Windows XP Multi Core config - Does your OEM do it for you?

Hmm... I just did the fix, and now in 3dMark06, I'm getting like 1 FPS all the way instead on just the CPU.... can someone let me know what's wrong?

 

Well I did the hotfix, and the reg edit and things didnt change too much. I scored slightly higher in 3dmark05, adn cut off about 3 seconds off my super pi score. I will test in game performance to see if it has changed. I am hoping thi helped, as empire earth 2 is a hog on cpu cycles.

 

Actually MCE is professional with the extra media software

 

And the Domain login capabilities removed.

Ya, I always get that backwards. Either way, anything that works for XP (either version) will work on MCE.

 

Wow, this really made my desktop (Athlon64 X2 3800+) more responsive... all kinds of annoying lags seems to have vanished...

Thanks...

 

I have had to do the dual-core config on every dual-core machine I have tested so far. Not one manufacturer does this. Many times I'll find that the benchmark scores are completely off (as in, half the 3DMark06 score indicating that the second core isn't being used) if it isn't done.

 

Wow, definitely going to do this to my Dell e1505 when I get home.

 

First of all I like to say this is an excellent first. Second what is "/noexecute=optin" I do not have that in my bootini. I know it has to do with preventing something from executing but what?

 

All i had to do (i have the e1505) was to edit the boot.ini

but now, with i boot windows, it asks me what to start up, even though i can only choose Windows XP MCE.

why?

 

The boot.ini probably thinks there is another OS on your PC/NB. Nothing critical abet a minor inconvienience.

 

Just did this fix on my lappy. After double checking the syntax, command spelling and crossing my fingers....the update seemed to go without a hitch. It's hard to tell how much my performance has improved since I haven't done any benchmarks, but it sure does boot up faster. Good times!

Just 1 question though, what does the boot.ini mod do? Microsoft doesn't mention it on their hotfix info page. Thanks!

 

The /usepmtimer in the BOOT.ini is a "switch" that forces the system to use a steady clock in place of the CPU frequency... for certain timing functions. Thats as basic as you can explain it.

This BOOT.ini "switch" does not really affect a single core system, while it is mandatory for multi core systems with the hotfix installed for Windows XP.

NOTE: The BOOT.ini that I have given was meant for an example only. I would not recommend you to copy all of it and overwrite your own, unless if you are specifically using XP Pro SP2. Everyone else (XP Home, MCE2005) should just add the /usepmtimer only.

Glad that everyone is finding this thread useful.

Game On People,
-Gophn

 

From Sysinternals:

/NOEXECUTE
This option is only available on 32-bit versions of Windows when running on processors supporting no-execute protection. It enables no-execute protection (also known as Data Execution Protection - DEP), which results in the Memory Manager marking pages containing data as no-execute so that they cannot be executed as code. This can be useful for preventing malicious code from exploiting buffer overflow bugs with unexpected program input in order to execute arbitrary code. No-execute protection is always enabled on 64-bit versions of Windows on processors that support no-execute protection. There are several options you can specify with this switch:

* /NOEXECUTE=OPTIN Enables DEP for core system images and those specified in the DEP configuration dialog.
* /NOEXECUTE=OPTOUT Enables DEP for all images except those specified in the DEP configuration dialog.
* /NOEXECUTE=ALWAYSON Enables DEP on all images.
* /NOEXECUTE=ALWAYSOFF Disables DEP.

 

Applied the patch on the Sager 5760 2GB Core duo, WinXP Pro sp2.
I did a test render sequence in AE (adobe after effects) before and after the patch.
I ran the Total benchmark.
Before the patch my scores where :
1st render: 1min 24sec
2nd render: 19min 49sec

After Patch installed:
1st render: 1min 29sec
2nd render: 20min 59sec

As you can see, rendertime took longer after patch applied, AND in task manager one core was getting the load, the other one was way lower in the graph. I did look into rgistry and boot.ini, all settings are there as supposed....

 

I have a Samsung X60 which has a Core Duo (T2500), so I have read this thread with much interest... Right now, I have not made any changes to my system setup, though I do have the hotfix 896256 installed (having checked the add/remove programs list).

