Comments Locked

115 Comments

Back to Article

  • coburn_c - Friday, September 5, 2014 - link

    Wonderful chips, hopefully this will strange ARM in the market. Of course it also strangles Atom... but still wonderful chips.
  • azazel1024 - Friday, September 5, 2014 - link

    Not really to both. Unless Intel is slashing profits to the bone, it is a MUCH larger die than Atom. Atom is targeted at more budget models and is also at a higher TDP than Atom. Especially once Braswell and Airmont hit.

    I doubt you'll see many/any Core M in 10.1" tablets, despite the possibility.

    When you are talking probably $150-200 in 1000 trays for the low end processor (Intel could surpise me and price is sub $100, but I doubt it), you aren't going to be finding them in $400-500 tablets.

    Lets also keep in mind, Atom is getting more advanced too. Silvermont appears to have roughly 35% of the IPC per core that Haswell did and it also generally had a clock speed disadvantage. Which is why in single threaded tasks you'd see something like an i5-4200 turning in roughly 2.5x the single thread performance of the z3770, but only about a .8x advantage in heavily multithreaded applications.

    Now here, the upcoming top end Airmont process and even the current z3770 will actually have a clock speed advantage, and might also have some IPC tweaks (granted, looks like Broadwell is likely to have a 5-8% IPC gain over Haswell). So, yeah, Core M is likely to stay ahead of Atom/Airmont...but I'd bet that the performance delta between the two, especially if you look at the more affordable/probably more common 5Y10 is probably going to shrink a lot.

    Graphically...well it looks like improvements, but clock speed reductions for Core M, and Airmont is supposed to get 16EU, though probably at even further reduced clock speed...but it'll be a GPU performance delta shrink from roughly 4x advantage for Haswell right now, to maybe at most 2x for Broadwell M, maybe more like 1.5x.

    So if anything, this means that Core M is more suitable for tablets, but maybe not the smaller 10.1" tablets...and that the performance between Core M and Atom is only likely to shrink, while still having a pretty wide gulf in cost (seeing has how the latest z3775 is only about $33 a pop, comapred to i5-4200 which I think is around $200 a pop).
  • coburn_c - Sunday, September 7, 2014 - link

    I don't even... these match the lowest TDP Atom currently available and easily quadruple its performance, as it is a dual core chip. You have a lot of eclectic information in your post that you then use to draw outlandish conclusions. The only relevant fact you state is that these are expensive. These also are capable. You can actually run useful software on these, not just your starbucks app. This is scaling Desktop performance down into the mobile space and doing it properly, for the first time. This is huge.
  • duploxxx - Monday, September 8, 2014 - link

    this is huge indeed. :) and expect a lot from these chips although they will be expensive

    it has a huge turbo fo both CPU and GPU and a low TDP.
    so how much will it throttle and decrease performance of both cpu/gpu once it requires both resources ????
  • Wolfpup - Monday, September 8, 2014 - link

    I can't wait for this to hit Surface Pro 3! Maybe a fanless design, or more performance...
  • PEJUman - Tuesday, October 7, 2014 - link

    I own z3745, i3 3127u, i7 4770k, and AMD A10 7850k & c-60. Azazel's remarks are quite similar to my own findings from benchmarking the above toys. I find his claim very reasonable. What data do you have that contradicts his opinions?
  • name99 - Friday, September 5, 2014 - link

    "hopefully this will strang[l]e ARM in the market"
    At $281 a pop? Yeah sure it will.

    Meanwhile, say goodbye to any dreams of a Surface Pro that matches, let alone is cheaper than, iPad prices.
  • mkozakewich - Saturday, September 6, 2014 - link

    It was never meant to be directly compared to an iPad in features. There are huge differences. The iPad has a denser, higher-accuracy display and the Surface Pro has things like pen and application support.

    (Though, to be honest, it's the fault of the developer culture that the iPad hasn't seen more productivity apps. Office, Photoshop, Autocad, and more could all have been ported to the iPad by now, but they just haven't.)
  • name99 - Saturday, September 6, 2014 - link

    My point was that every time a new Surface Pro is released, the NUMBER ONE complaint (just look through the AnandTech reviews) is the cost...
  • peterfares - Sunday, September 7, 2014 - link

    That's because people are being silly and comparing it to the iPad instead of realizing it's an ultrabook spec tablet, not a weak ARM or ATOM tablet
  • Alexey291 - Sunday, September 7, 2014 - link

    No no that's because people don't want a shitty 2 in 1 that's crap at either of the specific tasks (ultrabook OR tablet) AND THEN it costs an arm and a leg.
  • Wolfpup - Monday, September 8, 2014 - link

    Surface Pro is excellent at both. Where are you getting the idea that it's not? It's a WAAAAAAAAAAAAAAY better tablet than an iPad. I've got an iPad and an Android tablet, but there's no way I'd buy another of either since Windows 8 tablets started launching.
  • Alexey291 - Thursday, September 11, 2014 - link

    Well I'm glad that you found yourself something you like.

