J'ai le même prolbème... le iddle du CPU est désactivé et ma temp monte.. arg!
t'as trouvé une solution?

ben il suffit de remettre l'acpi  

héhé en fait mon pbme est plus compliké ke ca
c'est pas l'acpi ki me vire le cpu iddle mais le patch 0.14 qui enlève le bug IDE du via686B...

passe au 0.15

le 0.15 me fait grésiller la sb live comme pas possible  

On peut meme plus manger tranquille maintenant
 
Non c'est pas resolue mon CPU n'est plus en Idle ... Ce ne suis pas encore sur que ca vienne de L'ACPI d'ailleur ...
 
J'ai mi un Rad dans mon Circuit est ca va mieux .... mais c'est bof bof ...
 
J'ai installe CPUIDLE 5.8 pro ... AUCUN EFFET .... je comprends pas ? voila a mediter ...

au passage, c quoi l'ACPI?  car c pas la premiere fois que j'en entend parler, ms je sais tjrs pas c que c!!

-> DOWN !
 
C'est la gestion d'energie de Win2k ...
 
Mais en plus de ca c'est L'ACPI QUI GERE LES IRQ ... pas le BIOS ... donc pour regler mes pb j'ai du la virer azfin de definir mes IRQ moi meme !

Pims a écrit a écrit : On peut meme plus manger tranquille maintenant   Non c'est pas resolue mon CPU n'est plus en Idle ... Ce ne suis pas encore sur que ca vienne de L'ACPI d'ailleur ...   J'ai mi un Rad dans mon Circuit est ca va mieux .... mais c'est bof bof ...   J'ai installe CPUIDLE 5.8 pro ... AUCUN EFFET .... je comprends pas ? voila a mediter ...

 
Pims c'est quoi ton Circuit ?
sinon un truc tres louche si je desactive la sblive
dans le gestionaire de peripherique de win2k
la temperature de mon cpu baisse de 8° et arrive
au meme niveau que sous win98se

a ok!! merci
ms c que ds wink car moi dans mon bios ça y est acpi et c desactvé!!

Pims pour tes prob de surchauffe, repasse a la version 0.10 non?
 
Sinon ou trouves-tu les 0.15, je les trouves pas?

>> XOR
 
mon circuit:
http://forum.hardware.fr/sqlforum/ [...] inc&cache=
 
>> astro
 
De quoi parles tu?
 
Mon CPU ne se mets plus en IDLE depuis que j'ai virer l'ACPI ...

Pims ta pas mis le patch via latency toi?
parce kil paraitrait ke ca vienne de ca les prob de temperature.

 
Ben non c'est Nasdak ... par contre zavez pas un lien histoire que je vois un peu a quoi ca sert ...

PCI Latency patch
for VIA Chipsets:
Questions and Answers
Copyright (C) 2001, George E. Breese. All Rights Reserved.
 
Version 0.14 4/20/01
 
Important notes:
This product could, in a rare case, damage a computer. Use it carefully and at your own risk.
 
If you have installed a previous version of this driver, uninstall it before installing this version.
 
All credit for research on this patch is given to the site au-ja! Their page review-kt133a-1.html describes the problem and solution in German.
 
I wish to thank the staff of www.viahardware.com for promoting my work. They have been supportive of my drivers since I first began creating them.  
 
License: This software is licensed, not sold. The author of this product has granted you a license to use this product, subject to the following conditions. By possessing, using, or attempting to use this product, you assume all liability for its use. You agree never to take legal action, civil or criminal, against its author for any reason. You may redistribute this product in its original form only. You may not charge money for distribution of the product, unless all such charges are remitted to the author immediately upon payment. The author retains ownership of all intellectual properties embodied in this product.  
 
This driver has one known bug. If run on Windows 2000, it will not re-apply the patch when the computer returns from "hibernation".  
 
