Actually I had it backwards. In terms of average power draw reduction between the WiFi and Video playback tests Shield shows a 0.583W difference while Surface 2 shows 0.769W. Though there is the slight wrench of the companion core apparently only being enabled in Surface 2 during video playback.What drives this conclusion that this result is a Windows property? If you're comparing against the Shield, don't forget the screen size difference.
Especially interesting that Baytrail is showing more performance despite having only a ~10% frequency advantage. But yeah, it's definitely interesting to get a better comparison point for Baytrail that basically just shows it being on par with Tegra 4 on power consumption in mostly-idle situations.What I find interesting is that the A15 everyone was saying is a power hog behaves similarly as a Bay Trail. That said Silvermont has more performance, though quite ironically it's being destroyed on the benchmark old Atom fanatics used to love: Sunspider :biggrin:
I wonder if Haswell video block is just poor or if Windows drivers are to be blamed.Edit: On the upside, at least the T100 with Baytrail is showing a battery life increase similar to Surface 2 (albeit not quite as good) going from the WiFi to Video battery life tests... unlike Surface 2 Pro which again shows a slight decrease.
I wouldn't call web surfing mostly-idle. But I agree we need some more tests than what we've seen at the moment.Especially interesting that Baytrail is showing more performance despite having only a ~10% frequency advantage. But yeah, it's definitely interesting to get a better comparison point for Baytrail that basically just shows it being on par with Tegra 4 on power consumption in mostly-idle situations.
Doubtful that it's windows considering that the macbook air actually saw a 1.37x decrease in battery life between the WiFi and Video playback tests in the tablet battery life suite. (Which actually is indicative of how much worse Windows is at handling the 'active idle' of the WiFi test.) I'm really not certain what exactly the issue with Haswell video decoding is, it may well be the simply fact that it's too tied into the GPU itself rather than being an entirely separate block? It definitely seems like some other logic block isn't able to turn off while doing video decode since it'd be difficult for Intel to make the decode block itself that inefficient.I wonder if Haswell video block is just poor or if Windows drivers are to be blamed.
Eh, it depends entirely upon the pace of the web browsing. Unfortunately the explanation of the Tablet Web Browsing test doesn't go into much detail... But we know that the normal mobile Light battery life test does the following:I wouldn't call web surfing mostly-idle. But I agree we need some more tests than what we've seen at the moment.
The light test hits four different websites every minute, pausing for nearly the entire time to simulate reading time. Flash is enabled and present on three of the sites.
When is 14nm atom comming?
When is 14nm atom comming?
What drives this conclusion that this result is a Windows property? If you're comparing against the Shield, don't forget the screen size difference.
What I find interesting is that the A15 everyone was saying is a power hog behaves similarly as a Bay Trail. That said Silvermont has more performance, though quite ironically it's being destroyed on the benchmark old Atom fanatics used to love: Sunspider :biggrin:
You can't imagine how disappointed I am, Sunspider is such an important and representative benchmark! My world comes tumbling down :biggrin:The Sunspider tests is misleading in the Surface 2 review because the A15 score on the Surface 2 is running on IE11 while the Baytrail scores were running on Chrome. You can go back on the T100 review and Z3770 preview to see this. We know that IE11 runs Sunspider the fastest and other sites have benched Baytrail running IE11 scoring in the mid to low 300 ms so sorry Atom still wins in Sunspider.
If browsing , wich will hardly use more than a core lightly ,
get 8.68h battery life we can assume without much risk
to be significantly off that video playing will hardly last
more than 5 hours or so , i m even sure that Anand has
the numbers but didnt publish them for some reason , heck ,
despite the wonderfull numbers doctored by Intel he didnt
seem to give as much care for what was announced
as the great advantage of this new chip and even relegated
this spzecification to negligible quantity benchmarks wises..
There s no smoke without fire , that s for sure , and all
evidences point to average battery life , far , very far
from what was ultra hyped.
Oops.
Primary difference I see is that they've calibrated the display to 100 cd/m^2 while Anandtech tests at 200 cd/m^2.
Wow didn't realize brightness made that big a difference.
It definitely does - in this case with the 31.5 Wh battery we can calculate that the extra 100 cd/m^2 requires about 0.5W. Amount of power required for a given brightness level can vary a lot between devices due to both the transmission factor of the LCD (light output by the screen divided by light output by the backlight) and screen size. Both result in battery life comparisons typically telling more about the specifics of the device rather than the platform it's built on. Which is fine considering that consumers buy the device, not the platform it's using, but it does mean that one product is not necessarily representative of all that use the same platform.
What would 100 represent on a brightness percentage for the t100? I usually keep my iPads 4 brightness on minimum a reference
according to anandtech, max for T100 was 228nits
iPad 4 (late 2012), max was 407 nits
so only half
Pretty old driver. Can you install the newest and try again?
http://downloadmirror.intel.com/23334/a08/
If this driver doesn't work tell me your device id from your Intel HD graphics.
I ask bc I usually run on minimum on ipad so if 100 cd /m2 is in that ballpark I'm a happy dude
Great, can't wait to see how Ubuntu 64-bit behaves on such a device, keep us informed :thumbsup:I just got my device... backing up the recovery, then nuking all the bloatware and dual-booting ubuntu+w8
I was told that screen resolution has an impact since backlight has to be increased with higher native resolutions due to higher density of LCD components.It definitely does - in this case with the 31.5 Wh battery we can calculate that the extra 100 cd/m^2 requires about 0.5W. Amount of power required for a given brightness level can vary a lot between devices due to both the transmission factor of the LCD (light output by the screen divided by light output by the backlight) and screen size.
Exactly, and that's why TDP, claimed W for SoC, etc. don't matter that much in the end.Both result in battery life comparisons typically telling more about the specifics of the device rather than the platform it's built on. Which is fine considering that consumers buy the device, not the platform it's using, but it does mean that one product is not necessarily representative of all that use the same platform.