Home » Mailing list archives » linux-kernel » 2008 » March » 25

Re: performance differences: "maxcpus=1" vs. "echo 0 > /sys/devices/system/cpu/cpu1/online"

Previous thread: Re: What to do about the 2TB limit on HDIO_GETGEO ? by Mark Lord on Tuesday, March 25, 2008 - 9:34 am. (8 messages)

Next thread: RE: Performance changes between 2.6.13 and 2.6.23 by Sanders, Rob M. on Tuesday, March 25, 2008 - 9:52 am. (2 messages)
[view original message]
From: Michael Meyer <mike65134@...>
To: <linux-kernel@...>
Subject: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Tuesday, March 25, 2008 - 9:47 am

Hi, 


what is the difference between booting a dual core

machine with "maxcpus=1" or by deactivating the second

core at run time with "echo 0 >

/sys/devices/system/cpu/cpu1/online"?


I observed a funny behaviour of apache ant: although

it uses javac which is single threaded, a compile run

with "maxcpus=1" is actually faster than a compile run

with both cores activated. But with the second core

deactivated using "echo 0 >

/sys/devices/system/cpu/cpu1/online" it is even slower

than with both cores.


Is here any method to get the exact same behaviour of

"maxcpus=1" with disabling the second core only

temporarily? So that the second core could be disabled

before the ant execution and enabled after the ant

execution?


Thanks a lot!


      E-Mails jetzt auf Ihrem Handy.

www.yahoo.de/go


--
[view original message]
From: Bernd Eckenfels <ecki@...>
To: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Saturday, March 29, 2008 - 7:22 pm

I dont have much insight, but I do remember some java performance problems

on Multi-Core with some Dell Desktops. I think they had been fixed with a

Bios upgrade. Sorry I dont remeber the chip and revision details anymore. I

just felt reminded, because MAXCPU=1 helped back then, also.


Gruss

Bernd

y

--
[view original message]
From: Pavel Machek <pavel@...>
To: Michael Meyer <mike65134@...>
Cc: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Friday, March 28, 2008 - 9:59 am

maxcpus=1 : core stays powered off


Patch pushing 0 > online cores into C4 would be nice.

--

(english) http://www.livejournal.com/~pavelmachek

(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

--
[view original message]
From: Michael Meyer <mike65134@...>
To: Pavel Machek <pavel@...>
Cc: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Saturday, March 29, 2008 - 4:55 pm

I have an extremely well cooled case and a Thermaltake Typhoon sitting

on the E6600 (Intel TDP 65 Watt). At a room temperature of 20°C both

cores idle at 21°C and after one hour of two instances of prime95

running they are both at 28-33°C. As the E6600 is specified until 60°C


      E-Mails jetzt auf Ihrem Handy.

www.yahoo.de/go


--
[view original message]
From: Pavel Machek <pavel@...>
To: Michael Meyer <mike65134@...>
Cc: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Saturday, March 29, 2008 - 5:22 pm

Ok, and your chip is too old to have "Enhanced Dynamic Acceleration

Technology" (Intel, can't you use some reasonable names?!)... both

would explain effects you see, and it is neither...


You could try implementing deep sleep (C4) state for cpu hotplug....

									Pavel

--

(english) http://www.livejournal.com/~pavelmachek

(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

--
[view original message]
From: Michael Meyer <mike65134@...>
To: Pavel Machek <pavel@...>
Cc: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Saturday, March 29, 2008 - 6:16 pm

It is too old to have that. I bought it at the beginning of 2007 and it

is from the first line of Core 2 Duo processors (released in summer

2006). I think "Enhanced Dynamic Acceleration Technology" is only


I have to confess that I sneaked unto LKML illegitimately, as I am not

a kernel hacker.


But thank you all very much for all the hints!


      Lesen Sie Ihre E-Mails jetzt einfach von unterwegs.

www.yahoo.de/go


--
[view original message]
From: Pavel Machek <pavel@...>
To: Michael Meyer <mike65134@...>
Cc: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Wednesday, April 2, 2008 - 7:31 am

You can still play one, and submit a patch. Over internet, noone can

tell if you are hacker or not ;-). I would be happy to help with good

advice ;-).

									Pavel

--

(english) http://www.livejournal.com/~pavelmachek

(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

--
[view original message]
From: Luciano Rocha <luciano@...>
To: Michael Meyer <mike65134@...>
Cc: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Tuesday, March 25, 2008 - 10:08 am

maxcpus=3D1 should turn off the SMP alternative and switch to UP only,

optimising some locks and instructions.


