Quik test

**Bemk** · 11-27-2011

Hey people,

I'm doing my own OS, and I'm trying to figure out what the quickest memcpy is.

I've written a small test project you can run to help me finding out.

The instructions are quite simple.

1) get the test from here: https://github.com/bemk/memcpytest
2) run make and post a reply with the exact output
3) (if you are running 64-bits), run make CFLAGS=-m32 and post a reply with the exact output.

The output you should see is like the bit below. It's basically the time it takes to run the newly compiled binary, which itself is a test for one of the memcpy functions.

No. 5 is from the native library. Don't be surprised if that's the fastest one.

Code:

time -p ./1
real 0.24
user 0.24
sys 0.00
time -p ./2
real 0.96
user 0.95
sys 0.00
time -p ./3
real 0.21
user 0.20
sys 0.00
time -p ./4
real 0.13
user 0.12
sys 0.00
time -p ./5
real 0.05
user 0.04
sys 0.00

**atreyu** · 11-28-2011

crappy pc #1:

Code:

#
# uname -r
2.6.30
#
# cat /etc/redhat-release
Red Hat Enterprise Linux WS release 4 (Nahant Update 3)
#
# cat /proc/cpuinfo 
processor       : 0
vendor_id       : AuthenticAMD
cpu family      : 6
model           : 6
model name      : AMD Athlon(tm) Proswssor
stepping        : 2
cpu MHz         : 1462.866
cache size      : 256 KB
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 1
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mp mmxext 3dnowext 3dnow
bogomips        : 2925.73
clflush size    : 32
power management: ts
#
# gcc --version
gcc (GCC) 4.0.2 20051130 (Red Hat 4.0.2-14.EL4)
Copyright (C) 2005 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
#
# make
gcc  -o 1 memcpy1.c
In file included from memcpy1.c:4:
vars.h:2:22: warning: no newline at end of file
memcpy1.c:41:2: warning: no newline at end of file
time -p ./1
real 0.98
user 0.98
sys 0.00
gcc  -o 2 memcpy2.c
In file included from memcpy2.c:4:
vars.h:2:22: warning: no newline at end of file
memcpy2.c:29:2: warning: no newline at end of file
time -p ./2
real 2.30
user 2.30
sys 0.00
gcc  -o 3 memcpy3.c
In file included from memcpy3.c:4:
vars.h:2:22: warning: no newline at end of file
memcpy3.c:27:2: warning: no newline at end of file
time -p ./3
real 2.71
user 2.71
sys 0.00
gcc  -o 4 memcpy4.c
In file included from memcpy4.c:4:
vars.h:2:22: warning: no newline at end of file
memcpy4.c:38:2: warning: no newline at end of file
time -p ./4
real 2.35
user 2.35
sys 0.00
gcc  -o 5 memcpy5.c
In file included from memcpy5.c:5:
vars.h:2:22: warning: no newline at end of file
memcpy5.c:19:2: warning: no newline at end of file
time -p ./5
real 0.83
user 0.83
sys 0.00

slightly less crappy pc #2:

Code:

#
# uname -r
2.6.38.6-26.rc1.fc15.i686.PAE
#
# cat /etc/redhat-release
Fedora release 15 (Lovelock)
#
# cat /proc/cpuinfo
processor       : 0 (1 of 4)
vendor_id       : GenuineIntel
cpu family      : 6
model           : 28
model name      : Intel(R) Atom(TM) CPU D510   @ 1.66GHz
stepping        : 10
cpu MHz         : 1662.555
cache size      : 512 KB
physical id     : 0
siblings        : 4
core id         : 0
cpu cores       : 2
apicid          : 0
initial apicid  : 0
fdiv_bug        : no
hlt_bug         : no
f00f_bug        : no
coma_bug        : no
fpu             : yes
fpu_exception   : yes
cpuid level     : 10
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc arch_perfmon pebs bts aperfmperf pni dtes64 monitor ds_cpl tm2 ssse3 cx16 xtpr pdcm movbe lahf_lm dts
bogomips        : 3325.11
clflush size    : 64
cache_alignment : 64
address sizes   : 36 bits physical, 48 bits virtual
power management:
#
# gcc --version
gcc (GCC) 4.6.0 20110603 (Red Hat 4.6.0-10)
Copyright (C) 2011 Free Software Foundation, Inc.
This is free software; see the source for copying conditions.  There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
#
# make
time -p ./1
real 0.56
user 0.56
sys 0.00
gcc  -o 2 memcpy2.c
time -p ./2
real 2.02
user 2.01
sys 0.00
gcc  -o 3 memcpy3.c
time -p ./3
real 1.80
user 1.79
sys 0.00
gcc  -o 4 memcpy4.c
time -p ./4
real 1.80
user 1.79
sys 0.00
gcc  -o 5 memcpy5.c
time -p ./5
real 0.09
user 0.08
sys 0.00

**Bemk** · 11-28-2011

Thanks, this is actually a piece of useful result.

First of all, I noticed your compiler was complaining about newlines (fixed that in a new commit), and secondly, your results are quite different from mine. Already my reason for asking here seems to be valid.

As for the CPU specifications, I'm not really that interested. All I want to know is what your architecture is, 64-bits or 32-bits.

The rest can remain private.

