How stable are the intel HT CPU, OEM spec

[wingman99]

Can the Intel HT CPU handle this test

stress test dual prime 95 and one round of SPI 32m and surf the web or play a games or run 3dmark loop all at the same time. looking for 100% stable system.

Depending on your ram amount adjust both prime 95 custom total usage so there is not to much file swapping

This simulates stress testing for multitasking.

It would be nice to see some people take this test and see if it is possible to score a 100%

[M Powered]

I can run dual prime and still score relatively high from PCmark04.

If that is not a test of multitasking, I don't know what is.

[fordf250]

Thats what I use to test my overclocks.I dual prime,32m superpi and atitool find max mem to test stability on my prescotts all running at the same time and I surf as its running, cant stand to watch the screen.

http://community.webshots.com/photo/312249557/1315587780067863468IRasSu

http://community.webshots.com/photo/312249557/1315588034067863468rOBMEJ

http://community.webshots.com/photo/312249557/1312256240067863468LyAEkj

This gives max cpu ad gpu heat with some memory testing, sorry these are old shots and dont have any with all four running.

[Sjaak]

If a cpu cannot handle that at stock speed its either

- overheating
- faulty, to be returned to intel, also nearly impossible since intel stresstests them for weeks.

[MadHyeNa]

If a cpu cannot handle that at stock speed its either

- overheating
- faulty, to be returned to intel, also nearly impossible since intel stresstests them for weeks.Indeed.

Overheating, most of people think that their case is good, just by being tower.. that's not true, by placing your rig in a place, where that can't get airflow, and just having stock PSU [exhaust] fan (if it's a generic/non-branded one), than do expect overheating with prescotts.

I remember my prescott running onload ~60C everything stock (stock heatsink/fan/stock thermal paste), and then I mod my case, get aftermarket heatsink/cooler, and now I get oc-ed almost ~56C (max 60C) doing stress tests, or just folding.

Just try to see if you have airflow or not, and then stress things, and yes IDE cables can really stop airflow in your case..

[Tserrof]

Indeed.

Overheating, most of people think that their case is good, just by being tower.. that's not true, by placing your rig in a place, where that can't get airflow, and just having stock PSU [exhaust] fan (if it's a generic/non-branded one), than do expect overheating with prescotts.

I remember my prescott running onload ~60C everything stock (stock heatsink/fan/stock thermal paste), and then I mod my case, get aftermarket heatsink/cooler, and now I get oc-ed almost ~56C (max 60C) doing stress tests, or just folding.

Just try to see if you have airflow or not, and then stress things, and yes IDE cables can really stop airflow in your case..


Agreed. Exactly the case with my Prescott Celey D, back when it was in a generic case with the stock cooler.

[wingman99]

If a cpu cannot handle that at stock speed its either

- overheating
- faulty, to be returned to intel, also nearly impossible since intel stresstests them for weeks.
Yes i can surf the web fine, and play a few games.

I need opinions on what is wrong with the 6 new P4 HT, 5 new mother boards, 6 new memory kits i have tested. I'm looking to bee 100% stable, if I'm going to be able use it to it's absolute maximum potential.

The amount of testing a trying all new parts that i have done in the past 3 years would make any person ****ed off that multitasks like i do


all Intel P 3 are 100% stable same wall socket

all p4 without HT 100% stable same wall socket

all p4 with HT slightly unstable same wall socket 3 PSU


I'm afraid that new technology just does not hold up, that is why i started this post and put together a simple test that my system fails at.


yes my case cover has been off for 3 year
Just last month i tested a new Deathstar T7K250 hard drive that has new improved ECC technology , that was my 5 drive Ive tried.

With all the changes Ive installed 2 xp approximately 30 times, i have backup drive keeps all essentials for quick setup.

I'm waiting for a phone call from FSP to see if there is problems with my
power supply model, like signal noise past speck, because Ive only tired 3 power supplies.

This is the only test my PC fails and if i don't file swap it runs spi and dual prime without fail

As soon as i set prime memory usage to middle amount of file swap, it fails SPI.

Ive done more consistent testing, I'm narrowing it down it's taking allot of time.

I'm going to test affinity dual prime spi several times to see what happens.

The reason i have been doing this off an on for the past 3 year because hosting and playing games, some times there are small failures like my keyboard keys getting corrupted and crashes.

So i needed a failure test that was quick

Then i lost hope for multitasking with Intel HT,


Measurement is with Fluke 87 v in recording mode before, during, after, 2 stress test failures

18 AMP 12v rail under MAX load
12.27v MAX
12.20v Average
12.18v MIN



28 AMP 3.3v rail under MAX load
3.330v MAX
3.329v Average
3.328v MiN

[greenmaji]

prescots need better cooling..

having your case cover off releaves all air preasure IE not good..

