Linux – How to mearsure the average cpu frequency in linux

central-processing-unitelectrical-powerfrequencylinux

I'm trying to measure the average cpu frequency in an DVFS enabled cpu for specific
interval , the obvious way of periodically sampling /proc/cpuinfo has
very large variants. The cpufreq-stats driver gave me some hope ,e.g.

cat /sys/devices/system/cpu/cpu0/cpufreq/stats/time_in_state 
2600000 118148
2000000 8562
1600000 11041
1200000 3428602    # lots of ticks is idle ticks

but unfortunately it is mixed with idle ticks(see patch "Do not account for idle time when tracking
time_in_state"). During idle time it is probable that cpu will fall back to the lowest frequency ,thus skew the frequency distribution significantly .

The patch above utilize the account_idle_tick function to remove the idle
ticks from statics, but it seems only works for xen kernel. Is there alternative way to measure average cpu frequency?

Best Answer

The new development version Sysstat 9.1.6 includes a new option (-m FREQ) to report average cpu frequency ,e.g.

$ sar -m FREQ -P ALL 0
Linux 2.6.30.10-105.2.23.fc11.i686.PAE (palmer.localdomain)     10/23/2010     _i686_        (2 CPU)

02:36:09 PM     CPU   wghMHz
02:36:09 PM     all   1042.23
02:36:09 PM       0   1039.43
02:36:09 PM       1   1166.65

The average weight is the time spent in that frequency , for example in a 10 second reporting interval , first 8 seconds the frequency is in 1GHz, the last 2 seconds is in 2GHz, then the average frequency 1.2 GHz

Related Solutions

Linux Bash – How to Sort du -h Output by Size

As of GNU coreutils 7.5 released in August 2009, sort allows a -h parameter, which allows numeric suffixes of the kind produced by du -h:

du -hs * | sort -h

If you are using a sort that does not support -h, you can install GNU Coreutils. E.g. on an older Mac OS X:

brew install coreutils
du -hs * | gsort -h

From sort manual:

-h, --human-numeric-sort compare human readable numbers (e.g., 2K 1G)

The average lifespan of a CPU

Usage of a CPU does result in wear at the atomic/electronic level.

The actual lifespan of the silicon transistors of a consumer CPU is typically in the range of 20-30 years before there is a failure, not 3-4years. It is asssumed by then that the item would be obsolete.

How does Intel/AMD know if there's no way to test a chip for 30years? It is tested under load under higher than normal conditions (heat voltage, clock) and the variables and failure data are then extrapolated and calculated backwards for typical use.

Of course there are other failure points such as the chip packaging wires and things of that nature, but low decades would be a fair assumption.

3-4 years is more of the practical obsolescence of the product due to Moore's law and all that.

Source: Masters degree in EE, and learned in a course on chip failure design.

Best Answer

Related Solutions

Linux Bash – How to Sort du -h Output by Size

The average lifespan of a CPU

Related Topic