--=20

Luciano Rocha <luciano@eurotux.com>

Eurotux Inform=E1tica, S.A. <http://www.eurotux.com/>
[view original message]
From: Andi Kleen <andi@...>
To: Luciano Rocha <luciano@...>
Cc: Michael Meyer <mike65134@...>, <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Tuesday, March 25, 2008 - 1:16 pm

CPU hot unplug will do the same. But it is unlikely it accounts

for that much performance difference.


If he used maxcpus=0 it would make sense. maxcpus=0 disables

the IO-APIC which likely makes a large difference. But it should

be actually slower.


There should be actually no difference in theory between max_cpus=1

and hot unplug to one CPU. Might be some bug.


-Andi

--
[view original message]
From: Pavel Machek <pavel@...>
To: Andi Kleen <andi@...>
Cc: Luciano Rocha <luciano@...>, Michael Meyer <mike65134@...>, <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Friday, March 28, 2008 - 10:00 am

Or thermal effect.  I'd expect that behaviour with bad cooling.


							Pavel

--

(english) http://www.livejournal.com/~pavelmachek

(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

--
[view original message]
From: Michael Meyer <mike65134@...>
To: Andi Kleen <andi@...>
Cc: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Tuesday, March 25, 2008 - 1:23 pm

I had the following time values:


maxcpus=1:

real	0m1.642s

user	0m1.528s

sys	0m0.068s


maxcpus=2 and

echo 1 > /sys/devices/system/cpu/cpu1/online:

real	0m2.579s

user	0m4.096s

sys	0m0.160s


maxcpus=2 and

echo 0 > /sys/devices/system/cpu/cpu1/online:

real	0m3.757s

user	0m3.632s

sys	0m0.112s


      Lesen Sie Ihre E-Mails jetzt einfach von unterwegs.

www.yahoo.de/go


--
[view original message]
From: Len Brown <lenb@...>
To: Michael Meyer <mike65134@...>
Cc: Andi Kleen <andi@...>, <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Tuesday, March 25, 2008 - 7:27 pm

this above is the baseline, yes?

it is same as if you used no boot param


Please post the contents of

# grep . /sys/devices/system/cpu/cpu*/cpufreq/*

and also

grep . /proc/acpi/processor/*/power


My guess that the maxcpus=1 case benefits from turbo mode, aka EIDA.

That benefit, however, is subject to this bug:

http://bugzilla.kernel.org/show_bug.cgi?id=5471

because for a single thread to run faster than the marketing MHz,

the other thread must be in deep-idle, which is prevented

by the bug above.


If your scaling_available_frequencies includes 2401000

then you probably have a turbo-mode enabled processor.


one way to verify this would be to disable turbo mode

by pegging the MHz like so:


# echo 2400000 > /sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_max_freq

# echo 2400000 > /sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_max_freq


-Len

--
[view original message]
From: Michael Meyer <mike65134@...>
To: Len Brown <lenb@...>
Cc: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Wednesday, March 26, 2008 - 3:26 am

