DATE | WHEN | WHERE |
2009-10-22 | From 12:00 to 21:00 UTC (8am -> 5pm EDT) | #fedora-test-day (webirc) |
What to test
We'd like to check power management tuning available for Fedora-12.
If you come to this page after the test day is completed, your testing is still valuable. You can check whether you save some energy after applying tuning.
Who's available
The following cast of characters will be available testing, workarounds, bug fixes, general discussion ...
- Development - Phil Knirsch, Marcela Maslanova (marcela on irc), Petr Lautrbach, Jiri Skala (skalnik on irc)
- Quality Assurance - Jan Scotka
Prerequisite
- An installed Rawhide (development version of Fedora) with graphical environment. Alternatively, you may install Fedora 12 Beta and update your system.
- SELinux isn't needed, users don't have to enable it.
- Internet connection
- The special-purpose Test Day RPM package which includes necessary dependencies and provides all testing scripts.
- Download the package for your architecture: (i686, x86_64, ppc, ppc64)
- Download the package public key and import into RPM:
su -c "rpm --import testday-20091022.asc"
- Lastly, install the package using:
yum --enablerepo=rawhide-debuginfo localinstall testday-pm-0.6-1.fc12.YOURARCH.rpm
- Kernel debugging info:
- Install the required package:
debuginfo-install kernel
- Alternatively, you can install the debuginfo kernel using
yum
:yum --enablerepo=rawhide-debuginfo install kernel-debuginfo
- NOTE - be sure that the kernel-debuginfo version matches your kernel version. You can confirm by running the command
rpm -q kernel-debuginfo-$(uname -r)
- Install the required package:
- Upload your hardware to Smolt according to these instructions.
Running the tests
Update your machine
It will be needed rawhide machine. Howto install or update Rawhide.
Tests
The test cases are split into test groups. Some of the tests are a bit time-consuming. Please perform as many tests as possible. If you aren't able to run them all, please remember that the tests in each table are sorted by importance in descending order.
You don't have to download any special scripts for each test. Our Testday package provides following commands:
testday-run-testname
(performing tests)testday-pack-results
(packing results of all tests in .tag.gz file)
There is a note about permission in the description of each test. Please follow this information.
X Window System
These tests should be performed on all systems in X11 environment.
Testcase | Description | ETA |
---|---|---|
personal | Typical use of computer (do whatever you are doing usually). CPU states and frequencies. | 10 minutes |
scomes | Ticks, syscall count, read/written bytes when performing some specific tasks. | 1 minute |
Single mode
These tests should be performed in single mode.
Testcase | Description | ETA |
---|---|---|
tuned | CPU states and frequencies when tuning daemon enabled | 7 minutes |
basic | Stress situations (BZIP2 compression, disk copying, vsftpd test) | 2 minutes |
Laptops
Following tests are intended only for laptops. If you don't have a laptop, don't perform these tests because they are based on capacity of a battery.
Testcase | Description | ETA |
---|---|---|
bltk | Power consumption during three simulated workloads (idle, OpenOffice, reader). | 15 minutes |
tunedprofiles | Tuned profiles effect on power consumption. | 5-20 minutes |
alpm | A simple test to determine effect of SATA link power management settings on power consumption. | 10 minutes |
Results
- Send your Smolt profile if you haven't done it already (see Prerequsite).
- Create compressed file with results by running:
testday-pack-results
- Upload created file to this wiki.
- Add your results into the table below.
User | Smolt Profile | Results file | ALPM result | Comments&problems | |
---|---|---|---|---|---|
User:ExampleUser | HW profile | results | |||
guaneryu | HW profile | results | |||
User:czhang | HW profile | results | |||
User:hdong | HW profile | results | |||
User:jbao | HW profile | results | |||
User:Liam | HW profile | results | |||
User:Jvcelak | HW profile | results | |||
User:ykopkova | HW profile | results | |||
User:jscotka | HW profile | results | max: 37.2W, medium: 36.3W, min: 36.6W | It seems that changing powerstates was more less without effects | |
jkoten | HW profile | results | max: 16.1W, medium: 16.0W, min: 15.6W | ||
mmaslano | HW profile | results | |||
alich | HW profile | results | |||
User:tpelka | HW profile | results | |||
urban | HW profile | results | |||
User:rrakus | HW profile | results | No SATA on machine | Well, stap and battery things doesn't work on ppc (iBook G4) | |
Michal Nowak | HW profile | results | |||
Phil Knirsch | HW profile | results | |||
James Laska | HW profile | results | No such files - /sys/class/scsi_host/host*/link_power_management_policy
|
||
Geert Janssens | HW profile | results | No such files - /sys/class/scsi_host/host*/link_power_management_policy
|
||
Mike Gahagan | HW profile | Results | min_pwr:21.2 max_perf: 21.9 med_pwr: 21.2 | noticed some inconsistancy in pwr readings, with 1019 USB image, usage is about 2.5W less.. HD likely spun down |