    I am on the other hand very glad that I don't have to use windows for ANYTHING in my life. Gaming is literally the very last thing. And even that I usually stream on to something more convenient.

    And also the thought of having to run a security suite (don't kid yourself - windows needs an antivirus, a firewall and a malware scanner, and ms security essentials don't work for shit) on a tablet which has fans and next to no usable space (all thanks to windoze) is so thrilling I'm going to pee myself from excitement.

    All that for the sake of doing what exactly? Running ms office? What other real benefits are there from running windows on a tablet again?

    Oh and the keyboard is shit. I've tried all 3 iterations of it so far (and 2 and 3 are pretty much the same kb) and they are all pretty much comparable with 25 - 30 quid bluetooth keyboards from amazon. SP's kb is better but not by much. You can't even try to compare it to a proper keyboard on an ultrabook.
  • Krysto - Saturday, September 6, 2014 - link

    Ultrabooks have been a dead-end for Intel and Microsoft. Nobody wants them over Macbook Air. Surface Pro is essentially an ultrabook, that's a competitor to Macbook Air. I bet it doesn't sell even 1/100 of Macbook Air. In fact, I'd be surprised if Satya Nadella releases a Sufrace Pro 4 anymore.
  • peterfares - Sunday, September 7, 2014 - link

    Plenty of people want ultrabooks. Lots of people want Macbook Airs for some reason (terrible screen) but I doubt Intel cares which people choose.
  • peterfares - Sunday, September 7, 2014 - link

    The iPad doesn't have good enough I/O for real productive use. It's a content consumption device and that's all Apple wants it to be.
  • Wolfpup - Monday, September 8, 2014 - link

    It's doesn't have enough RAM either, and the OS really isn't up to it either. iOS is really a terrible tablet OS.
  • Laxaa - Friday, September 5, 2014 - link

    So, the M-5Y70 in the Surface Pro 4 next year?
  • kchase731 - Friday, September 5, 2014 - link

    Would be awesome to see a vPro surface.
  • Thermogenic - Friday, September 5, 2014 - link

    Their earlier marketing material talked about the need for a fanless core cpu in a 9mm chassis - essentially talking about the Surface Pro 3. I have no doubt you'll see a Surface Pro in February 2015 using these chips.
  • mkozakewich - Saturday, September 6, 2014 - link

    I kinda hope MS will include a fan just to improve turbo performance. Like, it doesn't even have to be on when the workload is light.
  • MrSpadge - Saturday, September 6, 2014 - link

    I agree.. I'd rather have the option to turn on a fan if I need performance now (maybe I'm not on battery?). Building a PC around such a wonderful SoC like Braswell and then "crippling" its performance by demanding "always passive cooling" does not seem right for power users.
  • peterfares - Sunday, September 7, 2014 - link

    That would be cool. Even take it to the next level and have it be user-controlled. Have a performance mode where the fan can turn on and a quiet mode where the fan never turns on.
  • Samus - Saturday, September 6, 2014 - link

    The problem is these chips are substantially lower performing than all Surface Pro CPU's have been thus far (perhaps with the exception of the elusive i3 edition of Surface Pro 3)

    They are also slightly more expensive than the current i5-4300U used in Surface Pro 2/3. Yes, the 4300U is triple the TDP (15W) but the Surface actually does a remarkable job handling the heat (in my opinion) so I think until Intel can product a chip on par with the performance of current Ultrabook CPU's in Broadwell, the Surface may alienate its design model by REDUCING performance in the next model.

    Which leaves Microsoft in an interesting position. I love the Surface, but its pretty clear its been a failure for Microsoft financially. Are they willing to risk a THIRD, middle-ground model to fit between Surface "RT" and Surface Pro? That's what they should do, but will they do a Surface "Premium" edition for ~$600?
  • hero4hire - Sunday, September 7, 2014 - link

    Microsoft's entire business model is indirect sales. Xbox aside...I don't think they want or need a 30% margin air or even a 5% margin successful pc builder into the fold. Surface was born out of the poor designs and crapware loaded laptops while they looking down the coast at the "revolutionary" design wins at Apple. If a partner put out a ultraviolet, tablet , lappy, that sells better then osx at the high end, I think ms would happily let surface retire.
  • Krysto - Saturday, September 6, 2014 - link

    Well ask yourself this question: do you want a 800 Mhz CPU in a $1,000 laptop?
  • Klimax - Saturday, September 6, 2014 - link

    Not thermally constrained like Haswell units already present? Since it will likely massacre any toys like Ipads, then your question is nothing but nonsensical.
  • Samus - Saturday, September 6, 2014 - link

    He said $1000 laptop. iPad's are $300-$500 and not comparable.

    However, $800 Surface 2-3 Pro's with i5-4300U's are 2x the performance of these parts, and although 15w, are still incredibly light and thin. Did I mention 2x the performance of these parts?

    Yeah, nearly $300 for these chips is robbery. They're only slightly faster than Bay Trail, and for burst performance, the previous Haswell U chips will be substantially faster, even if "thermally constrained."

