Back to Article

  • jjj - Monday, September 12, 2011 - link

    april-may? slidong further 1 more month? Reply
  • tuklap - Monday, September 12, 2011 - link

    they are really going to match the next gen APUs at the ivy bridge release...price cuts!!! Reply
  • yankeeDDL - Monday, September 12, 2011 - link

    I think this is a neat feature, but I really don't see the big fuss about it.
    Basically, you take a 3GHz CPU with a TDP of 65W, underclock it and undervolt-it (if that's a word) and get a 35W configuration.
    When mobile you use the latter, when docked you use the first.

    What I find a bit tricky is how are you going to do, in practice, the extra cooling. Fans are not exactly power hogs, so keeping them at lower RPM, may spare you some noise, but power?
    On the contrary: docking usually limits (slightly, but it does limit) the cooling solutions, as it partially wraps around the mobile device.
  • quiksilvr - Monday, September 12, 2011 - link

    You're not factoring in the energy the CPU itself takes. Reply
  • yankeeDDL - Monday, September 12, 2011 - link

    Either I don't understand your comment, or you're wrong.
    I am accounting for the energy that the CPU itself takes: that is the TDP ...
    The system needs to have a cooling arrangement that enables it to dissipate the heat generated at the highest TDP. So this feature, does not bring advantages there.
    Setting lower TDPs, makes you save energy, and therefore increase battery life. But that is already covered by SpeedStep and similar techniques. I guess this gives you another knob, but I cannot imagine in which practical application this makes any difference: today I cal already set the max frequency of the CPU (and possibly lower the voltage too).
    The only example that this article brings is when a laptop is docked, the CPU speed can be cranked IF higher cooling is available. This is, I think, the only reasonable scenario where this makes sense. Now think about the cooling solution of a normal desktop CPU with 65W TDP, and imagine implementing something like that in a laptop. You can't.
  • Alexvrb - Tuesday, September 13, 2011 - link

    I think quiksilvr was trying to say that a laptop might have the cooling capability, in that it can ramp up fans even more. But they wouldn't do that when it is mobile (undocked) because it's not necessary because the CPU is running at a lower frequency/TDP combo to save power.

    So when you dock it, no more battery power concerns, ramp up TDP, ramp up clocks, ramp up fans. You could even use larger diameter (possibly TMD) fans that run at very low speeds when mobile, so that when you dock they don't need to scream.

    Alternatively the dock itself will utilize large fans and push air into the system similar to using forced induction on an engine. I don't think they'll get as elaborate as an intercooler, however. But they could if they darn well wanted to - they could even use a compressor and blow chilled air into the laptop intake(s). The dock itself could even be heat conductive, hollow and finned, and pull heat away from the entire laptop chassis.

    It all depends how elaborate the OEMs want to get. I think in the end only very high end boutique vendors will really make good use of this. Maybe determined modders too.
  • yankeeDDL - Tuesday, September 13, 2011 - link

    I think I understand your point: you say that the cooling solution does not need to change, in principle, but you cup the TDP when mobile to avoid draining your battery.
    With the current solution (SpeedStep), this is only up to the user, and in any case the CPU has no way to know whether it is docked or not, in order to enforce the TDP limitation.
    I can see a small added value here. Although, if you think about it, all OSes today are aware of whether a laptop is docked or not, so it would be easy to implement a TDP capping in SW. I would think so at least.
    In any case, thanks for your explanation.
  • mathew7 - Tuesday, September 13, 2011 - link

    Maybe this will open the market to some docking stations with small AC units, so that not only will air be forced into the laptop (fans), it will be also cooler than ambient.

    There are plenty of ideas.

    PS: I think it would be nice to have a 12"-13" laptop with good battery life which would still be usable as a gaming PC while docking with an external GPU (so the CPU could "upgrade" itself...of course this would require more than 10% TDP boost).
  • blueeyesm - Monday, September 12, 2011 - link

    There'd be a few ways. One would be an intake fan over the exhaust port of a laptop, for example, that force draws air into an airway and exhausted out the dock itself. Or, say two small posts that the laptop docks onto could be used (although somewhat inefficiently).

    Use your imagination and dream up a solution :)
  • iwodo - Monday, September 12, 2011 - link

    Sorry how is this different from SpeedStep? And Turbo Boost? Isn't it basically a "power usage" marketed version of the two above. Where as Speed Step and Turbo Boost were marketed as a performance oriented features? Reply
  • dagamer34 - Monday, September 12, 2011 - link

    At least on the high end of things, a processor that SpeedSteps can never go over it's configured TDP. The other thing is that I'm not sure the CPU is aware at what power state a laptop is in (either plugged in or battery powered). Reply
  • SlyNine1 - Monday, September 12, 2011 - link

    Dell seemed to manage it, look at the 1545 XPS studio's. They underclocked if it was using to much power for the 90watt PSU, and boy did it underclock and use clock modulation.

