

    http://www.eyepuzzles.net/dimensions.pdf

    The battery life on this device is about 8+ hours (with a 6 cell lithium ion). Planning on migrating to a 1.7GHz Haswell Mobile CPU. What would be the expected TDP on this processor, since today the above device (which is a Windows 8 tablet/laptop/docked workstation with a built in GPS, GSM phone & Accelerometer) has a max power consumption of around 25 watts.

Log in

Don't have an account? Sign up now