    It has yet to be seen if these chips will even survive prolonged load without throttling like Haswell does in certain applications. And God forbid they throttle, who knows how slow they'll get...for 4.5w parts (that's 1/3 the TDP of Haswell) I'm frankly shocked how unsubstantial the battery life improvements are. I just don't know why they didn't bump the power envelope to 6-7w to get them clocked at 2GHz base.
  • Speedfriend - Monday, September 8, 2014 - link

    "He said $1000 laptop. iPad's are $300-$500 and not comparable."

    Lets be realistic, a 128gb iPad Air is $800, well into laptop price territory.
  • Kidster3001 - Friday, October 3, 2014 - link

    Intel parts don't throttle below Base speeds. They will always run at at least the Base speed. Throttling refers to limiting Turbo speeds.
  • Morawka - Friday, September 5, 2014 - link

    can you guys not afford a dslr for the front page photo? This looks like it's been taken with a flip phone.
  • fokka - Friday, September 5, 2014 - link

    haha, what i was thinking. like it has been run through a bad hdr filter.
  • haukionkannel - Friday, September 5, 2014 - link

    It would be interesting to see these tested against Intel Atoms. How much more there is speed in these and how much more these eats electricity compared to atoms... These would be very interesting in for example Surface pro mini tablets!
  • azazel1024 - Friday, September 5, 2014 - link

    I'd guess the power delta is pretty minor, at least 22nm Bay Trail versus 14nm Broadwell M. Now, against the upcoming 14nm Airmont/Braswell, it would probably be a pretty significantly delta.

    IIRC Bay Trail-T is a 4.5w TDP as well (or roughly that), though I suspect in normal use is likely to bat a fair amount lower than Broadwell M.

    Performance wise, see my comment above. The proof is in the puddling, but some basic extrapolation I'd guess that both are going to edge closer to Atom than they are to increase the current performance gulf. However, Airmont especially in Cherry Trail configurations, is likely to reduce the performance delta rather significantly, especially in GPU ability, but probably in CPU as well (its a 200Mhz increase, and probably/possibly IPC improvements...and they might be more than very minor IPC improvements, though I am betting on minor). Where as Broadwell M is a step back in CPU frequency at least (compared to ULT Haswell, which is the only baseline I have for Atom versus Core performance). Though compared to Haswell Y, it is probably a significant increase in performance.
  • mkozakewich - Saturday, September 6, 2014 - link

    Considering the CPU in the charts above was running at 1 W and was only a fifth of the total platform power, things are definitely getting to a point where we have to shift our focus to the high-power aspects. Like, at full power, how much performance per Watt do we get with Broadwell? How about with Atom. Yeah, I'm definitely waiting for that.
  • Krysto - Saturday, September 6, 2014 - link

    Who cares about "full power", whatever that's supposed to mean. What matters is how it does in its TDP range. I don't care if Broadwell can be 10x faster at 200W, if 99 percent of the Broadwell chips they want to sell are limited to 4.5W TDP. THAT'S where you wanto compare them. The rest is irrelevant.
  • Krysto - Saturday, September 6, 2014 - link

    Exactly. That's what I want to see too. Let's compare this $280 chip to an Atom and also the $25 Nvidia Denver CPU and Apple A8. Then we'll see how truly pathetic these chips are.

    I bet they aren't more than 50 percent faster.
  • Klimax - Saturday, September 6, 2014 - link

    Another jump to conclusions. My prediction is: Even massively constrained Broadwell will make short work of any of those chips.
  • darkich - Saturday, September 6, 2014 - link

    Lol, no.
    Just no. Mark my words it won't be even 50% faster.
    And just wait the hilarity of the 8 core A57+A53 Mali T760Mp 12 20nm/16nm chips 6 months from now, almost matching the Core M..in a phone.
  • Speedfriend - Monday, September 8, 2014 - link

    An A57+A53 8 core will be up to 1.5x faster than an octa A15 chip, which was probably 10-20% slower than a Baytrail. So you are assuming that Core-M will only be 30% odd faster than a Baytrail? And lets not forge tthat the graphics are likely to be significantly better in Core M. That seems ridiculous, but I suppose we will have to wait and see. The only performance guidelines for core M that I have seen is that it will be 2x a fast as a 2010 Nehalem at 1/4 the TDP.
  • Wolfpup - Monday, September 8, 2014 - link

    I'm baffled by all these comments that don't seem impressed by this and seem to think 28nm phone chips are in some way comparable to 14nm chips based on Intel's latest and greatest :-O
  • Alexey291 - Thursday, September 11, 2014 - link

    Considering how badly intel has been doing in the mobile space I'm not baffled by these comments at all.