The v0.11 driver is identical to the v0.10 driver, except that v0.11 resets the PCI bus arbitration to 96 CLKs where v0.10 disabled PCI bus arbitration.  
 
The v0.12 driver is identical to the v0.11 driver, except that v0.12 resets the "Disconnect enable when STPGNT Detected" flag in register 52. This register was first mentioned by 'Wotsupnow' on the viahardware.com forum on 4/18/01.  
 
Tracey-Anne in New Zealand gets special thanks for her contribution to version 0.12 of this driver. At my request, she checked every setting of register 52 and determined which one solved problems with crackling audio on her SoundBlaster Live card.  
 
The v0.14 driver corrects a bug in v0.12. The changes to register 52 should only be performed on Athlon chipsets. The 598 and 691 specifications show different uses for register 52. Thanks to simonv in Holland for reporting the bug.  
 
What is it?
This is a driver for Windows 9X, NT, and 200X. It is intended to stabilize the PCI bus of computers that have certain Via chipsets.
 
NOTE: This driver does not actually configure PCI latency. It refers to PCI latency because it is the nickname that users have given to the problem.
 
Do I need it?
If your computer has any of the following symptoms:
 
Files copied from one IDE hard disk drive to another are corrupted  
The computer stops ("freezes", "hangs", "locks up" ) when playing sounds, especially to a Creative SoundBlaster LIVE! card  
The computer stops ("freezes", "hangs", "locks up" ), or files are corrupted, when copying to or from an IDE DVD, CDROM, or CDRW drive  
Sounds from a SoundBlaster LIVE! card are distorted when other PCI settings are adjusted in the computer  
AND, if your computer has a VT82C686B chip on its motherboard, along with one of the following chips:
 
Apollo Pro, Pro Plus, Pro133, Pro133A; labeled "VT82C691" or "VT82C693" or "VT82C693A" or "VT82C694X"  
KX133, labeled "VT8371"  
KT133, labeled "VT8363"  
KT133A, labeled "VT8363A"  
Pro266, labeled "VT8633"  
KT266, labeled "VT8366"  
then you might need this driver.
 
VIA has acknowledged a problem with their VT82C686B controller chip. The VT82C686B is a chip that is responsible for, among other things, controlling the IDE bus. When overloaded by an incorrectly-configured motherboard, the chip can cause corruption and/or lockups. The SoundBlaster LIVE! card seems to make the problem worse.
 
To confirm that your computer has the correct chipset, try any of the following.
 
Open the computer and examine the motherboard. A large, flat, square chip near the CPU will have the VIA logo and the part number. This chip may be under a heat sink and may be difficult or impossible to read, in which case you will have to try a different method.  
Download Sisoft Sandra. Examine its Mainboard Information. The 2001 edition of Sandra should identify the chipset clearly.  
Refer to your motherboard manual.  
How do I install it?
You should have received this in a .ZIP compressed file. Extract the entire .ZIP file to a suitable location.
 
If your computer is running Windows 95, 98, 98 Second Edition, or ME, then run SETUP_9X.BAT .
 
If your computer is running Windows NT 4.0 or 2000, then run SETUP_NT.BAT.
 
After running the appropriate .BAT file, restart the computer.
 
How do I know that it is working?
Copy large files to and from multiple hard disks while playing sound files. The files should be at least 100MB in size. If the computer freezes during such activity when this driver is absent, but works when the driver is present, then the driver is working.
 
What could go wrong with this driver?
Just about anything can go wrong. Your computer may display a "blue screen of death". The computer may become unstable. This driver may not help your computer at all.
 
A common problem reported by users of the v0.10 driver was distortion of sound coming from SoundBlaster LIVE! cards. This v0.11 version may improve the problem. Please report your findings to the email address at the end of this document.
 
How do I uninstall it?
The driver can be uninstalled from the Add / Remove Programs section of Control Panel. The driver will be removed immediately, but the PCI bus settings will continue to be in effect until the system is restarted.
 