Yes. I just repeated it - once without the commands and once with the


# grep . /sys/devices/system/cpu/cpu*/cpufreq/*

/sys/devices/system/cpu/cpu0/cpufreq/affected_cpus:0

/sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_max_freq:2400000

/sys/devices/system/cpu/cpu0/cpufreq/cpuinfo_min_freq:1600000

/sys/devices/system/cpu/cpu0/cpufreq/scaling_available_frequencies:2400000

1600000

/sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors:ondemand

userspace conservative powersave performance

/sys/devices/system/cpu/cpu0/cpufreq/scaling_cur_freq:1600000

/sys/devices/system/cpu/cpu0/cpufreq/scaling_driver:acpi-cpufreq

/sys/devices/system/cpu/cpu0/cpufreq/scaling_governor:ondemand

/sys/devices/system/cpu/cpu0/cpufreq/scaling_max_freq:2400000

/sys/devices/system/cpu/cpu0/cpufreq/scaling_min_freq:1600000

/sys/devices/system/cpu/cpu1/cpufreq/affected_cpus:1

/sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_max_freq:2400000

/sys/devices/system/cpu/cpu1/cpufreq/cpuinfo_min_freq:1600000

/sys/devices/system/cpu/cpu1/cpufreq/scaling_available_frequencies:2400000

1600000

/sys/devices/system/cpu/cpu1/cpufreq/scaling_available_governors:ondemand

userspace conservative powersave performance

/sys/devices/system/cpu/cpu1/cpufreq/scaling_cur_freq:1600000

/sys/devices/system/cpu/cpu1/cpufreq/scaling_driver:acpi-cpufreq

/sys/devices/system/cpu/cpu1/cpufreq/scaling_governor:ondemand

/sys/devices/system/cpu/cpu1/cpufreq/scaling_max_freq:2400000


# grep . /proc/acpi/processor/*/power

/proc/acpi/processor/CPU0/power:active state:            C0

/proc/acpi/processor/CPU0/power:max_cstate:              C8

/proc/acpi/processor/CPU0/power:bus master activity:     00000000

/proc/acpi/processor/CPU0/power:maximum allowed latency: 8000 usec

/proc/acpi/processor/CPU0/power:states:

/proc/acpi/processor/CPU0/power:    C1:                  type[C1]

promotion[--] demotion[--] latency[000] usage[00000000]

duration[00000000000000000000]

/proc/acpi/processor/CPU1/power:active state:            C0

/proc/acpi/pr...
[ message continues ]
[view original message]
From: Wander Winkelhorst <w.winkelhorst@...>
To: Michael Meyer <mike65134@...>
Cc: Andi Kleen <andi@...>, <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Tuesday, March 25, 2008 - 1:49 pm

What kind of CPU are you using? Some Intel CPU's do "funny stuff",

like dynamically overclocking itself when working on a single thread,

or using all of the 2nd level cache instead of sharing it with the

second core.


Regards,

Wander Winkelhorst.

--
[view original message]
From: Michael Meyer <mike65134@...>
To: Wander Winkelhorst <w.winkelhorst@...>
Cc: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Tuesday, March 25, 2008 - 1:56 pm

--- Wander Winkelhorst <w.winkelhorst@gmail.com>


Intel Core 2 Duo E6600 (2.4 Ghz). I do not think that

it is capable of dynamically overclocking.


$ cat /proc/cpuinfo

processor	: 0

vendor_id	: GenuineIntel

cpu family	: 6

model		: 15

model name	: Intel(R) Core(TM)2 CPU          6600  @

2.40GHz

stepping	: 6

cpu MHz		: 1600.000

cache size	: 4096 KB

physical id	: 0

siblings	: 2

core id		: 0

cpu cores	: 2

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 pni monitor ds_cpl vmx est tm2 ssse3 cx16

xtpr lahf_lm

bogomips	: 4791.87

clflush size	: 64


processor	: 1

vendor_id	: GenuineIntel

cpu family	: 6

model		: 15

model name	: Intel(R) Core(TM)2 CPU          6600  @

2.40GHz

stepping	: 6

cpu MHz		: 1600.000

cache size	: 4096 KB

physical id	: 0

siblings	: 2

core id		: 1

cpu cores	: 2

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 pni monitor ds_cpl vmx est tm2 ssse3 cx16

xtpr lahf_lm

bogomips	: 4788.35

clflush size	: 64


      E-Mails jetzt auf Ihrem Handy.

www.yahoo.de/go


--
[view original message]
From: Andi Kleen <andi@...>
To: Wander Winkelhorst <w.winkelhorst@...>
Cc: Michael Meyer <mike65134@...>, Andi Kleen <andi@...>, <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Tuesday, March 25, 2008 - 1:57 pm

P0 should also work fine with cpu hotplug. At least in theory.


-Andi

--
[view original message]
From: Michael Meyer <mike65134@...>
To: <linux-kernel@...>
Subject: Re: performance differences: "maxcpus=1" vs. "echo 0 &gt; /sys/devices/system/cpu/cpu1/online"
Date: Tuesday, March 25, 2008 - 12:38 pm

Can this be achieved during runtime also? And is it

possible to switch back to SMP after that?


      Lesen Sie Ihre E-Mails jetzt einfach von unterwegs.

www.yahoo.de/go


--
Previous thread: Re: What to do about the 2TB limit on HDIO_GETGEO ? by Mark Lord on Tuesday, March 25, 2008 - 9:34 am. (8 messages)

Next thread: RE: Performance changes between 2.6.13 and 2.6.23 by Sanders, Rob M. on Tuesday, March 25, 2008 - 9:52 am. (2 messages)