    And given the price of these chips I'm not seeing them sell any significant quantity.
  • fteoath64 - Tuesday, September 9, 2014 - link

    "interesting to see these tested against Intel Atoms." These parts are clocked at 800Mhz at 4.5w with 2 cores : the BayTrial clocked at 2Ghz with 4 cores at 10w. Granted the gpu of BT is just poor compared to GT3e in these chips. ie 40EUs with eDram. SO on cpu, the BT would be 40% as fast approx.
  • ArthurG - Friday, September 5, 2014 - link

    Dear Yan,
    it's a nice article but reality is a bitch. These "wonderful" (irony) products are already end of life !!!
    see here: http://qdms.intel.com/dm/d.aspx/824A06BE-D3AD-47AC...
    this Product Change Notification document states that you can't order these models after September 26 !!! The official reason is that the market moved to other products ; politically correct way to say that nobody wants it !!!
  • harpocrates - Friday, September 5, 2014 - link

    Intel is releasing a new Core M stepping to fix the TSX bug:
    http://www.kitguru.net/components/cpu/anton-shilov...
  • MartinT - Friday, September 5, 2014 - link

    Are you sure the F stepping is supposed to fix that bug? Seems awfully early for a fix to be in production so soon after the public acknowledgement of the bug.
  • harpocrates - Friday, September 5, 2014 - link

    I'm not 100% sure but it's what been reported at the moment.

    From Techreport: "Intel has a fix in the works for Broadwell's next stepping. We don't yet know when Broadwell production will transition to the new stepping or how prevalent the TSX erratum will be among the first wave of Broadwell-based systems."

    http://techreport.com/news/26911/errata-prompts-in...
  • extide - Friday, September 5, 2014 - link

    No, it's because the 14nm process delays caused this launch to get delayed so far back that it's almost time for skylake to come out.
  • Krysto - Saturday, September 6, 2014 - link

    For real.
  • Shadowmaster625 - Friday, September 5, 2014 - link

    4.5W in a 7" device? OK Intel, now take a rockchip ARM SoC and one of your tiny modems and stick them in there and make me a dual boot device. Better yet, just add the modem and the ARM cores to the broadwell SoC. Get rid of some of those EUs because lets face it, you just cant use that many at that TDP. Imagine Android + windows all from one SoC. Intel already has everything they need to do it, except for the desire to make money, apparently.
  • REAVER117 - Friday, September 5, 2014 - link

    What benefit would you get by adding an ARM SoC to an x86/64 processor? Android already supports x86/64.
  • Krysto - Saturday, September 6, 2014 - link

    Native Android apps, can run up to 40 percent slower and use up to 90 percent more battery life on Intel chips. Also, only 20 percent of the top 100 Android apps even have support for Intel chips, which means it's probably more like 5-10 percent for all Android apps, if the top ones don't even have it:

    http://www.theregister.co.uk/Print/2014/05/02/arm_...
  • Kidster3001 - Friday, October 3, 2014 - link

    "Native Android Apps" is a relative term. Native for what ISA? Most Android apps that include a native binary include support for ARM, ARMv7a and MIPS. Many now include native support for X86. You are of course, referring to Intel's ability to translate ARM native binaries to X86 native code "on-the-fly". Try having the developer of your favorite app release it with an x86 binary (easy with Android NDK) and your story goes away. The article you are referring to is about an experiment performed by ARM Ltd. Of course they are not going to let X86 run native X86 binaries. They are forcing X86 to run ARM binaries. Intel Android devices are very much equal to ARM Android devices for performance AND power consumption when they are both running binaries compiled for their respective ISA's.

    Intel Android can run well over 99% of all apps on the market using this binary translation. Your % compatibility numbers are from the 2010-2011 era, not current reality.
  • Kidster3001 - Friday, October 3, 2014 - link

    sry, Intel can translate native armv7a binaries on the fly, not 'arm'.
  • fokka - Friday, September 5, 2014 - link

    there are already products combining windows and android, from asus most probably
  • pav1 - Friday, September 5, 2014 - link

    Just one or two additional hour(s) of battery life.. after all that trouble of a new process node. Is Broadwell M a game changer? Not at all. This evolutionary trajectory has been followed by Intel for years. Yawn.
  • OrphanageExplosion - Friday, September 5, 2014 - link

    The top-end chip looks very interesting, and the turbo on all parts is intriguing, but the data sort of makes me more interested in what's going to happen with the ULT parts at 15 and 28W.

    Personally I'd really like to see a lower-clocked (vs existing Haswell) 28W quad for the 13-inch rMBP, but I doubt that's going to happen.
  • fokka - Friday, September 5, 2014 - link

    how about a 25 watt quadcore with massive turbos? that's what i would like to see for a machine like the rmbp.
  • Krysto - Saturday, September 6, 2014 - link

    The Turbo is there to take your eyes off the fact that the baseline is 800 Mhz. But you'd be much better off with a Haswell Celeron at 1.6 Ghz than a Broadwell Y "Core M" at 800 Mhz that goes to 2 Ghz - performance wise.
  • Black Obsidian - Saturday, September 6, 2014 - link

