Benchmarks¶

CPU Info¶

Architecture	x86_64
CPU op-mode(s)	32-bit, 64-bit
CPU	4
On-line CPU list	0-3
Thread(s) per core	1
Core(s) per socket	4
Socket(s)	1
Vendor ID	AuthenticAMD
CPU family	16
Model	4
Model name	AMD Phenom(tm) II X4 965 Processor
Stepping	3
CPU MHz	3400.000
BogoMIPS	6823.10
Virtualization	AMD-V
L1d cache	64K
L1i cache	64K
L2 cache	512K
L3 cache	6144K
VGA	GF114 [GeForce GTX 560 Ti] (rev a1)

Software¶

	Debian	Gentoo	Gentoo
Distr ver	Wheezy (7.0)	Rolling-release (2.2)	Rolling-release (2.2)
Kernel ver	3.2.46-1	3.10.1-gentoo	3.10.1-backbone
VGA ver	304.88	325.08	325.08
DE	Gnome 3.4	Awesome 3.5.1	Awesome 3.5.1

OS Specific Tests¶

The tests show results independence of OS/Gcc/CFLAGS/LDFLAGS/Libs and dependence on kernel patches only!

Unigine Heaven Benchmark 4.0¶

Render:	OpenGL
Mode:	1920x1080 fullscreen
Preset	Custom
Quality	High
Tesselation	Disabled

	Gentoo	Debian
Min FPS	7,6	7,5
FPS	34,1	33,4
Max FPS	64,6	64,3
Score	859	842

Gtk performance test¶

	Gentoo	Debian
GtkEntry	0,02	0,07
GtkComboBox	0,79	0,7
GtkComboBoxEntry	0,76	0,76
GtkSpinButton	0,06	0,9
GtkProgressBar	0,05	0,06
GtkToggleButton	0,06	0,3
GtkCheckButton	0,06	0,05
GtkRadioButton	0,12	0,08
GtkTextView — Add Text	0,14	0,12
GtkTextView — Scroll	0,08	0,15
GtkDrawingArea — Lines	0,28	0,36
GtkDrawingArea — Circles	0,27	0,36
GtkDrawingArea — Text	0,13	0,17
GtkDrawingArea — Pixbufs	0,04	0,05
Total time	2,86	3,62

Insignificant differences due to different nVidia drivers versions and enabled compositing on Debian.

SysBench¶

CPU

		Gentoo	Debian
total time, s		10,4996	10,4263
	total number of events = 10000
total time taken by event execution, s		10,4893	10,4252
	per-request statistics
min, ms		1,04	1,02
avg, ms		1,05	1,04
max, ms		4,48	4,65
approx. 95 percentile, ms		1,04	1,09
	Threads fairness:
	events (avg/sttdev): 10000.0000/0.00
exec time (avg/stddev), s		10,4893	10,4252

Mutex

	Gentoo	Debian
total time, s	0,0035	0,0126
	total number of events = 1
total time taken by event execution, s	0,0034	0,0122
	per-request statistics
min, ms	3,4	12,2
avg, ms	3,4	12,2
max, ms	3,4	12,2
approx. 95 percentile, ms	10000000.00	10000000.00
	Thread fairness
	events (avg/sttdev): 1.0000/0.00
exec time (avg/sttdev), s	0,0034	0,0122

On Gentoo Sources threefold acceleration observed when using mutexes. The reason is not determined. But this difference is not strongly correlates with the other tests results.

Threads

	Gentoo	Debian
total time, s	1,7616	2,2473
total number of events = 10000
total time taken by event execution, s	1,7513	2,246
per-request statistics:
min, ms	0,16	0,21
avg, ms	0,18	0,22
max, ms	0,8	0,94
approx. 95 percentile, ms	0,2	0,24
Threads fairness:
events (avg/sttdev): 1.0000/0.00
exec time (avg/stddev), s	1,7513	2,246

UnixBench (Main Bench)¶

Shows efficiency of the sources.
I recommend to use CFS (not BFS) if You have more than 1 CPU/Core!

1 CPU

	Debian	gentoo-sources	pf-sources	backbone-sources-3.10.1	3.11.4-backbone-r2
Dhrystone 2 using register variables	2628.9	2652.6	2660.9	2656.7	2657.8
Double-Precision Whetstone	728.9	767.9	757.1	764.1	764.9
Execl Throughput	236.1	282.8	274.1	582.7	621.2
File Copy 1024 bufsize 2000 maxblocks	2385.6	2735.4	2795.7	2779.7	2775.6
File Copy 256 bufsize 500 maxblocks	1710.4	1979.3	2055.2	2052.7	2042.8
File Copy 4096 bufsize 8000 maxblocks	2568.0	2847.4	2820.1	2818.0	2808.4
Pipe Throughput	1905.6	1979.0	2016.1	1922.1	2023.9
Pipe-based Context Switching	115.9	116.8	113.2	187.9	181.7
Process Creation	250.8	319.0	294.8	833.0	715.5
Shell Scripts (1 concurrent)	715.3	575.1	562.9	1882.7	2149.0
Shell Scripts (8 concurrent)	6137.4	4875.4	4812.4	4858.7	5479.5
System Call Overhead	2627.9	2704.2	2716.0	2540.8	2707.1
System Benchmarks Index Score	1096.8	1142.9	1132.0	1503.1	1533.2

4 CPU

	Debian	gentoo-sources	pf-sources	backbone-sources-3.10.1	3.11.4-backbone-r2
Dhrystone 2 using register variables	10332.9	10583.9	10518.7	10595.1	10527.2
Double-Precision Whetstone	2893.4	3070.0	3036.3	3065.7	3061.2
Execl Throughput	4020.7	3536.5	3394.3	3546.3	4652.7
File Copy 1024 bufsize 2000 maxblocks	1783.5	2253.7	2213.0	2301.0	2318.6
File Copy 256 bufsize 500 maxblocks	1265.9	1478.2	1475.8	1589.1	1546.8
File Copy 4096 bufsize 8000 maxblocks	2741.4	3251.5	3168.1	3453.9	3494.4
Pipe Throughput	7528.3	7812.0	7937.8	7625.5	7937.2
Pipe-based Context Switching	3829.0	3331.6	3064.1	3984.1	3189.0
Process Creation	4328.3	3792.4	3473.3	3424.4	2434.2
Shell Scripts (1 concurrent)	7343.1	5530.2	5383.9	5350.2	6060.3
Shell Scripts (8 concurrent)	6940.2	5749.5	5580.7	5463.8	6157.3
System Call Overhead	2253.6	2301.5	2386.7	2365.0	2608.8
System Benchmarks Index Score	3851.7	3797.4	3709.8	3845.7	3869.6