C2Q + Linux: which core is which?

fes016 · 2008-01-01, 06:27

I have been running 4 instances of prime95 on my Xeon X3210 (c2q) with the -An switches under Windows XP x64. I do TF on cores 0 & 2, and LL on cores 1 & 3. Now with the new year upon us, I figure I'd make a resolution to limit the use of M$ products ;) so I installed Fedora 8. I've got everything up and running, but my question is, does Linux count the cores the same way? I'm wondering because work seems to be done a little slower, at least for LL. The 2 LL instances went from 0.73/0.73 to 0.77/0.77. I don't know how TF is affected because now it outputs after roughly every 77 seconds as opposed to every 250 seconds. Another reason I'm wondering is because I noticed that the processor number and the core id's don't match in /proc/cpuinfo:

Code:

processor	0 1 2 3
physical id	0 0 0 0
siblings	4 4 4 4
core id		0 2 1 3
cpu cores	4 4 4 4

So when I set the affinity for each instance of prime95, what does it go by, the processor number or the core id? Does this mean I should switch the work for -A1 and -A2?

FES

fes016 · 2008-01-02, 00:26

Well, I was finally able to access my Xeon X3210 system and swapped the work for -A1 and -A2, and the iteration times for the LL tests are now faster at 0.073/0.073 (oops, wrote the times wrong in the previous post, should have been 0.073/0.073 and 0.077/0.077

). So now I have TF on cores 0 & 1, and LL on cores 2 & 3. Strange.

What's also strange is that for TF I still get outputs to the screen every ~77 seconds instead of ~250 seconds when I was running prime95 under Windows XP x64. Any idea why the change? I didn't edit any of the configuration files between installs. Is there a way to increase the time between outputs for TF? (The iteration output setting under Options->Preferences is for LL tests, isn't it?)

FES

patrik · 2008-01-02, 22:39

On my Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz "processor" and "core id" are the same. I'm running

Code:

---@nu:~> uname -a
Linux nu 2.6.18.2-34-default #1 SMP Mon Nov 27 11:46:27 UTC 2006 x86_64 x86_64 x86_64 GNU/Linux
---@nu:~> more /etc/SuSE-release
openSUSE 10.2 (X86-64)
VERSION = 10.2

2008-01-02, 00:26	#2
fes016 Oct 2007 USA 1000₂ Posts	Well, I was finally able to access my Xeon X3210 system and swapped the work for -A1 and -A2, and the iteration times for the LL tests are now faster at 0.073/0.073 (oops, wrote the times wrong in the previous post, should have been 0.073/0.073 and 0.077/0.077 ). So now I have TF on cores 0 & 1, and LL on cores 2 & 3. Strange. What's also strange is that for TF I still get outputs to the screen every ~77 seconds instead of ~250 seconds when I was running prime95 under Windows XP x64. Any idea why the change? I didn't edit any of the configuration files between installs. Is there a way to increase the time between outputs for TF? (The iteration output setting under Options->Preferences is for LL tests, isn't it?) FES Last fiddled with by fes016 on 2008-01-02 at 00:27

2008-01-02, 22:39	#3
patrik "Patrik Johansson" Aug 2002 Uppsala, Sweden 651₈ Posts	On my Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz "processor" and "core id" are the same. I'm running Code: ---@nu:~> uname -a Linux nu 2.6.18.2-34-default #1 SMP Mon Nov 27 11:46:27 UTC 2006 x86_64 x86_64 x86_64 GNU/Linux ---@nu:~> more /etc/SuSE-release openSUSE 10.2 (X86-64) VERSION = 10.2 Last fiddled with by patrik on 2008-01-02 at 22:41 Reason: Deleted login name for security reasons

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Core i5 2500K vs Core i7 2600K (Linear algebra phase)	em99010pepe	Hardware	0	2011-11-11 15:18
exclude single core from quad core cpu for gimps	jippie	Information & Answers	7	2009-12-14 22:04
Startup config for Quad Core on Linux	dswanson	Hardware	4	2008-01-28 16:29
Optimising work for Intel Core 2 Duo or Quad Core	S485122	Software	0	2007-05-13 09:15
Linux FC3 - mprime v23.9 : Segmentation fault (core dumped) nohup ./mp -d	T.Rex	Software	5	2005-06-22 04:22

2008-01-01, 06:27	#1
fes016 Oct 2007 USA 8₁₀ Posts	C2Q + Linux: which core is which? I have been running 4 instances of prime95 on my Xeon X3210 (c2q) with the -An switches under Windows XP x64. I do TF on cores 0 & 2, and LL on cores 1 & 3. Now with the new year upon us, I figure I'd make a resolution to limit the use of M$ products ;) so I installed Fedora 8. I've got everything up and running, but my question is, does Linux count the cores the same way? I'm wondering because work seems to be done a little slower, at least for LL. The 2 LL instances went from 0.73/0.73 to 0.77/0.77. I don't know how TF is affected because now it outputs after roughly every 77 seconds as opposed to every 250 seconds. Another reason I'm wondering is because I noticed that the processor number and the core id's don't match in /proc/cpuinfo: Code: processor 0 1 2 3 physical id 0 0 0 0 siblings 4 4 4 4 core id 0 2 1 3 cpu cores 4 4 4 4 So when I set the affinity for each instance of prime95, what does it go by, the processor number or the core id? Does this mean I should switch the work for -A1 and -A2? FES