Intel Dual Core 830 D and Windows XP Home

[zeemo]

I just got my new system with a Intel Dual Core 830 3.2Ghz
In Windows there is an option in the Task Manager/ Processes if you right mouse click on a process to set the process either to CPU0 or CPU1. I cannot find any screenshot or guides or info on how to configure this part.
Basically I would like to set the OS and Virusscan to run on 1 processor and games for example to run on the other processor? Anyone come accross the same idea / question?

 

[Guest]

I beleive its usually best to let the OS's scheduler take care of that part. Even after the stress test where we noticed strange thing.
The OS is program as to balance the load on the different ressources available.
If you really want to do this, THG made a little program to change the CPU affinity in a easier way, cant find it but ill look and ppl will probably post it.
But it would be quite easy, just put everything started by the user SYSTEM on CPU 0 plus the apps you want to run there. And then, all the other threads should(if your lucky heh) default on CPU1 (thats if the system is really using a good amount of ressource on the first core, the scheduler of windows should allocate them on the other core).

Its been shown that windows scheduler is sometimes skecthy, I suggest not playing to much with it heh

Asus P4P800DX, P4C 2.6ghz@3.25ghz, 2X512 OCZ PC4000 3-4-4-8, MSI 6800Ultra stock, 2X30gig Raid0

 

[zeemo]

Thanks for your input. I am interested in this tool now 🙂

 

[slvr_phoenix]

Its been shown that windows scheduler is sometimes skecthy

Where has it been shown that?

<font color=purple> یί∫υєг ρђœŋίχ
<i>The devil is in the details.</i></font color=purple>
@ 193K -> 200,000 miles or bust!

 

[zeemo]

I also found this with more details about setting affinity on multi processors:
http://www.tomshardware.com/cpu/20040528/task-assignment-01.html

 

[Guest]

<A HREF="http://www.tomshardware.com/cpu/20050714/index.html" target="_new">Rember this test =)</A>

Its sometimes sketchy because wtih 4 threads running on four(2 logical) cores, the scheduler give lots of time to the low priority thread (divx encoding)...IMO it makes it "sometimes skechy" =Þ

Gotta love this quote muhahah

Four applications ran in tandem on both platforms over a period of 14 days. In this case Intel's top model - the Pentium 840 EE - easily outperformed its rival from AMD. The double-core CPU from Intel achieved this result primarily through Hyperthreading - the division of the two physical processors into four virtual CPU units. We knew this to be the case because when the HT function was deactivated, the AMD Athlon 64 X2 4800+ came out on top.

This last finding is more theoretical in nature as Pentium EE systems are always shipped with activated Hyperthreading. It is not possible to say precisely what performance edge this gives AMD because of the different load distribution.

Asus P4P800DX, P4C 2.6ghz@3.25ghz, 2X512 OCZ PC4000 3-4-4-8, MSI 6800Ultra stock, 2X30gig Raid0

 

[slvr_phoenix]

Its sometimes sketchy because wtih 4 threads running on four(2 logical) cores, the scheduler give lots of time to the low priority thread (divx encoding)...IMO it makes it "sometimes skechy" =Þ

That's not sketchy. That's exactly what the OS should be doing with one 'free' processor and one unassigned process, regardless of priority. Low priority just means give more room for other processes, but if there are no other processes (that don't already have their own processor) to give way to...

I'm sure that had the review run a fifth high-usage normal-priority process, then you'd have seen the divx processes take a huge performance hit.

If you want to blame anything, blame Intel for not internally limiting the resources of HT to something more like 75real/25fake than 50/50.

I've found HT at times useful, and at times annoying. Far more often than not it's useful to me. But either way it's not M$'s fault for that. (As much as I hate not being able to blame M$ for yet something else...)

<font color=purple> یί∫υєг ρђœŋίχ
<i>The devil is in the details.</i></font color=purple>
@ 193K -> 200,000 miles or bust!

 

[Guest]

Umm point taken and noted! You are right its more in the HT implementation tahn the widnows scheduler. But Windows Xp does know that its a virtual CPU. The Os doesnt think it has four real core, so they could still do something about it but calling it skecthy or defective is too much I agree...At least until you argument with P4man on this

Asus P4P800DX, P4C 2.6ghz@3.25ghz, 2X512 OCZ PC4000 3-4-4-8, MSI 6800Ultra stock, 2X30gig Raid0

 

[slvr_phoenix]

The Os doesnt think it has four real core, so they could still do something about it

I'm not entirely sure how much control the OS would have at that level, or if it even should be the OS's responsability. I prefer to blame Intel for not giving more options or handling it better internally I think.

At least until you argument with P4man on this

I don't think I want to go there. I'm already exhausted and have a killer headache from the flu that I've been fighting for two days. For me banging my head against a wall figuratively is hardly better than literally at the moment. :O

یί∫υєг ρђœŋίχ @ 193K
Pleased to meet you. Hope you guessed my name.
But what's puzzling you is the nature of my game.

 

[P4Man]

>At least until you argument with P4man on this

Fat chance, slvr is right, and basically wrote what I wrote a couple of weeks ago in that much-too-long thread. Its not a bug, its a feature. If you dont like how it works out for you, change process priorities.

That said, I do believe MS Windows scheduler F*s up at times, giving way to big priority (if not realtime priority) to certain core threads involved in I/O (?), which makes windows 'freeze' quite a bit more often than Linux on single threaded cpu's. I'll see if I can find some more info on it later

= The views stated herein are my personal views, and not necessarily the views of my wife. =

 

[dunklegend]

At least until you argument with P4man on this

Lets get ready to RUMBLEEEEEEEEEEEE!

<b><font color=blue>If you try to please everybody, nobody will like you<font color=blue></b>

 

[slvr_phoenix]

That said, I do believe MS Windows scheduler F*s up at times, giving way to big priority (if not realtime priority) to certain core threads involved in I/O (?), which makes windows 'freeze' quite a bit more often than Linux on single threaded cpu's.

Alright, this is really true in my experience, especially so with the networking protocols. At their worst they can even lock out a three finger salute. :O (Though I also believe that drivers can cause significant mayhem as well.)

یί∫υєг ρђœŋίχ @ 193K
Pleased to meet you. Hope you guessed my name.
But what's puzzling you is the nature of my game.

 

[Guest]

Oh im somewhat disapointed!
But I stand corrected on my opinions about windows scheduler!

And I like the bug vs feature! Typical developper versus tester!Altough I agree it was meant to be like that wich makes it a feature!

Asus P4P800DX, P4C 2.6ghz@3.25ghz, 2X512 OCZ PC4000 3-4-4-8, MSI 6800Ultra stock, 2X30gig Raid0