The article on Microsoft's KnowledgeBase ( http://support.microsoft.com/Default.aspx?id=896256 that discusses HotFix 896256 suggests that the Registry entry

HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\Throttle\PerfEnablePackageIdle​

is only necessary to disable the effect of the hotfix - this information is included in the KB article under the heading "How to disable the new performance state policy behavior".

So, I guess my question is (something like) this:
If you have the HotFix 896256 installed, does adding the Registry entry Throttle\PerfEnablePackageIdle=1 affect (improve?) your system performance? And does it affect battery life?

 

Yourmum does have a point. On the site provided above it stated:

"This solution favors performance gains over power savings. Although benchmark performance scores may improve, battery life could be negatively affected. Accordingly, this kernel policy change may be disabled by a registry key to allow for maximum flexibility."

Then it listed the Throttle Key change.

 

I just ran 3dmark and scored 2152, with the reg value set to 1. Then changed it to 0 and ran 3dmark again and got the same score. Doesn't seem to do much in that department.

And I am using a T2400.

 

Did you restart before running it with 0?

 

Strange to here that, I guess you should try outher benchmarks though to be sure, such as SuperPI, PCMark, and 3DMark. Others have all shown improvement with these benchmarks.

 

I didn't. After restarting though I'm getting scores of 2141. Minimal difference at any rate.

On another note though, I'd like to see if battery life takes a hit with this thing.

 

This registry change from 0 to 1 will increase performance using the state performance policy. As for the power saving, its has been shown that the battery life differences are very nominal at best, not even noticeable to most users... a few minutes less.

 

Zyphere,

Is that 3DMark05 or 06?? Or is that your CPU Score??? I would try more than one benchmark.... namely SuperPI, PCMark, and 3DMark06. You can compare these scores with others in this forum.

 

If anything will show a difference, it will be 3DMark06's CPU benchmark. It's multithreaded, and if you don't set up dual-core properly, then you may be getting scores that are way off. For example, I ran it on a previous review machine, and got roughly 860 in the test - after the dual-core patch, I got a proper 1600 something.

 

After further review:

" You can type 1 in the Value data box to enable the new performance state policy behavior."

Everything is A-OK.

 

That is 3DMark06.

My CPU Score the first time was 1547. With the reg value set to 0 it was 1544. SM2.0 also dropped from 766 to 758.

However I did notice a difference in the load times for the tests (Return to Proxycon loaded a lot more quickly when it was set to 1).

 

My understanding of that line is that if you have the Throttle\PerfEnablePackageIdle value in place, because you have previously set it to 0, then 1 re-enables it. i.e. effectively has no effect - it only exists so that you can enable the new performance state policy behavior after it has been disabled. The alternative would be to remove the registry entry, which is a bit harsh!

 

I'd like to keep this discussion going and ask others to submit their experiences with this patch. I did it and my computer doesn't really "feel" any faster but it doesn't seem to have detracted from its performance either.

Some people are reporting gains, others are reporting losses. This is rather confusing for me and I would like to know why there doesn't seem to be uniform results from all who have installed this patch. Even Microsoft appears to admit that WinXP is not initially configured to properly utilize the data handling capabilities of dual core processors.

I remain very interested and hope others continue contributing to this thread.

BTW, is there any way this thread can be made available to owners of other brand laptops besides Sager? This is an issue that obviously has relevance to all owners of dual core machines. Just a thought.

 

Agreed. Originally when I made this thread, i just wanted to show my findings from observations on my recon trips to computer stores. Didnt realize that it was going to turn into a Sticky.

I would like to change the title to:

Windows XP Multi Core config - Does OEM do it for you?

and stick it in the forums somewhere that it would get the word out more.

Chaz, what do you think?

 

I'll sub in for Chaz *clears throat* Sounds like a great idea. I'll stick it in the front page of NBR and NBR Forum and use 72 sized font with red highlights! That should get the word out.

 

Sorry I missed that, it is now changed.

 

No harm done

 

Thanks Chaz, but I was a little hasty in the change.

it should have a "your" in it.

Windows XP Multi Core config - Does your OEM do it for you?

-OR-

Windows XP Multi Core config - Do OEMs do it for you?

thanks again.

 

hi, will this decrease battery life?

 

it has been shown that the battery life does decrease by a few minutes (avg. 10mins)... its pretty nominal at best.

 

where do you know that from, may I ask?

