CTD- No Error!

[unmerged(18552)]

Hi!

I have a CTD problem like some other users here.

Like them i wont get any error message and the crash doesn´t occures at a certain time or game speed, but seems to arise more often when i play Austria, but i most play with this nation so this mustn´t be important.

Sometimes the game runs fine for 2 hours and crashes then, but sometimes i have to restart the game 5 times in a hour.

I habe no mods or player created scenarios installed.

here is my sys:

AMD AthlonXP 2400XP
512 MB Ram

NVIDIA GeForce4 Ti 4800 SE ->DirectX 9.0b
Sound:Terra Tec 128PCI
Sound and Graphics driver are new one and no prob´s at other games

Running a Windows 98SE
EU2Version 1,07

Thx for all help!

btw: is ther any chance to get the autosave option to less then one year for the time period?

If ther would be a monthly autosave the damage wouldnt be so big!

 

[AndrewT]

Rename the AVI directory to AVI_OLD

Run eu2_settings.exe and turn sound off, let us know if that helps.

Do running at a lower resolution or a slower speed help?

Try uninstalling EU2 (moving your save files elsewhere first), do a defrag of your hard disk, reinstall EU2 then 1.07 immediately. See if that helps.

 

[Castellon]

I assume since you reffered to the other thread that you have tried the AVI rename and also renaming the drivers.

It could be related to the VIA chip set, there are a couple of threads open on this subject. Just search support forum for posts by JPD, he is our resident expert on that problem.

 

[Castellon]

Okay so I am a nice guy and did it for you.

Originally posted by jpd
AMD Athlon + VIA chipset + complete lockup = infinite loop.

Had to fight this problem myself early this year. It is caused by a faulty AGP driver from VIA.

Uninstall the VIA 4in1 driver set completely, and use the Microsoft stock IDE drivers and AGP drivers. Then the lockups should no longer happen.

A brief explanation maybe in order about what is going on:

The AGP portion of the northbridge maintains a command/data queue for pending transactions, which works both ways. AGP transactions are always 64 bits. The problem starts when the software initiates a transfer starting on an odd DWORD boundary. The transaction must be completed in two cycles instead of one.

When, after the first cycle is complete but before the second cycle starts, the AGP card itself also request a transfer transaction, and this transaction for some reason requires the one from the CPU to be completed first, both transactions deadlock each other.

The interrupted transaction can no longer complete, as the video card will reject it, pending completion of it's own transaction. The AGP card's transaction cannot complete either, as it requires completion of the one initiated from the CPU. Because of the bouncing of the second part of the split CPU transaction, Microsoft has named this problem 'infinite loop'. Because of the lockout in the northbridge of the chipset, the processor is permanently locked out of all memory access, and thus cannot execute any other instructions, nor can it respond to interrupts. This will cause a soundcard to repeat it's last loaded sound fragment ad infinitum, for example.

Intel has documented this particular sitiuation in it's official AGP specification, with an explicit warning to software/driver developers to make sure the above described situation is to be avoided at all times, as it will completely freeze the system. It can be avoided completely by assuring that software drivers initiate memory transfers on even DWORD boundaries only, as the AGP transaction will then only need one uninterruptable cycle and thus avoids the deadlock situation. Unfortunately, the VIA programmers haven't made sure their drivers follow this advice.

Jan Peter