Tag Archives: powersaving

Asus x99 Deluxe standby mode not working properly

Update Jan. 2016
Try disabling these settings:

 

160110191334

 

Update 06.11.2014: When using Corsair Power Supplies:

Go to http://www.corsair.com/en/rm-series-rm1000-80-plus-gold-certified-power-supply and click “Haswell compatibility” under the FAQ & Support Tab.

Quote “According to Intel’s presentation at IDF, the new Haswell processors enter a sleep state called C7 that can drop processor power usage as low as 0.05A. Even if the sleeping CPU is the only load on the +12V rail, most power supplies can handle a load this low. The potential problem comes up when there is still a substantial load on the power supply’s non-primary rails (the +3.3V and +5V). If the load on these non-primary rails are above a certain threshold (which varies by PSU), the +12V can go out of spec (voltages greater than +12.6V). If the +12V is out of spec when the motherboard comes out of the sleep state, the PSU’s protection may prevent the PSU from running and will cause the power supply to “latch off”. This will require the user to cycle the power on their power supply using the power switch on the back of the unit.”

Also check out their forums (http://forum.corsair.com/v3/showthread.php?t=118456). Possible that you need a new PSU but at first you could try to disable C7 state (if possible) in your BIOS settings and see if it helps.

 


 

You may know or not that i bought a new system this year. However the Asus x99 Deluxe standby mode seems to confuse some of the components of this system or its the mainboard itself messing around with it.

 

Infinite loop

When i put my system into standby or suspend mode i can’t really wake it from there. When i try to wake it: the hardware responds, fans are starting to spin, harddrives etc, but the system is not really recovering from the standby mode, instead the system goes to Q-CODE bF and sits there until it starts to reboot automatically. Even a 4-sec-press-hold the power button doesn’t help to cold boot the system. The system sure goes down, but it does keep the system to do the bF loop. It keeps infinite (re)booting.

 

Q-CODE: [bF]

 

The first thing i did was CMOS reset but it is NOT necessary!

Instead: turn off your power supply wait a few seconds and turn it on again. Then the system should boot normally.

I already did play with some BIOS settings and there was something strange happening when i enabled ASPM Support in “Advanced->Onboard Device Config->ASPM Support” ….

 

Asus x99 Deluxe standby mode messed up?
Asus x99 Deluxe standby mode messed up?

When i enabled ASPM Support, i was able to put the system in suspend/standby mode and wake it with either keyboard or power button, right on the first attempt. This made me think i found a solution. It did work 3 times in a row. I then wanted to know if i can reprocude this by undoing the changes to see the bF code again and redoing the changes to see a working standby mode, and failed. Disabling and reenabling ASPM Support didn’t come up with a working suspend/standby mode.

 

 Asus Technical Support

I already wrote to Asus Technical Support and the first “conclusion” was, that i may have defect hardware with the aim at power supply.

 

Do you have same or similar problems? Different Q-CODE? Let me know!!

 

Update: Well sometimes its working and sometimes not. And when i try to record it with a camera, take a guess, its working…. to be continued

 

Home-Server™ with MSI J1900I and debian Linux

This is kind of a review but at the same time it's not. I'm not going to test audio or video right now nor will i benchmark anything down. The MSI J1900I has an onboard Intel® Celeron® Processor J1900 (2M Cache, up to 2.42 GHz, 4 Cores) Chip, which lacks instruction sets compared to other mainboards in this category, but comes with a good clockspeed and low energy consumption.

I decided to get me a new little server running at home. For fun and entertaining myself for sure.

Sorry, i flipped the wrapping of the board. Well here is the hardware list.

 

Components:

  • MSI J1900I Motherboard, Intel® Celeron® J1900 (2M Cache, up to 2.42 GHz), mini-ITX
  • Kingston SDD V300 120 GB
  • be quiet 300 Watts Power Supply
  • Xigmatek Case Nebula, mini-ITX
  • and of course some SO-DIMMs 4 GB in total (Teamgroup)

 

Once assembled i started to install debian with USB sticks and a lot of patience and problems and patience. Why debian? well i made this decision like 17 years ago. It's an old love.

An argument maybe would be that debian appears very clean to me. Some people maybe in turn would say the opposite. However...

Now i have a running system with cpufreqd (this changed), dynamic dns or dyndns, wordpress of course and the independency and freedom that comes with it.