    I hope this isn't a way from OEMs to sell higher rated chips then the cooling system can handle.

    Yea I just bought a 3.0ghz chip, why is it only running at 1.9ghz.
  • silverblue - Monday, September 12, 2011 - link

    ...don't the server parts of Bulldozer have a configurable TDP as well? AMD would beat them to market by a good 6 months. The question is, whose implementation is superior? Intel's sounds more seamless whereas AMD's is far more granular but requires software support. Reply
  • KPOM - Monday, September 12, 2011 - link

    Did Intel let OEMs know this when they designed the Ultrabook spec? My concern here is that Ultrabook manufacturers are designing chassis right now to get products to market based on the existing TDP, and that while the chips might be ready in April or May, actual products taking advantage of the scalable TDP might take longer. Reply
  • retrospooty - Monday, September 12, 2011 - link

    All major OEM's are well informed of what is coming and have designs based on it. This is true for all releases. Reply
  • Le_Tobe - Monday, September 12, 2011 - link

    i can see an elevated docking station with external fans cooling down a laptop enough to disipate maybe 10w of additional heat, but i wouldn't trust it enough. especially if this is to make your laptop a desktop replacement. for all i know, i could be ignorant to the amount of heat that is contained in a laptop by placing it on your lap or some other insulating surface. Reply
  • Le_Tobe - Monday, September 12, 2011 - link

    just to clarify, i simply don't trust a whole lot of OEMs to make a product that can safely and fully take advantage of the variable TDP Reply
  • Iketh - Monday, September 12, 2011 - link

    yes... oem "engineers" make me facepalm all too often... how do those morons get in the positions they are in? Reply
  • gramboh - Monday, September 12, 2011 - link

    Do we know what the TDP of the desktop IVB parts will be? This articles seems to allude to 65w? I thought they would be 95w at the top end (2600k) like Sandy Bridge? Reply
  • xxtypersxx - Monday, September 12, 2011 - link

    I agree with the poster commenting on how this differs very little from speedstep. Basically they are taking a 33w chip and making the default speedstep profile 17w, essentially locking the chip into a underclocked/undervolted mode for normal use. However while it may not be anything ground breaking technically, I do think it is pretty exciting from a design possibility standpoint and I will certainly happily accept the added performance potential.

    What I'm not sure of is exactly how the OEM's are going to implement additional docked cooling beyond simply having the fans sping faster. You can put fans in the docking station but simply having them blowing into the stock inlet vents is only going to be a marginal improvement. Alternatively you could have excess metal or fins on the bottom of the laptop that are connected to the main heatsink with heatpips but this would likely become uncomfortably warm during mobile use as well. I am a bit worried that what we will see come out of this is laptops designed with internal cooling that can handle 33w but simply run very quietly when mobile and consuming 17w. While this is nice in its own regard, it would be a tremendous waste of the potential here.

    Where I think this becomes even more exciting is on Windows 8 tablets that dock like the Asus EEE Pad. If the chip can run at a low TDP without fans while in tablet mode but somehow make thermal contact with cooling devices in the laptop base when docked and clock up to a 13w or 17w tdp it would be fantastic.
  • jah1subs - Monday, September 12, 2011 - link

    Did Intel ever say officially whether or not they would be using the same sockets for Ivy Bridge that they are using for Sandy Bridge?

    If they said something, I have apparently forgotten :-)
  • faizoff - Monday, September 12, 2011 - link

    I think I remember that Ivy Bridge will have the 1155 socket that SB currently employs. I forget exactly what the socket 2011 was intended for. Reply
  • faizoff - Monday, September 12, 2011 - link

    Just remembered it's the SB-Extreme series. Reply
  • jah1subs - Monday, September 12, 2011 - link

    If the 1155 socket will support Ivy Bridge, it would seem that the P7, H7 series chipsets will basically be little more than P6 + Z68, H6 + Z68.

    Or, is there something else that I am missing?
  • DanNeely - Monday, September 12, 2011 - link

    Just speculating, but perhaps integrated USB3, more sata 6GB, or PCI3.0 on the southbridge and a PCIe3.0 derived DMI? Reply
  • DanNeely - Monday, September 12, 2011 - link

    Leaked slides from earlier this year indicate 4x USB3, and support for triple display configurations. Sata 6GB support will remain at 2 ports. The consumer oriented chipsets will drop legacy PCI support; although I suspect that like with PATA most OEMs will add it back via bridge chips: Unless they drop 3rd party audio/nic parts for Intel's more expensive versions they'll have a hard time getting enough PCIe lanes free to fill all 7 spaces on a full ATX board.

Log in

Don't have an account? Sign up now