    Can you expand on that? Either you're making unstated assumptions (about TDP, or perhaps thermal constraints) or you're mistaken. That you seem to feel baseline clockspeed has any real relation to actual performance seems to suggest the latter, but maybe you've got a valid argument that you just forgot to make.
  • tipoo - Friday, September 5, 2014 - link

    So, with no eDRAM (and the low tdp), is it likely the HD Graphics 5300 would perform under the 5200? Talk about confusing naming for laypeople.
  • JarredWalton - Friday, September 5, 2014 - link

    The existing Iris Pro 5200 is in a different brand (Iris vs. HD Graphics), plus we're comparing across power envelopes. Iris Pro 5200 is only available in 47W quad-core laptop CPUs or 65W desktop CPUs; these are 4.5W TDP parts, which is an entirely different market. The products they're replacing (e.g. i5-4300Y) were 11.5W TDP with HD Graphics 4200, and compared to those we should see an improvement in graphics performance thanks to the updated process, clocks, and architecture.
  • fokka - Friday, September 5, 2014 - link

    thanks for explaining, that's what i was thinking. another example of intels genius naming schemes.
  • rootheday3 - Friday, September 5, 2014 - link

    so.. NVidia had 650, 680, etc. Then they came out with 750... By your reasoning, since 7xx > 6xx, then 750 must be faster than 680 - right?

    And the mobile parts that are called 650M, 680M must have the same number of shader cores, etc as the non-M variants - right?

    Oh wait.... well AMD surely gets it right don't they?

    Nope...- Mullins/Beema and Low end Kaveri parts are 7xxx while Richland is 6xxxx. By your reasoning, Mullins/Beema should be faster than the fastest Richland
  • Drumsticks - Saturday, September 6, 2014 - link

    I don't think that was his logic. He wasn't comparing a 4xxx part to a 5xxx part, he compared a 5xxx part under one brand to a 5xxx part of another.

    A more apt comparison would be a GTX 720 With a GT 730. One is GTX but they "appear" from the same generation. An uninformed onlooker might not give any thought to the Iris Pro designation.
  • fteoath64 - Tuesday, September 9, 2014 - link

    "with no eDRAM (and the low tdp)". Did you NOT not see the separate package on chip itself ?. That's the eDRAM!. Manufactured by a Japanese company and integrated to the memory controller. The HD 5300 is greater than Iris Pro 5200. They are reducing the power consumption of the gpu cores due to 14nm process and even clocking it lower than before. All just to make the 4.5w TDP rating.
  • CaedenV - Friday, September 5, 2014 - link

    So 2 general questions/observations.
    1) Are we really still expecting Skylake in 2015? I mean these low power chips just launched and won't even really be available in products for some time yet. I was expecting to maybe hear about Skylake towards the end of 2015, with a product launch in mid 2016.
    2) A while back there was talk of Intel working with Creative Labs... could these audio improvements be part of that collaboration?
  • DanNeely - Friday, September 5, 2014 - link

    Intel's repeatedly said that the trouble they've had with double patterning (the same thing that's severely delayed everyone else's 20nm process) will only delay/truncate broadwell's market lifetime; because they weren't moving the skylake launch date.
  • CaedenV - Friday, September 5, 2014 - link

    Interesting, you would think that they would want to milk it a bit longer after having so much trouble. I mean it isn't like there is a whole lot of competition now that they are starting to beat ARM offerings. On the plus side Skylake holds some promise for desktop users to look forward to, so it not being delayed is some pretty good news.
  • JarredWalton - Friday, September 5, 2014 - link

    No idea on point 2, but I suspect Skylake will have a similar launch one year from now -- there will be a few parts shipping at the end of 2015, but many of the Skylake SKUs will come in 2016. That's just my guess, of course, so don't hold me to that. :-)
  • arkhamasylum87 - Friday, September 5, 2014 - link

    I am waiting for the Core-M vs Apple A8 comparison soon. I wonder how close are Apple to achieving Intel Core M performance in the mobile SoCs. This would have a huge impact. On that nide, do you see Apple ever using Core M for their Macs? It would be a performance regression if they adopt this for the Air lineup.
  • kyuu - Friday, September 5, 2014 - link

    Even if the A8 is competitive with Core M performance-wise (which I rather doubt), it wouldn't mean a damn thing. They won't ever be in anything but iPhones and iPads (and maybe a future iOS-based AppleTV product).

    Hell, on the iPad Air I use for work, web browsing is still an awful experience due to tabs continually being dropped and having to refresh when I switch between them. And I'm talking 2-5 tabs at once, not some absurd amount. So the nice SoC inside isn't really doing much when the lack of RAM and crappy OS design get in the way. I get a far better browsing experience on my Venue 8 Pro.
  • nathanddrews - Friday, September 5, 2014 - link

    On the topic of V8P, how do you like it? They seem to be around the $200 mark and I've been contemplating picking one up to play with. The VP11 is what I really want (i5 power, real SSD, large screen), but the reviews have been abysmal by comparison.
  • nathanddrews - Friday, September 5, 2014 - link

    V11P
  • Drumsticks - Friday, September 5, 2014 - link

    If you want a V11P, why not look into a SP3? With that configuration, they'll run similar in pricing anyways.
  • kyuu - Friday, September 5, 2014 - link

    The V11P is fine, but the trouble is that once you're configuring with an i5 and SSD, the pricing is up there in SP3 territory. Once you're up there, you might as well go SP3 (or SP2 for a bit less), as the SP is a superior device.

