Windows 10 [2014 - 2017]

Status
Not open for further replies.
@Kaotik
thanks for the warning, but my home internet is down for almost 2 days now and its still around 20% last time i have an internet lol.
now its, last day for my monthly quota using mobile phone. I download Life is Strange instead of W10 lol.
 
any idea why my WINDOWS + P button on keyboard not working on WIn 10?

i also cant "windows + search" in win 10...

EDIT:
generally "keyboard search" is totally broken in windows 10 new menus. works fine on old menus.

EDIT:
the music app does not support "scroll wheel" to change volume. weird :/

EDIT:
Cortana cannot listen to me while classic speech recognition works

EDIT:
managed to get Crota working but... its sux at recognizing my voice command. i know my english is bad, but Xbox 360 recognize my voice command well.
 
Last edited:
That's why it's in preview. You're going to find all kinds of oddities and broken or half-finished features.
what im afraid is, the half baked feature will be like that for the final release.

for example, the odd search feature of Windows 8 was already there since preview/beta/whatever it was called. But fixed in Windows 8.1
 
what im afraid is, the half baked feature will be like that for the final release.

for example, the odd search feature of Windows 8 was already there since preview/beta/whatever it was called. But fixed in Windows 8.1

You probably shouldn't be using Windows10 right now if that's the angle you're coming from. You are able to participate in feedback if there are bugs and things you'd like to see improved.
 
Very cool, free Windows 10 for owners of Raspberry Pi 2...

http://techreport.com/news/27758/raspberry-pi-2-is-out-gets-free-windows-10

Not particularly interested in an ARM install of Windows 10, but for that price it might be fun messing around with it. Would make for a decent media streamer to TV while staying within the Windows ecosystem, however. And I'm assuming ARM version of Windows will get better support this time around since it's one Windows for multiple devices now with unified apps.

Regards,
SB
 
If I can end up with a Pi2 headend with W10 that can run Kodi and Steam Streaming then I'll be happy. The 100Mbit ethernet wouldn't be enough for Steam though unfortunately, and I have my doubts about it pushing h265 very well.
 
100Mbit is likely bigger than the decoding capability and GPU is the same one as first Raspberry Pi so you will do no H265 whatsoever.

This Windows 10 is very interesting though ostensibly for developers and I have to wonder what personality you're supposed to use for apps : Win32 or Metro? can Metro apps even access GPIO et al. hardware? or just some .NET.
I would find it funny if it just boots to a command prompt. It is for "IoT" i.e. networked embedded so if you want to run it for that intended purpose it should be able to run headless. Then you can develop command-line/server apps, and/or glue between them with some Powershell. Display output is there for public display and ATM types of use.
 
You could make a nice little Windows media box if someone writes a windows driver for the GPU. And 100mb should be plenty good for Steam streaming (RaspBmc does not have any problems playing 1080p 60 fps videos).
 
Playing or streaming 1080p60? I'm not referring to performance as I know Pi can play high bitrate videos. I'm mainly concerned with the high bandwidth requirements of quality Steam streaming. At worst probably stream at 720p I guess. And it's as much "Windows" as Windows RT is since it's on an ARM chip.
 
I'm not sure why you think steam streaming requires in excess of 100mbit. Have you ever used it before? We're talking ~20mbit, tops.
 
You know something that would be awesome?
Ability to set permanent Core Affinity profile.

I would really like to be able to set things like browsers to not use core 0 so that single-threaded games don't have their performance messed up.
I can do this via Task Manager but only manually each time I run the program, should be possible to make this permanent & automatic.
 
I can do this via Task Manager but only manually each time I run the program, should be possible to make this permanent & automatic.
I believe the reason for this is so that you don't screw with affinity, then forget you messed with it, and then suffer performance issues because other factors changed in your system (you changing affinity for some other piece of software as well perhaps, and both then clashing over the use of the same CPU cores, thrashing each others' caches, for example.)

Anyway, should not someone have made a launcher of some sort, to automatize these kind of changes? I would think there's more people than you with these desires. :)
 
I would really like to be able to set things like browsers to not use core 0 so that single-threaded games don't have their performance messed up.
The Windows process scheduler doesn't work the way you think it does; pegging things away from ACPI CPU 0 wouldn't "solve" anything.

The thread scheduler will rotate threads through all available cores, even if that thread is "single threaded." You're better off pinning the single threaded job to a single CPU than trying to evacuate the first enumerated ACPI CPU. You can see this for yourself by running a single threaded, highly intensive application on a multicore CPU with a bunch of other tasks (that are "less busy") running simultaneously. You'll see your 100% CPU peg float around between logical cores within task manager, if you have it set to show ALL cores versus the CPU average.

Part of this pathological behavior sometimes works against the best interests of your power management, specifically talking about core parking on "inactive" cores. As Windows stupidly revolves threads around your processor package, it will force a wake-up state on parked cores, which can result in stuttering on low-threaded CPU-intensive tasks.

If you search my post history for "core parking" on this forum, you'll find a larger thread that goes through quite a bit of this in larger detail.
Much later edit: The thread in question is here - https://forum.beyond3d.com/threads/...neys-on-our-pc-cpus.53809/page-3#post-1711294

it wasn't as detailed as I remembered ;) hehe...

Oh, and unrelated to the core parking issue...

I've had the Windows 10 preview build inside a Hyper-V guest for a few months now, just to tinker with it. I finally decided that I wanted to play with it on a physical box, so I dug through the random junk pile and dragged out my wife's old 2009 Dell Mini 10v -- an Atom N270 powered, 945GMSE chipset netbook with 2GB of ram and a 500GB Western Digital hybrid drive plugged into the onboard SATA1 port. Let's be blunt: this is a very slow box, even with the hybrid drive.

Equally blunt: Windows 10 performance on this box is nothing short of astonishing. It goes from power button to desktop in about 12 seconds, and is remarkably usable after it takes about 20 more seconds to "settle down". As you might expect, it's massively CPU bottlenecked, but outside of the boot process, you really can't tell.

It's markedly faster than Windows 7 Pro on this craptastic hardware. Color me impressed.
 
Last edited:
Part of this pathological behavior sometimes works against the best interests of your power management, specifically talking about core parking on "inactive" cores. As Windows stupidly revolves threads around your processor package, it will force a wake-up state on parked cores, which can result in stuttering on low-threaded CPU-intensive tasks.

Windows process scheduling has improved a lot since the Windows XP days. It now understands Intel Hyperthreading, it understands AMD module/cores and it to some extend understands the cost of migrating from one core to another (or from one socket to another)

It also bounces processes around a lot less than it used to. In the past it tried to match each core's run queue to the same length, which meant your single threaded job got round robin scheduled to all cores in the system (Linux did the same). Today a process can pin to a single core for a long time (until burst of system activity shakes it up)

Optimum process scheduling is NP-hard, which is why we see fast and simple algorithms used in just about all OSs, where fairness and robustness is valued over outright aggregate performance.

Cheers
 
Status
Not open for further replies.
Back
Top