 

Few words about the Xigmatek Nebula Case

I like it. You can take off 3 side panels and everything is easy to access. The bottom has a lot of tiny holes for good airflow. For installing a mini-itx board i can recommend a short screw driver and be patient when you install the power supply. Everything is new and sits tight. The hole case itselfs look pretty clean and "cool". I wish i had the money to get me a stack of them to create a wall. But my internet connection just couldn't make use of them.

If you buy one, be gentle! The panels sit very justified/flush. You may have to use your nails or some plastic stick to get the first panel off. Once you  have the first panel off, its easy. Additionally i bought some purple LED to light the undersurface.

 

 

 

Realtek you Schlitzohr (translate that please for me)

During installation you may  get in trouble with the network drivers. Possible that your kernel is shipped with Realtek 8169 support (grep for r816 in lsmod) and even that a driver is loaded your network is not working properly? Then open your brower and search for "Realtek 8168".

I ended up here and tried some mirrors for the UX drivers and as you can see i got it to work - otherwise you wouldn't see this website.

When compiled and installed using the autorun.sh there still might be a r8169.ko left. Rename it to something that doesn't end with *.ko and run:

update-initramfs -u -k $(uname -r)

otherwise your kernel may keep loading the "old" driver during boot.

root@bahuett:~# lsmod|grep -i r8
r8168                 248409  0

 You didn't even get to that point? Debian is not even booting? What device are you using to deploy debian love on your system?

 

CD/DVD/NetInst?

The painful truth about this system is, that even when you manage it to install the netinstall system or base system using USB sticks you end up with dead network drivers and an USB stick and no way to compile a new driver (when using 1 CD ISO Image only).

So i recommend using an DVD Image to make sure every package that could be useful is available. And please, check the Image you downloaded for corruption/integirity or the files you need are not there or not complete.

For "burning" my USB sticks on Windows i tested some programms and in the end i can recommend this one:

WinSetupFromUSB-1-4 - Click

 

Quiet but not cold?

The MSI J1900i comes with a passive heatsink and the Nebula case has a fan rotating at ~ 1000 rpm. You can't hear it thats what we want - but - under load the CPU can reach up to 60 ° C or 158° F. (Just keep in mind that this CPU comes with a turbo boost mode!) I dont like that but for now i leave it as it is. The first thing i did was to play with cpufreqd and start makeing powersaving modes for my little hamster.

governor "powersave" -- didn't work well for me

... because i want both, saving power when nothing to do and launch rockets when necessary.

At first i tried the governor (you will know what this is, when you start business with cpufreqd) "powersave". The result was that the CPU got throttled the way i wanted but didn't dethrottle the way i wanted. Or there was an "incredible" higher stress required to get this governor switch to the next higher frequency while in powersaving mode. Don't get me wrong, the CPU got clocked smoothly higher and higher but not to fullspeed.

And i had  5 Rules and 5 Profiles to cover the frequencies (9 steps) cpufreq can move in.

Steppings

2.0 GHz (100%) 1.91 GHz (95,5%) 1.83 GHz (91,5%) 1.74 GHz (87%) 1.66 GHz (83%) ->

1.58 GHz (79%) 1.49 GHz (74,5%) 1.41 GHz (70,5%) 1.33 GHz (66,5%)

 (and the turbo-boost frequency 2.4 GHz)

 

However some of the governor parameters in cpufreqd are not available for governor "powersave" thus i changed to governor "ondemand" and am happy with it. Name of the rules do not have to contain the name of the governor in it. And frequencies are allowed to overlap. Also its not necessary to provide exact frequencies. You can use % values instead in your config.

current active rule is being selected by cpufreqd according to it's rule parameters which scores most

root@bahuett:~# cpufreqd-get

Name (#1):      ondemand High
Active on CPU#: 0, 1, 2, 3                       (this line implies which profile is active which in turn gets selected by the rules you create)

Governor:       ondemand
Min freq:       1328000
Max freq:       1328000

Name (#2):      ondemand Med
Governor:       ondemand
Min freq:       1328000
Max freq:       1494000

Name (#3):      ondemand Low1
Governor:       ondemand
Min freq:       1411000
Max freq:       1660000

Name (#4):      ondemand Low2
Governor:       ondemand
Min freq:       1577000
Max freq:       1826000

Name (#5):      ondemand Low3
Governor:       ondemand
Min freq:       1826000
Max freq:       1993000

model name and its official clock speed and the actual speed of each core-#

root@bahuett:~# watch "grep Hz /proc/cpuinfo"