 

i only had a chance to test it on my Clevo D900K with Dual Core Opteron, and on a number of Core Duo notebooks from Acer, HP/Compaq, and Dell... didnt have a chance to try it out on a mobile Core 2 Duo (Merom) notebook yet.

i have 2 sets of data: one for pre-hotfix, and one for post hotfix.

i leave all notebooks to idle on Windows XP desktop with default power scheme (without monitor turning off or system going into standby).

and then I put on the hotfix and make sure batteries are recharged to 100%.

I log the shutdowns in the Event Viewer when battery hits critical around 3%.

The difference for most of the Core Duo's were little at best, some went as high as 19 minutes being lost to as little as 4 minutes. Still very nominal at best.

If you have time to test your battery life with the hotfix, please post them.

Game on people,
-Gophn

 

That is a bit too basic for me. Could somebody explain a bit further what the /usepmtimer switch actually does?

 

I am not a Kernel programmer, but as I know it, the USEPMTIMER has to do with processor throttling causing the QueryPerformanceFrequency API to return values that cause the applications or games to run too slow (or even too fast at times)

Such as an application to run only one core rather than knowing that 2 are available to use... the Affinity for each program might see 2 cores but the load will only go onto one unless that particular switch is added into the BOOT.INI and the Throttle added into the registry.

-Gophn

 

does dell do this too? this thing that sager does?

 

Sager doesn't do it. See this post earlier in this same thread- http://forum.notebookreview.com/showpost.php?p=1438195&postcount=15

I highly doubt Dell would do it since MS says it's not "necessary".

 

MS says alot of stuff isn't necessary. They also say alot of stuff IS necessary that ends up borking your system. What I took note of in the MS article was their acknowledgement that the OS isn't using multi-core CPU's as efficiently as possible. That's why i decided to go ahead and implement this patch. It sure didn't hurt. My last 2M Super Pi score was 1:12 and my computer boots faster and runs like a champ. If your satisfied with the way your computer is running, then I would agree that this patch is not necessary. With any mod and registry change, if you don't do things exactly right, you may turn your machine into a boat anchor. That's prolly the main reason MS isn't crazy about people trying this en masse.

 

true that its not necessary for typical (average) users that just web browser, listen to music, watch DVDs, etc...
but the main reason for the hotfix is to increase performance, especially for those that are enthusiasts and gamers that want the most out of their systems.

 

This statement is alarmist and totally untrue.
Hosing the registry will NOT turn your machine into a boat anchor.

The registry is just software. At worst, you will have to reload windows. And if you lose your data because you don't have it backed up, that's your problem, not the hotfix's.
The vast majority of the time however, you can do a restore and your system will be absolutely fine.

I've worked on computers for nearly 15 years and I actually completely corrupted my registry 5 days ago. Worse than you could EVER do with a hotfix. I look up the error on the MS knowledgebase and was given explicit directions on how to restore my machine using my windows CD and the Recovery Console. An hour and several reboots later, my machine is running perfectly. And still is to this day.

 

Don't go ballistic on me dude. I am well aware that a software or registry problem can be corrected with a restore or OS reload. But for many people, this is a very intimidating process and until their systems are brought back on line, they are as useless as a "boat anchor". For people like you, me and others with experience this is not a problem, merely an inconvenience. For many others, it is a major problem and that was the point of my cautionary advice. This is not alarmist thinking. It's just a fact that many people without alot of experience who decide to start "tweaking" their gear can encounter issues that essentially leave them "dead in the water" until they receive knowledgeable assistance. Some may say "if you don't know engines, don't open the hood". I say computers are supposed to be fun and people learn by doing. Sheeesh.....sorry if you misunderstood what I was trying to say.

 

there i added some pics, hopefully that will decrease the chances of screwing it up.

Game On People,
-Gophn

 

Good job on the pics, Gophn. The hotfix worked like a charm for me.

C.

 

no prob, glad to see that its helping.

 

Excellent Gophn, you've really made the whole process noob proof.

 

Tried it on my W3J CoreDuo, installed hot fix and did the reg thing, exactl the same benchmarks... any clue?

 

I have 3 dell xps 1710 laptops they have intel duo processors in them

One laptop show up correctly in task manager.

The other 2 laptops only show one cpu in the task manager. I have applied this hot fix with no luck. I have windows xp and sp2 installed. Any ideas?