London event

Discussion in 'Microsoft' started by Kumabjorn, Sep 6, 2017.

Thread Status:
Not open for further replies.
  1. Steve S

    Steve S Pen Pro - Senior Member Super Moderator

    Likes Received:
    Trophy Points:
    << how in God's green earth is QC getting that with Windows?..>>

    ...Because they have it plugged into the wall...???
    Bronsky and dstrauss like this.
  2. sonichedgehog360

    sonichedgehog360 AKA Hifihedgehog Senior Member

    Likes Received:
    Trophy Points:
    Apple never claimed 29 hours; their official claim is only up to 10. I can easily see a Snapdragon 835 getting 20 hours if it were paired with a laptop size battery. In addtion, Apple A series processors are power hogs compared to most other ARM SoCs. So getting 20 hours battery life should be a piece of cake.
  3. dstrauss

    dstrauss Comic Relief Senior Member

    Likes Received:
    Trophy Points:
    Will believe it when I see it (and keeping fingers crossed it is true)...
  4. Step666

    Step666 Pen Pal - Newbie

    Likes Received:
    Trophy Points:
    So, bit of background, I used to work for Samsung has part of their retail set-up in the UK.

    When they launched the Tab S2, the internal training material made a huge thing about how the Tab S2 offered the same battery life as the iPad of the day (the Air 2, I think but I'm not sure) despite being nearly 10% thinner.
    The thing they seemed oblivious to was that the iPad had a much larger battery, something like 25% higher capacity.

    This leads us to two conclusions.
    Firstly, Samsung are idiots - for the sake of less than 1mm, they should've given the Tab S2 the same capacity battery as the iPad and blessed it with better battery life.
    Secondly, on the face of things, Apple's SoCs aren't anything special in terms of power efficiency compared with the likes of Samsung or Qualcomm (there were variants of the Tab S2 that utilised both companies' chipsets with, apparently, no significant variation in battery performance).

    As for the potential battery life of a Qualcomm equipped Windows tablet, given that most flagship smartphones using the Snapdragon 835 have batteries of around 3000mAh (some are larger but most are around that) and can run for around a day on a single charge with data connections switched on etc, if you were to put it in a tablet with a similar capacity battery to the Surface Pro (~6000mAh) then it doesn't seem particularly far-fetched that it could last around 24 hours even with a larger screen to power.
    The big.LITTLE architecture and HMP aspects of ARM SoCs mean that they can be very power-efficient when there's little-to-no-load.

    I for one though cannot help wondering whether the first round of WinARM tablets will be worth getting given that the Snapdragon 845 is rumoured to be unveiled before the end of the year.
    It should be both more powerful and more power-efficient and due to hit the market not long after WinARM, so it would seem sensible to hold off for a second wave of devices.
    Marty likes this.
  5. burningorange

    burningorange Pen Pal - Newbie

    Likes Received:
    Trophy Points:
    In case you are interested, here's a tip for estimating battery life:

    battery life [h] = battery size [Wh] / the power consumption of the device [W].

    The hard part is estimating the power consumption of the device, and since it is in the denominator, it has a strong influence on battery life. The simplest is just to assume the power consumption is the Thermal Design Power (TDP) or Scenario Design Power (SDP): for example, the Surface Pro has a 42 Wh battery, and a TDP of 7W for the Core i7-U series, which gives a 42/7 = 6h battery life; If you replace the chip with a Snapdragon 835 with TDP of 2W, then you get 21h battery life. For running Netflix only, I can see it use less, and 29h life just means the system consumes a reasonable 1.5W.

    Now, TDP is just the maximum wattage of heat that must be exhausted from the system to maintain integrity. Depending on how efficient your chip is, it consumes more than that. For example, at 50% efficiency, the system consumes 14W and exhausts 7W as heat, with 7W useful for work (running that all important Excel spreadsheet while rendering that even more important TabletPCReview webpage). But if it consumes 14W, the Surface Pro would only last 42/14=3h. That is the typical runtime when playing a game at full tilt and full brightness. Hence the TDP ! The reason I typically use the TDP is that it is a number that gives me a good estimate for a "typical use" of a given system and a "maximum use" of a system: if I don't often use the computer under full load (i.e. don't exhaut 7W of heat), so assuming 50% efficiency, 3.5W is exhausted as heat and 3.5W is consumed as something useful, e.g. CPU takes 1W, display takes 1W, other parts take 1W, for a total of 6h battery life. And at full load, I exhaust up to 7W of heat and consume 7W, for 3h battery life.

    Clearly, you need to find out how much your system consumes for the kind of workload you want. You need to include display power consumption, SSD power consumption, radio antenna power consumption, etc. It is hard to get a detailed measurement of all this, but it is easy to get total power consumption, you can use the inbuilt Windows Performance Monitor on your own computer to follow the battery discharge rate under various scenarios. It is very useful, and on my Surface Pro or Dell XPS 9365, I typically get discharge rates of 6-7W under normal workload. Turning screen brightness up/down, antennas on/off, etc, changes the battery discharge rate that is easy to follow. The difference in power consumption when things are turned on/off then gives you the power consumption of that component ! Easy. For the screen, you can then divide by the number of pixels (so you get W/pixel) and then go to another screen with different numbers of pixels to estimate the power consumption of that screen... etc.

    But for a quick rough estimate, the published TDP or SDP of the chip is good enough.

    Next, a note on the numerator: many batteries are not listed in Wh (which would be the proper unit of energy, Joules) but in Ah which is a unit of charge (Coulombs). It is mainly because batteries store a certain amount of charge, not energy, but a device converts the charge into energy by making the charged particles move around in electronic circuitry using various voltages. So you have to multiply by the voltage of the system using the battery. I typically take the CPU voltage.

    You can also estimate typical power consumption of various chips if we assume all screens are approximately the same) just by looking up the battery life reported by people and knowing the battery size. Then if that chip is used in another system, you can get the battery life.

    Anyway, all this to say that I'm not surprised that if they make a laptop/2-in-1 Surface Pro with a Snapdragon, that they would get all day/multi-day battery life. I'm more interested in their rumoured folding penabled full fat Windows 10 on arm folio device.
    Azzart likes this.
Similar Threads - London event
  1. Steve S
Thread Status:
Not open for further replies.

Share This Page