My hobby benchmarks

From ThorstensHome
Revision as of 22:48, 26 November 2008 by WikiSysop (Talk)

Jump to: navigation, search

Contents

Performance

my QEMU benchmark

QEmu guest system openSUSE 10.2, command is

1:time (for i in $(seq 1 1 10000); do echo $i>>file;done)
2:time (for i in $(seq 1 1 10000); do echo $i>>/dev/null;done)
host command times without kqemu times with kqemu times native
XEON 1       17-7-2     
XEON 2       16-7-1
PIV  1       20-8-3              6-2-2            1-0-0

my easy benchmark

Goal: Get a rough impression of system power Command is

time (for i in $(seq 1 1 10000); do echo $i>>file;done)

Or if bash is not available, write a script that does this and call it using time

system     real time  user time  system time
Nokia N810      2.8s       0.8s         0.9s

KDE compilation benchmark

Computer command time   Env remark
1        1       2:39h  1
2        1       0:55h  1   HT on
2        1       0:59h  1   HT off
3        2       41:31m 2   
2        2       1:35h  2   HT on
4        3       16:24m 3   make -j2 
2        3       43:23  3 
5        2        2:55m 2   make -j32
6        4       15:36m 4   make -j8
6        5        1:13h 5   (virtualbox)
6        5       38:32m 6   make -j2

Computer
1 1.5 GHz P-IV, 256 MB RAM
2 3.0 GHz P-IV, 512 MB RAM
3 1.73GHz P-M, 512 MB RAM
4 2x1.83GHz P-M, 2 GB RAM
5 16x2.92GHz Xeon, 128 GB RAM
6 4x2.4GHz x64, 4 GB RAM

Environment
1: GCC 3
2: GCC 4, more and newer code
3: GCC 4, more and newer code, CMake, QT 4, KDEPIM 656855
4: GCC 4, KDEPIM Revision: 773158, Automake, QT 3
5: GCC 4, KDEPIM Revision: 854206, CMake, QT 4 on VirtualBox
6: GCC 4, KDEPIM Revision: 854206, CMake, QT 4 on VMWare VM, 2 vCPU
Command 
1 date >> date.dat && rm -rf bench && tar -xvf bench.tar && cd bench/kdepim/ && make -f Makefile.cvs && ./configure && make && make install && cd ../.. && date >> date.dat
2 date >> date.dat && rm -rf bench && tar -xvf bench.tar && cd bench/kdepim/ && make -f Makefile.cvs && ./configure && make -j4 && make install && cd ../.. && date >> date.dat
3 make clean; rm CMakeCache.txt; date >>date.txt && cmakekde && date >>date.txt
4 make clean && date>>date.dat && make -f Makefile.cvs && ./configure && unsermake -j8 && unsermake install && date >> date.dat
5 make clean; rm CMakeCache.txt; date >>date.txt && cmake . && make && make install && date >>date.txt

disk I/O

USB disk on scorpio, the lower USB card.

scorpio:/media/disk # bonnie -o_direct -s 1
Bonnie: Warning: You have 250MB RAM, but you test with only 1MB datasize!
Bonnie:          This might yield unrealistically good results,
Bonnie:          for reading and seeking and writing.
Bonnie 1.4: File './Bonnie.4608', size: 1048576, volumes: 1
Using O_DIRECT for block based I/O
Writing with putc()...         done:    116 kB/s   5.0 %CPU

IDE disk on scorpio

scorpio:~ # bonnie -o_direct -s 1
Bonnie: Warning: You have 250MB RAM, but you test with only 1MB datasize!
Bonnie:          This might yield unrealistically good results,
Bonnie:          for reading and seeking and writing.
Bonnie 1.4: File './Bonnie.5050', size: 1048576, volumes: 1
Using O_DIRECT for block based I/O 
Writing with putc()...         done:  16079 kB/s  81.7 %CPU

USB disk on duffman (Latitude D610)

duffman:~ # bonnie -o_direct -s 1
Bonnie: Warning: You have 503MB RAM, but you test with only 1MB datasize!
Bonnie:          This might yield unrealistically good results,
Bonnie:          for reading and seeking and writing.
Bonnie 1.4: File './Bonnie.9098', size: 1048576, volumes: 1
Using O_DIRECT for block based I/O
Writing with putc()...         done:   8440 kB/s  52.8 %CPU

active USB disk on chris (Latitude D620)

duffman:~ # bonnie -o_direct -s 1
Bonnie: Warning: You have 2026MB RAM, but you test with only 1MB datasize!
Bonnie:          This might yield unrealistically good results,
Bonnie:          for reading and seeking and writing.
Bonnie 1.4: File './Bonnie.5974', size: 1048576, volumes: 1
Using O_DIRECT for block based I/O
Writing with putc()...         done:   9081 kB/s  49.7 %CPU

internal disk in bort (Inspiron 530)

bort:~ # bonnie -s 1000
Bonnie 1.4: File './Bonnie.13566', size: 1048576000, volumes: 1
Writing with putc()...         done:  67207 kB/s  92.1 %CPU

Internal 1TB 32MB Samsung disk in (bort=tweedleburg)

tweedleburg:/mnt/sdb1 # bonnie -s 1000
Bonnie 1.4: File './Bonnie.5710', size: 1048576000, volumes: 1
Writing with putc()...         done:  67345 kB/s  92.9 %CPU

Internal disk in (bort=tweedleburg) being /dev/sda, 1TB 32M Cache Samsung /dev/sdb:

tweedleburg:~ # hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   7116 MB in  2.00 seconds = 3561.63 MB/sec
 Timing buffered disk reads:  212 MB in  3.03 seconds =  70.05 MB/sec
tweedleburg:~ # hdparm -tT /dev/sdb

/dev/sdb:
 Timing cached reads:   7328 MB in  2.00 seconds = 3668.03 MB/sec
 Timing buffered disk reads:  280 MB in  3.01 seconds =  93.03 MB/sec

graphic cards

terminal server scorpio: 55-70 FPS with 10 rows from glxgears. Connection between scorpio and bort: 100Mb. bort: 5179-5199 FPS with 10 rows from glxgears.

netWork

Testing a 100Mb connection between scorpio and bort.

Latency

Connection between scorpio and bort: 100Mb over two switches. 10 pings from bort to scorpio: between 0.257 ms and 0.267 ms. Connection between scorpio and bort:1000Mb direct, no switches. 10 pings from bort to scorpio: between 0.074 ms and 0.103 ms.

ThroughPut

dd if=/dev/zero | netcat scorpio 8000
4.3 MB/s

KDE memory consumption

KDE 4

linux-gqit:~ # free
             total       used       free     shared    buffers     cached
Mem:       4060060     799476    3260584          0      30328     442632
-/+ buffers/cache:     326516    3733544
Swap:            0          0          0

KDE 3 tweedleburg:~ # free

            total       used       free     shared    buffers     cached

Mem: 4053152 2152388 1900764 0 89252 1520860 -/+ buffers/cache: 542276 3510876 Swap: 0 0 0

power consumption

                                 zero load   high load
3 GHz P IV with 512MB RAM          145.7 W     217.7 W
2.4 GHz Quad Core with 4GB RAM      72.4 W     152.4 W
CRT Monitor 14"                          57.5 W
TFT Monitor 22"                          30.4 W