Every 2,0s: grep Hz /proc/cpuinfo             Fri Aug 15 21:08:17 2014

model name      : Intel(R) Celeron(R) CPU  J1900  @ 1.99GHz
cpu MHz         : 1328.000
model name      : Intel(R) Celeron(R) CPU  J1900  @ 1.99GHz
cpu MHz         : 1328.000
model name      : Intel(R) Celeron(R) CPU  J1900  @ 1.99GHz
cpu MHz         : 1328.000
model name      : Intel(R) Celeron(R) CPU  J1900  @ 1.99GHz
cpu MHz         : 1328.000

 

 

 

 Sensors / Temperatur

Unfortunately ACPI can't find the CPUs temperature or at least not from stock. Just the systems temp can be measured - but - lm-sensors is doing well AND you can use the values in cpufreqd!

lm-sensor output

root@bahuett:~# sensor
sacpitz-virtual-0
Adapter: Virtual device
temp1:        +26.8°C  (crit = +90.0°C)

coretemp-isa-0000
Adapter: ISA adapter
Core 0:       +46.0°C  (high = +105.0°C, crit = +105.0°C)
Core 1:       +46.0°C  (high = +105.0°C, crit = +105.0°C)
Core 2:       +48.0°C  (high = +105.0°C, crit = +105.0°C)
Core 3:       +48.0°C  (high = +105.0°C, crit = +105.0°C)

ACPI output

root@bahuett:~# acpi -V
No support for device type: power_supply
No support for device type: power_supply
Thermal 0: ok, 26.8 degrees C
Thermal 0: trip point 0 switches to mode critical at temperature 90.0 degrees C
Thermal 0: trip point 1 switches to mode hot at temperature 85.0 degrees C
Thermal 0: trip point 2 switches to mode passive at temperature 85.0 degrees C
Thermal 0: trip point 3 switches to mode active at temperature 50.0 degrees C
Cooling 0: Processor 0 of 10
Cooling 1: Processor 0 of 10
Cooling 2: Processor 0 of 10
Cooling 3: Processor 0 of 10
Cooling 4: Fan 0 of 1

root@bahuett:~# acpi -t
Thermal 0: ok, 26.8 degrees C

 

The negative side is, that cpufreqd and turbo boost mode seem not to work together. However. Loading the msr module

modprobe msr 

and an additional stresstest discover that cpufreqd is the boss, but when you turn cpufreqd off, the turbo boost is working fine!

You could get cpufreqd working with "turbo" when you adjust the max frequency setting. But in my case i can now stick to msr only. So cpufreqd would just be extra overhead.

 

For plain powersavings i would recommend cpufreqd.  In my case, msr-only is the better choice. Frequency-changes during short stress-tests:

root@bahuett:~# turbostat
 CPU   GHz    TSC
 avg   2.35   2.00
   0   2.10   2.00
   1   2.41   2.00
   2   1.39   2.00
   3   1.33   2.00
 CPU   GHz    TSC
 avg   2.33   2.00
   0   2.12   2.00
   1   2.41   2.00
   2   1.33   2.00
   3   1.33   2.00
 CPU   GHz    TSC
 avg   2.30   2.00
   0   2.29   2.00
   1   2.38   2.00
   2   1.33   2.00
   3   1.33   2.00
 CPU   GHz    TSC
 avg   2.34   2.00
   0   2.22   2.00
   1   2.41   2.00
   2   1.33   2.00
   3   1.33   2.00
 CPU   GHz    TSC
 avg   2.34   2.00
   0   2.06   2.00
   1   2.41   2.00
   2   1.33   2.00
   3   1.33   2.00
 CPU   GHz    TSC
 avg   2.34   2.00
   0   2.06   2.00
   1   2.41   2.00
   2   1.33   2.00
   3   1.33   2.00
 CPU   GHz    TSC
 avg   2.34   2.00
   0   2.06   2.00
   1   2.41   2.00
   2   1.33   2.00
   3   1.33   2.00
 CPU   GHz    TSC
 avg   2.34   2.00
   0   2.19   2.00
   1   2.41   2.00
   2   1.33   2.00
   3   1.33   2.00
 CPU   GHz    TSC
 avg   2.36   2.00
   0   2.16   2.00
   1   2.41   2.00
   2   1.42   2.00
   3   1.43   2.00