    As far as the V8P goes, it's fantastic given its price. It's quick and responsive for everyday web browsing and app usage, and you can play a surprisingly large catalog of older games on it well (with a bluetooth controller, of course). My only disappointment was that it can't quite handle Gamecube emulation (though PSP and below works great). Would've loved to play some Metroid Prime on the go, among other things. And the 8" size is far more portable than 10/11", but I'm actually looking to get a 10" tab in the near future. At $200, you really can't go wrong with it, though.
  • Krysto - Saturday, September 6, 2014 - link

    But they could be in a $700 iPad Pro type LAPTOP that has the SAME PERFORMANCE as a $1,000 Macbook Air. And because of the price, and the MUCH bigger popularity of iOS, those could end up selling much better than Air, which means Intel is screwed either way.
  • Klimax - Saturday, September 6, 2014 - link

    This is called pipe dream. Also strongly illogical. (IOS apps are nothing like Metro apps nor OSX apps) And you can forget anything like Intel's performance. You just don't clock chips way beyond their original designed range.

    Such product would be anything but disaster.
  • Speedfriend - Monday, September 8, 2014 - link

    "But they could be in a $700 iPad Pro type LAPTOP that has the SAME PERFORMANCE as a $1,000 Macbook Air."

    For that to happen, Apple has to increase CPU performance by 2.5x and graphics by 4x between the A7 and A8. If they manage to do that, they should just give up making consumer devices and become a cpu company, flogging the A8 for $200 a chip.
  • Kevin G - Monday, September 8, 2014 - link

    Apple has been doubling CPU performance every generation it seems. It'll be tough to do a doubling of performance between the A7 -> A8 but I wouldn't say impossible. There is room for a clock speed increase here to ~1.8 Ghz which would be a ~25% increase by itself. Architecturally I don't think they can go wider in a mobile part but they can increase utilization via SMT (aka Hyperthreading on the Core M). More cache in both L2 and L3 areas would also offer a small boost in performance. A LPDDR4 memory controller would also give a tangible speed increase due to additional memory bandwidth.
  • darkich - Sunday, September 7, 2014 - link

    That's a funny argument.
    I can open more than 10 tabs on my Galaxy Note 3 in Dolphin.
    When I open more than 5 in Chrome, it starts to refresh.
    So it depends on the software more than hardware.

    Besides, the fact is that A8 will be far better than A7 AND come with 2GB of faster RAM, AND the safari in iOS8 will be improved to its swift core.
  • darkich - Sunday, September 7, 2014 - link

    *open more than 10 tabs without refreshing, that is
  • Speedfriend - Monday, September 8, 2014 - link

    @kyuu How dare you criticise the iPad Air, everyone knows it is going to take over the world....

    I can't believe ow may people still believe that it could be a useable business device when it has no removeable storage and no ability to properly multitask. No thanks, since buying a Trasnformer T100, I hardly touch my iPad despite its signficantly better screen.
  • Wolfpup - Monday, September 8, 2014 - link

    Yep, 1000% agree. iOS is a terrible tablet OS and the iOS devices are all ludicrously RAM starved.

    It's so bad that it's gotten to the point where I often remote in to one of my Windows machines...even going through remote desktop STILL gives a better experience than using my iPad natively. I'm done with both iOS and Android for tablets after seeing how well real Windows works now in the same form factors. Real Windows on a small device is literally what I've been waiting for since the first PalmOS devices launched...and astonishingly it's really here, and only going to get better.
  • joannecdinkins - Friday, September 5, 2014 - link

    just as Larry answered I didnt even know that people able to get paid $6104 in a few weeks on the internet .
    go to this site>>>>> paygazette.ℭOM
  • pav1 - Friday, September 5, 2014 - link

    Just one or two additional hour(s) of battery life.. after all that trouble of a new process node. Replacing Audio with a DSP gave the biggest gains. Poor Intel (pun intended)
  • Black Obsidian - Saturday, September 6, 2014 - link

    An extra 1-2 hours of battery life from nothing more than a die shrink isn't a bad outcome at all. Intel needed that new process for Skylake's new architecture, so why NOT have the current arch benefit from it as well? That's half the point of their "tick tock" strategy.
  • nofumble62 - Friday, September 5, 2014 - link

    Just bought a Macbook Air for my girl entering college last week. Having this chip would have be nice. There is always a high demand for high performance and efficient processor.
  • Krysto - Saturday, September 6, 2014 - link

    Except this isn't a high performance chip. Did you miss the part where its baseline clock is 800 Mhz? This will most definitely be slower than the already relatively slow 1.3 Ghz Haswell in the current Macbook Air.
  • Klimax - Saturday, September 6, 2014 - link

    I wouldn't say jump before benchmarks are in. There are some changes to microarchitecture like faster division (radix-1024)
  • Kevin G - Monday, September 8, 2014 - link

    We've already seen what impact a faster divisor has in real world scenarios. The change in the division unit was one of the alternations between Conroe and Wolfdale cores. In normal workloads this had less than a 1% impact in performance as division and square root are rarely used.
  • Drumsticks - Saturday, September 6, 2014 - link

    You realize turbo is what matters for responsiveness? My 4670k has a base clock of 3.4 GHz or whatever, but it still idles at 800 MHz.