***jayd512*** · 11-28-2011

32-bit Slackware

Code:

gcc  -o 1 memcpy1.c
time -p ./1
real 0.40
user 0.40
sys 0.00
gcc  -o 2 memcpy2.c
time -p ./2
real 1.24
user 1.22
sys 0.00
gcc  -o 3 memcpy3.c
time -p ./3
real 0.27
user 0.26
sys 0.00
gcc  -o 4 memcpy4.c
time -p ./4
real 0.18
user 0.17
sys 0.00
gcc  -o 5 memcpy5.c
time -p ./5
real 0.11
user 0.10
sys 0.00

**Bemk** · 11-28-2011

Thx, but I was more talking about the CPU bus width rather than what the OS supports, since it's not so much the OS that defines the impact on relative times. The CPU bus is the bottle neck here. The cache could exert some influence as well.

***jayd512*** · 11-28-2011

In that case, I'll just give you the whole spread

Code:

cat /proc/cpuinfo
processor	: 0
vendor_id	: GenuineIntel
cpu family	: 6
model		: 15
model name	: Intel(R) Pentium(R) Dual  CPU  T3400  @ 2.16GHz
stepping	: 13
cpu MHz		: 1000.000
cache size	: 1024 KB
physical id	: 0
siblings	: 2
core id		: 0
cpu cores	: 2
apicid		: 0
initial apicid	: 0
fdiv_bug	: no
hlt_bug		: no
f00f_bug	: no
coma_bug	: no
fpu		: yes
fpu_exception	: yes
cpuid level	: 10
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc arch_perfmon pebs bts aperfmperf pni dtes64 monitor ds_cpl est tm2 ssse3 cx16 xtpr pdcm lahf_lm dts
bogomips	: 4323.13
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual


processor	: 1
vendor_id	: GenuineIntel
cpu family	: 6
model		: 15
model name	: Intel(R) Pentium(R) Dual  CPU  T3400  @ 2.16GHz
stepping	: 13
cpu MHz		: 2166.000
cache size	: 1024 KB
physical id	: 0
siblings	: 2
core id		: 1
cpu cores	: 2
apicid		: 1
initial apicid	: 1
fdiv_bug	: no
hlt_bug		: no
f00f_bug	: no
coma_bug	: no
fpu		: yes
fpu_exception	: yes
cpuid level	: 10
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe nx lm constant_tsc arch_perfmon pebs bts aperfmperf pni dtes64 monitor ds_cpl est tm2 ssse3 cx16 xtpr pdcm lahf_lm dts
bogomips	: 4322.28
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual

**Bemk** · 11-28-2011

Thx, this makes things a lot more logical

**atreyu** · 11-28-2011

Originally Posted by Bemk

As for the CPU specifications, I'm not really that interested. All I want to know is what your architecture is, 64-bits or 32-bits.

That's why I showed /proc/cpuinfo, explained here.. Or do you mean kernel architecture? They're both 32-bit, though you could probably have inferred that.

**Bemk** · 11-28-2011

I basically meant, that only bus size/register size is enough

***elija*** · 11-28-2011

CPU: 64 bit running 64bit Debian

Code:

cat /proc/cpuinfo
processor	: 0
vendor_id	: GenuineIntel
cpu family	: 6
model		: 37
model name	: Intel(R) Core(TM) i5 CPU         650  @ 3.20GHz
stepping	: 5
cpu MHz		: 1200.000
cache size	: 4096 KB
physical id	: 0
siblings	: 4
core id		: 0
cpu cores	: 2
apicid		: 0
initial apicid	: 0
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 popcnt aes lahf_lm ida arat tpr_shadow vnmi flexpriority ept vpid
bogomips	: 6385.07
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 1
vendor_id	: GenuineIntel
cpu family	: 6
model		: 37
model name	: Intel(R) Core(TM) i5 CPU         650  @ 3.20GHz
stepping	: 5
cpu MHz		: 1200.000
cache size	: 4096 KB
physical id	: 0
siblings	: 4
core id		: 2
cpu cores	: 2
apicid		: 4
initial apicid	: 4
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 popcnt aes lahf_lm ida arat tpr_shadow vnmi flexpriority ept vpid
bogomips	: 6384.25
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 2
vendor_id	: GenuineIntel
cpu family	: 6
model		: 37
model name	: Intel(R) Core(TM) i5 CPU         650  @ 3.20GHz
stepping	: 5
cpu MHz		: 1200.000
cache size	: 4096 KB
physical id	: 0
siblings	: 4
core id		: 0
cpu cores	: 2
apicid		: 1
initial apicid	: 1
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 popcnt aes lahf_lm ida arat tpr_shadow vnmi flexpriority ept vpid
bogomips	: 6384.24
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

processor	: 3
vendor_id	: GenuineIntel
cpu family	: 6
model		: 37
model name	: Intel(R) Core(TM) i5 CPU         650  @ 3.20GHz
stepping	: 5
cpu MHz		: 1200.000
cache size	: 4096 KB
physical id	: 0
siblings	: 4
core id		: 2
cpu cores	: 2
apicid		: 5
initial apicid	: 5
fpu		: yes
fpu_exception	: yes
cpuid level	: 11
wp		: yes
flags		: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx rdtscp lm constant_tsc arch_perfmon pebs bts rep_good xtopology nonstop_tsc aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm sse4_1 sse4_2 popcnt aes lahf_lm ida arat tpr_shadow vnmi flexpriority ept vpid
bogomips	: 6384.23
clflush size	: 64
cache_alignment	: 64
address sizes	: 36 bits physical, 48 bits virtual
power management:

Results

Code:

make CFLAGS=-m32
time -p ./1
real 0.16
user 0.17
sys 0.00
time -p ./2
real 0.83
user 0.83
sys 0.00
time -p ./3
real 0.23
user 0.23
sys 0.00
time -p ./4
real 0.10
user 0.10
sys 0.00
time -p ./5
real 0.03
user 0.03
sys 0.00

Thread Tools

Display

Quik test

Posting Permissions