If the computer is unstable, the driver can still be removed. Boot the computer in Safe Mode and then use the Add / Remove Programs section of Control Panel to uninstall the driver.
 
How much memory does this driver use?
This driver uses almost no memory at all. Its only job is to reprogram the chipset one time at startup.
 
Windows drivers are special programs. They have a special section that only occupies memory while the driver is starting. Nearly all of this driver's program code is located in this special section.
 
How did you make this driver?
I used notes from the Internet. I stumbled onto a thread of conversation at www.viahardware.com on the subject and became curious. A user finally reported that the au-ja site had a potential solution to the problem. Their site said the only solution was to use WPCREDIT and WPCRSET. I knew that some people, myself included, have problems with those utility programs, so I offered to implement the solution in a friendlier form.  
 
To create the driver files, I used portions of a driver I'd already written. The previous driver speeds up Via-based motherboards by enabling memory interleave.  
 
The source code of this driver is a mix of assembly-language and C++ code. To compile it, I use Microsoft Visual C++ 6.0 and Microsoft's free Windows 2000/ME DDK. My Windows 9X VxD is based upon Microsoft's generic sample VxD, and my NT kernel driver is actually an Visual C++ AppWizard-generated DLL file with some carefully-chosen compiler and linker options.
 
Will you give us the source code?
I do not wish to share the source code at this time. It is not formatted cleanly, and I may elect to use it in a commercial product later.
 
But, for technically-inclined people, here is the key portion of the source code. This is the only portion that I consider to be key to this problem.
 
 //
 // Modify the necessary PCI flags in this Via chipset.
 //
 // In register 0x70, disable PCI Master read caching (bit 2)
 // and Delay Transaction (bit 1).
 //
 u70 = read_pci(0, 0, 0, 0x70, 1);
 u70 &= 0xF9;
 write_pci(0, 0, 0, 0x70, u70, 1);
 //
 // Reset PCI arbitration timeout to 3x32 clocks (set to 96 clocks). I hate to
 // force an entire byte of data into an already-set register, but oh well.
 // The only other bit set is described by Via as "(enable) Fair arbitration
 // between PCI and CPU".
 //
 write_pci(0, 0, 0, 0x75, 0x83, 1);
 //
 // Clear the "Disconnect enable when STPGNT Detected" flag. (v0.12) This
 // solve any problems with sound distortion caused by registers 70 and 75.
 // Thanks to Tracey-Anne in NZ for checking this register bit-by-bit!
 // (v0.14) Thanks to SimonV in Holland for noting that this is not for
 // Pro133's! It is for Athlons only.
 //
 if(bAthlon)
 {
  u52 = read_pci(0, 0, 0, 0x52, 1);
  u52 &= 0x7F;
  write_pci(0, 0, 0, 0x52, u52, 1);
 }
 //
 // All done!
 //
 
How do I contact you?
If you wish to offer feedback on this driver, send email to feedback@networking.tzo.com . Please note that I do not check this mailbox frequently, and I do not answer questions about installation or use of this driver.

tu peux le telecharger
ici http://www.viahardware.com/download/viatweak.shtm
 
le pb du patch v.14
c'est que le "disconnect enable when stpgnt Detected" est desactivé
(offset 52 bit 7)
la temperature du cpu augmente
par contre avec un bon refroidissement comme le tient
ton systeme devrait etre plus stable
car il y a moins de variation de voltage le cpu consommant
presque toujours la meme chose

Merci Astro je sais tout maintenant
 

 
Ok donc j'en ai pas besoin ... le Pb c'est que mon eau chauffe plus vite de ce fait
 
Avant en utilisation simple l'eau chauffait pas ...
 
C'est appartir du moment ou j'ai Desactive l'ACPI ... que mon CPU ne s'est plus mis en IDLE ...
 
J'hesite a reactiver l'ACPI

Pims tu as desactiver l'acpi a cause des irq partage?
Si c ca tu pe le reactiver car win2000 gere tres bien le partage d'irq meme si toute t cartes sont sur la meme.
Un pote l'a reactive et pas de prob.
Mais bon c toi ki voit.