    The new broadwell parts turbo past the i5 in the macbook air for instance, which I believe only turbos to 1.9.
  • The Von Matrices - Saturday, September 6, 2014 - link

    "The top of the line processor will be called the Core M-5Y70, which is a bit of a mouthful but the name breaks down similarly to Intel’s main Core series. ‘5’ is similar to i5, giving us a dual-core processor with Hyper-Threading; ‘Y’ is for Broadwell-Y; and ‘70’ gives its position in the hardware stack"

    Are you sure about the "5" relating to "i5"? I thought that the use of "5" was just because Broadwell is the 5th generation of Core processors, similar to how Haswell parts begin in 4, Ivy Bridge in 3, and Sandy Bridge in 2.
  • M.Q.Leo - Saturday, September 6, 2014 - link

    It seems so good.
  • Krysto - Saturday, September 6, 2014 - link

    800 Mhz ? That's beyond pathetic. Surely it must be a joke from Intel. But I'm not laughing, so it's a pretty bad one.
  • Klimax - Saturday, September 6, 2014 - link

    I didn't know you already got samples to test... Otherwise it is just horribly misinformed opinion. (Jumping to the conclusions...)
  • Galatian - Saturday, September 6, 2014 - link

    You keep posting here how bad the Broadwell chips are...are you an AMD or ARM employee? You have no performance figures yet you claim it's the worse chip ever.
  • Kevin G - Monday, September 8, 2014 - link

    The performance can be emulated rather well today with hardware out on the market. Broadwell is supposed to be a straight shrink of Haswell so IPC is the same. Take an unlocked Haswell Core i7, disable 2 cores, reduce the base clock to 800 Mhz, memory speed to 1600 Mhz and set the max turbo to 2 Ghz.

    GPU configuration is what is going to be difficult since the number of EU's and their throughput have changed. Giving the test chip's GPU a 20% boost in clock speed to compensate would give an approximation of the 20 -> 24 EU difference. The Core M's GPU should still be faster but such a test configuration could be seen as the minimal performance form the Core M.

    There will be some performance differences as the IO platform is going to be different than what the Z97 offers due to power consumption concerns. How much of that translates into real world performance difference should be minimal in terms of CPU benchmarking. (Think the difference between Z87 to Z97.)

    I'd also recommend altering the cooling solution as well as core voltages to further emulate the mobile environment. With desktop cooling, I'd expect this example to run the CPU at 2 Ghz virtually all the time but I'd still benchmark it as a reference point alongside the default desktop config. Note the power draw difference between these two and then reduce cooling capacity accordingly and keep going down until the 2.0 Ghz speed is truly a turbo setting but 800 Mhz is not throttled at all.
  • Krysto - Saturday, September 6, 2014 - link

    I'd like to see a direct comparison between Haswell M and "Core M" Broadwell Y, that they're trying to push for mainstream. I bet Haswell M is 50-100 percent faster than "Core M".

    Whatever you do, Anandtech, do not - I repeat - DO NOT, compare it to Haswell Y, which is a chip so pathetic, no OEM wanted to use it this past generation. But that's what Intel will want you to do, so they can claim "performance improvements over the last generation".

    But that would be an extremely misleading thing to do, since the last generation we had Haswell U as mainstream (which was quite weak, too, compared to previous generations), not Haswell Y, and before that we had SNB and IVB M, not U.
  • OrphanageExplosion - Saturday, September 6, 2014 - link

    I agree - the comparison should be with Haswell-U, assuming that Broadwell-Y is indeed meant for two-in-ones like the Asus Transformer Chi T300 and a supposed replacement for the U chip in the MacBook Air.