[wingman99]

prescots need better cooling..

having your case cover off releaves all air preasure IE not good..
I always test the new motherboards out of the case sitting on a foam paid on the desktop with a fan blowing on it. like they do when the test them on the reviews and also when they do prototype design at AMD and INTEL. I think that is cool enough at stock speed.

I do have a laser temp gun and my case gets 5 degree higher readings with the cover on with all the case fans going and my power supply temp goes up to max fan speed quickly.

I also have a temp probe for my Fluke 87 v i can measure air case temp and it is a little higher than ambient temp with the hot air blowing around in a closed case before it leaves.

I have to keep the cover off because I'm always tyring new parts in hope of 100% stability

Just last week i was trying new memory.

Well i was able to speak to a FSP engineer but he could not speak English LOL. The senior engineer will be back in a few days he speaks English, nice people.

[wingman99]

I called Intel they told me to run it out of my case, they said that they never have any stability problems with there products

I also talked to the FSP seiner engineer and he assured me that my PSU model was within ripple specs and that is not the problem because all of them are tested before they leave.

Yesterday i tried a new ANTEC Neo HE 500 ATX12v v2.2 and ESP 12 compliant
No improvement

I thought about trying one more motherboard and processor, but i better call it quits on Intel

I will probably try one more PSU and call it quits.

[wingman99]

found the stability problem in all p4 pc's

Pc stability problems here is couple fixes

First all memory is not perfect
That is what causes 70% of the problems with people that have bad memory

If you surf the web this is not for, your os will last without virus about 2-3 years but you cant count on that if your memory is oem and sector address deiced to stamp the kernel on the had drive.

Has memory always been a problem yes. I have tested this on older pc's to

If you want to start multitasking again and you have some tough games to keep running and your tired of reinstalling windows os all the time. this is for you.

Warning this dose not cover bad programs but i could swear i saw some bites straighten them selves out i am not ****ting you i did see it happen when running a bad program. I can tell when there is a change and freaked me out.




memory ECC unbuferd/unregster if your not going to use huge amount of memory, buffered is for sync. Server Memory: It's not the ECC, it's the quality of the memory. In testing i can get it to fail in about the 3-4 days that's wonderful because I know what it takes to mess up a pc usually it takes 1 minuet to 1 hour with poor memory, that will test good with mem86.

New last year Server Memory
Kingston Technology already performs on its server memory, we have developed Dynamic Burn-in Testing, a unique process designed to further screen out potentially defective modules before they are shipped. The process is part of our ongoing commitment to taking quality assurance to a new level of excellence. good timing too


1 :memory Server Memory ECC unbuferd/unregster.

2: memory with loose timing, you can use server memory as reference for timing.

3: loosen the timing up on high performance memory, not near as good as memory manufactured running loose specifications.

DDR2 800=5-5-5-15 ok DDR 400=3-3-3-8 ok

Bad = tight timing memory

Now here is the thing people miss on HT and Dual thread.


This will help about 50% giver take by applications and stress of single thread. You make the decision on this, i have not dun allot of application testing

for need four speed hi stakes needs the affinity needs to be set, other programs you can use this to


Flaws:
Imagecfg does not seem to recognize folders with spaces in it like "C:\Program Files". To get by this copy the Imagecfg.exe file to that directory and exclude the directory /path.


1. Put imagecfg.exe in the game directory or wherever the exe file is (C:\Program Files\Lineage II\Game.EXE )
2. Run this command in the dos prompt in that directory

imagecfg -a 0x1 game.exe

Then your .exe will be "patched" to run only on core 1. Backup your original .exe file though because you are changing it.
go to Start -> Run...

type cmd (press enter and it will bring up the dos prompt)
cd\ (enter at the end of every line
cd program files
cd lineage ii
cd system
(should say c:\program files\lineage ii\Game.exe)
OR
Cut from desk top icon or here C:\Program Files\Electronic Arts\Need For Speed High Stakes

paste at DOS prompt > CD C:\Program Files\Electronic Arts\Need For Speed High Stakes

Hit enter

Then type at DOS prompt > imagecfg.exe -a 0x1 nfshs.exe

Hit enter

Run game Test with Windos task Manager Cpu Usage History shoud be on the left one for cpu 0x1 = core 1


imagecfg -a 0x1 game.exe for core 1
imagecfg -a 0x2 game.exe for core 2




0x1 is CPU 0
0x2 is CPU 1

link for down load http://www.robpol86.com/Pages/imagecfg.php

[jordon]

rofl <3

[wingman99]

I mad mistake the Kingston memory i was testing. The new server memory for some reason it did not test as well as the high quality memory with loos timings, also it did some wired things, still it's better than OEM.