 
Oui Win2K peut etre ... mais le materiel pas sur ... j'avais des plantage avant ... et en la desactivant ... plus rien donc dans ma logique ca venait bien de la ...

 
ca c'est la theorie. en pratique on peut avoir des merdes avec l'acpi. chez moi avec acpi activé, mes transferts en ethernet 100Mb foirent dans les 20s. et j'ai moins d'ecrans bleus qu'avant (deja que j'en avais pas beaucoup)
pour la temperature du cpu j'ai "résolu" le probleme en installant cpucool qui fait passer mon proc, quand il ne fait rien bien sûr, de 53-54° a 35-36°

Pims a écrit a écrit : Merci Astro je sais tout maintenant   Citation : par contre avec un bon refroidissement comme le tient ton systeme devrait etre plus stable car il y a moins de variation de voltage le cpu consommant presque toujours la meme chose   Ok donc j'en ai pas besoin ... le Pb c'est que mon eau chauffe plus vite de ce fait   Avant en utilisation simple l'eau chauffait pas ...   C'est appartir du moment ou j'ai Desactive l'ACPI ... que mon CPU ne s'est plus mis en IDLE ...   J'hesite a reactiver l'ACPI

 
c'est bizarre, ton eau chauffe tellement ?
moi avec mon kit senfu et ses tuyaux au diametre ridicule
l'eau ne chauffe presque pas malgres les 2 ventillos en 5V
du gros radiateur (je ne pouvais pas les supporter en 12V)
le cpu chauffe jusqu'a 50° en pleine charge mais
la temperature de l'eau est presque constante
 
par contre la t° du cpu (1350 à 1.97V)
22° en idle sous win 98
30° en idle sous win 2k (22° si sblive desactivé )
41° en idle sous win2k (si PCI Latency patch  0.14)
50° en pleine charge sous win2k

 
C'est ce que j'ai essayé de faire .... mais sans succes  
CPUIDLE est totalement inactif !!
 
 

Elle par de 20° pour arrive à presque 30) en qq heures ...
 
Ce qui fait que mon CPU demarre à 26/27° ... et ensuite 36/37° ... voila c'est pas dramatique mais quand meme ...

 
je parle de CPUcool pas CPUidle (que je n'ai jamais utilisé)
 
http://www.the-ctrl-alt-del.com/utilities/CPUCOOL9.ZIP

Ok merci ... je connais mais j'ai pas teste ... je vais donc tester !

ACPI = pas mechant.
Fabrikants de cartes = salopiôts qui respectent pas les normes et pondent des drivers de merde.
 
Dans 50% des cas activer l'acpi pose des problèmes monstres parce que les constructeurs sont pas foutus de pondre des pilotes aux normes et des cartes aux normes. Sur plusieurs PC avec l'acpi des fois Windows ne se lançait même pas, gros freeze au chargement !
 
Seul truc étrange mais a priori pas génant que j'ai remarqué chez un pote : sa Radeon bouffe pas d'irq sans l'acpi, mais ca provoque pas d'instabilité nulle part... donc ben c bien comme ca...

le nouveau test pour compatibilité : verifier le power management ...
ha les cons , C pas deja assez compliké comme ca  

CPUCOOL FONCTIONNE
 
C'est bizarre quand meme
 
Rain 1.0 -> Fonctionne pas
Rain 2.0 -> Fonctionne pas
CPUIdle 5.8 Pro -> Fonctionne pas
CPUCool -> Fonctionne
Je suis à 25° maintenant au lieu de 32 tout a l'heure

Pinaise ya tjrs qq chose qui deconne :
 
avec CPUCOOL actif ... j'ai des perturbations enormes sur mon image avec MoreTv (image et son) ... Donc je suis obligé de le desactiver le temps de finir le film ...