    Haswell-Y was pretty awful, severely throttled - and Anand's SP3 Core i3 piece suggested that the battery savings were minimal there anyway on anything other than the most basic of workloads.
  • MrSpadge - Saturday, September 6, 2014 - link

    You state these would support TXT. Wasn't that canceled in the firsat batches of Broadwell due to a bug? At some point it will obviously get fixed.. but how do we know? Do these Core M all feature a newer stepping? I doubt it, since they're the first to market.
  • Alexvrb - Saturday, September 6, 2014 - link

    Am I the only one here wondering when a Tonga review is going to get published?
  • darkich - Saturday, September 6, 2014 - link

    You know what?
    Intel looks pretty much doomed.
    If it makes Atom as good as the upcoming ARM 16nm/14nm chips, it would kill this Core M because all three would end up comparable in performance per watt, with Core M being the most expensive by a huge margin.
    And Intel lives off of such high margin chips.
    Now, Intel can make Core M as frugal as ARM chips, but it can't make it nowhere near as cheap.
    So again it loses, in an increasingly changing paradigm of computing where Android and ARM will become good enough for 90% of the market.
  • Alexvrb - Saturday, September 6, 2014 - link

    They're not standing still, either. They'll continue to make competitive chips on the low-end, and high-end chips for high-dollar devices. By the time an ARM chip threatens these, they'll have something new. There's no way someone spending big money on a high-end device (especially Pro models) is going to want to settle. So having a competitive Atom doesn't kill off their premium line.

    Now granted, I'm talking about the foreseeable future, the near term. We don't know what will happen 10+ years down the road. MIPS could replace ARM. Who knows.
  • leomax999 - Sunday, September 7, 2014 - link

    SDP is gone,sounds good.
    Core is dangerously near arm power territory. 4.5w is close to the likes of K1,805 etc.
    Wonder what skylake will bring to the table.
  • darkich - Sunday, September 7, 2014 - link

    Actually SD 805 is a 1.5W TDP chip, and the most energy efficient one out of the Snapdragon 800 line.

    As for the Tegra, better compare this with the 20nm Tegra M1(Maxwell) that will probably reach the market just 2-4 months after the Core M.
  • leomax999 - Monday, September 8, 2014 - link

    I haven't seen a published TDP for Snapdragon lineup, So can't comment on that.
    I could argue skylake would follow 4-5 months after that.

    The point I was trying to make is, Intel has brought the power of their big 'core' down to few watts and close to ARM territory.
    We are yet to see what 14nm atom will bring, which will be followed by 3rd major uarch revision for atom by end of '15.
    I would think things aren't as bad for Intel as it sounds.
  • LetsGo - Monday, September 8, 2014 - link

    There not as bad?
    Its an unmitable disaster for Intel in the Tablet and Phone Space.

    Look at the picture of the SOC the GPU is now taking the lion share of the space this is where the other SOC vendors have an advantage other Intel, they only have to speed up their CPU 2x - 4x and Laptop devices will be able to use these SOCS with the added benefit of better graphics and battery life.

    As for code in the near future MOST applications will be cloud based with graphical generation done on the client, again playing to the strength of stronger GPU relative to CPU's.
  • leomax999 - Monday, September 8, 2014 - link

    I don't disagree that Intel have struggled. My comment was regarding the future.
    I don't see allocating more space for GPU as a bad thing. Speeding up CPU 2x - 4x is easily said than done.
    If it was easy we would have server's running with ARM SOC's. Speaking of which, As a fellow commentator pointed out, AMD's A1100 which has eight A57 cores has a TDP around 25W.
    So we'll have to see how things get played out.
  • Speedfriend - Monday, September 8, 2014 - link

    "Actually SD 805 is a 1.5W TDP chip, and the most energy efficient one out of the Snapdragon 800 line."

    Really, where do you get that info from, I saw Qualcomm saying that the 800 range targets 2.5W in a phone and 5W in a tablet?

    And the 8 core A57 server chip that AMD is building is going to be 25W, so it is not like the top end ARM designs have fantastic performance per watt.
  • mohsin1994 - Sunday, September 7, 2014 - link

    this chips will be the future of ours :)

    www.gadgetsalert.com
  • LemmingOverlord - Monday, September 8, 2014 - link

    Not sure if anyone's asked this, but... should we expect a "Xeon M" CPU for ultra low power servers (and to fight off ARM before it gains traction in the server business)?

    It sounds confusing to have an 5Y10a and a 5Y10 part, which seem identical, tho'...
  • fteoath64 - Tuesday, September 9, 2014 - link

    @LOverlord: " should we expect a "Xeon M" CPU for ultra low power servers". Yes, this is an area with predictable demand as power consumption (per 1U) gets factored as the key choice point. MIPS64 is coming into this space as Arm creeps up very slowly, showing still little interest in attacking the server space.
    Intel cleverly used BayTrial and Celeron-Haswell in the NAS space displacing the previous higher-end Arm chips. AMD will ramp up slowly and is not interesting to Intel.
    When Intel gets 10 cores or 20 cores of these 1Ghz server chips with TDP of 40w then the high-density server players will jump quickly!.
  • Altares13 - Tuesday, September 16, 2014 - link

    Hey AnandTech, can you please explain how come Apple's A8 SoC is supposedly a 2 billion transistors while Intel's Core M SoC is "only" 1.3 billion? Given the long pipeline of the X86 model and all, I really don't get it... The RAM maybe?
  • Pessimism - Thursday, September 25, 2014 - link

    I don't trust any of this stacked chip upon chip technology, almost every generation of laptop gpu using this has cooked and delaminated itself over time

Log in

Don't have an account? Sign up now