Surface Pro3 Benchmark issue(Only single-core can be used)in x86-r2

[Alfu s]

Hello

On cm-x86-14.1-r2 and  Android-x86 7.1-r2 run ALL application only can be used single core of CPU

Especially obvious in Benchmark applications e.g Geekbench4，HWBOT

Just like in the pcture....This is the case when running on multicore tests in HWBOT  and Geekbench4

And Run geekbench4  68%~100%，The progress is multi-core testing，in normal conditions all cores should be 100% Usage rate.

finally，Multicore Score is ？？？     Normal score is about 6000 Geekbench Browser SP3 with 4300U

Test devices:

surface pro3 with i5-4300U

msi GE60  with i7-4710HQ

But Previous versions of x86-r2 There is no such issue

Thanks

[Robert W]

Just tested on my Dell 5175 with Intel m3-6Y30 and I have same issue only core 0 is being used.

CPU has 2 real cores and 4 threads with hyperthreading but only 1 thread is used!

Using latest version of android x86 AOSP.

[Chih-Wei Huang]

2018-06-04 1:21 GMT+08:00 Alfu s <zgf1...@gmail.com>:
> Hello
>
> On cm-x86-14.1-r2 and Android-x86 7.1-r2 run ALL application only can be
> used single core of CPU
>
> Especially obvious in Benchmark applications e.g Geekbench4，HWBOT
>
>
> Just like in the pcture....This is the case when running on multicore tests
> in HWBOT and Geekbench4
>
>

> And Run geekbench4 68%~100%，The progress is multi-core testing，in normal
> conditions all cores should be 100% Usage rate.
>
>

> finally，Multicore Score is ？？？ Normal score is about 6000 Geekbench
> Browser SP3 with 4300U
>
>
>

> Test devices:
>
> surface pro3 with i5-4300U
>
> msi GE60 with i7-4710HQ
>
> But Previous versions of x86-r2 There is no such issue

Which previous version did you mean exactly?

I don't see any change could result in that.


--
Chih-Wei
Android-x86 project
http://www.android-x86.org

[Alfu s]

only r2 has this issue，The cm-x86-14.1-r2 and Android-x86 7.1-r2

All old versions no problem
x86 r1
cm-x86 r1
x86 rc1
cm-x86 rc1
...

[Patrick Wirth]

I have the same issue on my device - Acer Switch 3 (Intel Pentium N4200 Quadcore).
unfortunately i can only boot cm-14.1-x86_64-r2 on my device.
No aosp 7.1 and no older builds.
This behavior applies to many other linux distros for some reasons.
It gives a frozen cursor on the left top on the screen and nothing happens.

[youling 257]

That because cpuset，r2 kernel config enable cpuset.

    # this ensures that the cpusets are present and usable, but the device's

    # init.rc must actually set the correct cpus

    mkdir /dev/cpuset/foreground

    write /dev/cpuset/foreground/cpus 0

    write /dev/cpuset/foreground/mems 0

    mkdir /dev/cpuset/foreground/boost

    write /dev/cpuset/foreground/boost/cpus 0

    write /dev/cpuset/foreground/boost/mems 0

    mkdir /dev/cpuset/background

    write /dev/cpuset/background/cpus 0

    write /dev/cpuset/background/mems 0

[Patrick Wirth]

I found that entry in the file but what do i have to edit to get 4 cores useable?

[youling 257]

unpack pack ramdisk.img，edit init.rc

    # this ensures that the cpusets are present and usable, but the device's

    # init.rc must actually set the correct cpus

    mkdir /dev/cpuset/foreground

    write /dev/cpuset/foreground/cpus 0-3

    write /dev/cpuset/foreground/mems 0

    mkdir /dev/cpuset/foreground/boost

    write /dev/cpuset/foreground/boost/cpus 0-3

    write /dev/cpuset/foreground/boost/mems 0

    mkdir /dev/cpuset/background

    write /dev/cpuset/background/cpus 0-3

    write /dev/cpuset/background/mems 0

    # system-background is for system tasks that should only run on

    # little cores, not on bigs

    # to be used only by init, so don't change system-bg permissions

    mkdir /dev/cpuset/system-background

    write /dev/cpuset/system-background/cpus 0-3

    write /dev/cpuset/system-background/mems 0

    mkdir /dev/cpuset/top-app

    write /dev/cpuset/top-app/cpus 0-3

    write /dev/cpuset/top-app/mems 0

在 2018年6月10日星期日 UTC+8上午7:40:11，Patrick Wirth写道：

[Alfu s]

I have successfully edit new init.rc and unpack pack ramdisk.img，boot successfully but the issue  still have，benchmark only core 0 is being used.
在 2018年6月10日星期日 UTC+8上午9:51:50，youling 257写道：

[Alfu s]

I try delete Android-x86 7.1-r2 init.rc all of cpuset line.

find init.rc cpuset line in Android-x86 7.1-r1.iso and add to r2 init.rc then unpack pack ramdisk.img to boot system.

lucky it works so good :)

I guess r2‘s cpuset parameter have issues.

在 2018年6月10日星期日 UTC+8上午9:51:50，youling 257写道：

unpack pack ramdisk.img，edit init.rc

[Patrick Wirth]

how did you unpack and repack ramdisk.img?
I´m just trying the whole day to get it to work.
I followed this guide to do the job:

https://access.redhat.com/solutions/24029

After replacing the ramdisk.img with my modified version, android x86 doesn´t boot anymore...
Can you upload your image?
Maybe it works

[Alfu s]

i use ubuntu and run gunzip ramdisk.img, edit init.rc and cpio new img

or try this unpack image,it works on i5 4300u , 

i just test in CM-x86-14.1-r2. Google Drive New ramdisk.img

rm -r old ramdisk.img and cp new ramdisk.img or use the CM-x86-14.1-r2.iso replace ramdisk.img to reinstall android x86

 
在 2018年6月11日星期一 UTC+8上午1:10:44，Patrick Wirth写道：

[Patrick Wirth]

Your image works as expected! thank you!
Don´t know what was wrong with my image, maybe wrong compression or whatever.
My linux skills are very "noob" basics

[youling 257]

在 2018年6月10日星期日 UTC+8上午9:51:50，youling 257写道：

unpack pack ramdisk.img，edit init.rc

[Chih-Wei Huang]

Thank you for pointing it out.
I've applied a fix from Android Oreo.