Author Topic: Lockup/out after time (Read 658 times)

Tilt · « **on:** November 02, 2003, 05:45:40 AM »

AMD XP2400+ (266) (but at 133 x 12.5 = 1.666Mhz)
GF FX5600 256DDR (quincunx AA + anistropic off + medium intellisample + enabled fog emulation @ high image) DX9
512RAM (133 single bank)
SB5.1

All the above is new.

When playing on line I will get either a system freeze or Host connection lost (followed by system freeze if I re log and launch without reboot)

It will happen after about 60 to 90 minutes of play.

It does not happen if I am downloading or doing other stuff on line.

It is not a zone alarm thing (have ruled that out).

Voltages and temperatures seem OK

I suspected a memory drain as the memory is cheap stuff. However I have run "memtest" twice and it reports no faults (is it a falible test?)

Is there another way to check the memory?

Is there another possible fault?

Tilt · « **Reply #1 on:** November 04, 2003, 02:05:34 PM »

could any of this be the cause..............

Furious · « **Reply #2 on:** November 04, 2003, 03:35:13 PM »

I had a similar problem. It turned out to be a failing power supply.

The symptoms, CTDs and spontaneous reboots, only showed when the temp inside the box increased. When the box was cool the PS worked fine.

zmeg · « **Reply #3 on:** November 04, 2003, 04:13:00 PM »

I have the same processor, mine runs at 133 x 15 =1995. Could it be your mobo doesn't support a 2400+ or mabey needs a bios update to run it properly. Also check your network card and the fan on your vidio card.

Skuzzy · « **Reply #4 on:** November 04, 2003, 06:32:01 PM »

A coupe of things I see in that list could cause problems. You have a USB controller sharing and IRQ with the video card and you have a USB controller sharing an IRQ with your Ethernet card.

Either of the above could cause a problem if the drivers are/were not written to handle sharing on IRQ's. Even if they are, you are hurting your systems performance.

Also note, the FX cards drivers are not the most robust drivers around. I would suspect the sharing there to be the issue.

Tilt · « **Reply #5 on:** November 05, 2003, 04:52:30 AM »

Quote

Originally posted by Skuzzy
You have a USB controller sharing and IRQ with the video card and you have a USB controller sharing an IRQ with your Ethernet card.

Thanks.

I can (I think) disable the onboard USB host controllers (Via tech 3038).

The Belkin and NEC stuff is the result of a PCI to USB board I put in to get a better & more stable interface with my USB JS & pedals. I could try moving it to another slot.

This PC has no use for other Coms( 1 & 2)or a Printer (LPT1).......could I disable these in bios and free up the IRQ's?

MB was very cheap but Bios does not give multipliers beyond 12.5.

Skuzzy · « **Reply #6 on:** November 05, 2003, 07:10:09 AM »

Sure, disable all the unneeded devices in the BIOS. First disable them in the Windows device manager, then reboot to the BIOS and disable them. After rebooting, check the device manager and see if they are still there and disabled.

You should be able to move the PCI/USB board to another slot and with a dedicated IRQ. Normally, slot 1 is shared with AGP, slot 2 and 5 are shared, slot 3 is usually shared with some onboard device, and slot 4 could be by itself.
Check the motherboard manual and see if they list what slots are physically sharing IRQ's with other devices.

You may have to manually set the IRQ's in the BIOS roms, once you have freed the COM/LPT device IRQ's.

Tilt · « **Reply #7 on:** November 05, 2003, 09:00:01 AM »

Quote

Originally posted by Skuzzy
Sur First disable them in the Windows device manager,

To be sure just disable in WDM or remove them.?

Tilt · « **Reply #8 on:** November 06, 2003, 05:45:57 AM »

Done the above........ In fact I now have 3 free IRQ's

The FX5600 shares now only with steering. (The belkin 2.0 is also disabled)

I cannot separate the network card from the NEC USB OHC. What ever I do the NEC USB OHC follows it! However the other Via stuff is disabled.

Having done this yesterday after work I then had about an hour on line in the MA, followed by an hour off (without reboot) followed by another part of an hour or so on until I had another "host connection lost".

The strange thing about these is that i have to reboot prior to reconnect or else the program will lock up. (it does not appear to be a simple loss of connection, it does not follow a loss of "UDP").

Given this I return to the memory question and this memtest thing. Is it possible that memtest does not uncover faults such a a gradually degrading memory?

Also could the FX5600 DDR memory be at fault.........I note that whilst AH1 shows it as a 258M card, AH2 and DXDiag (DX9)only show it as a 130M card. (it is a 256M card).

Checking the temps and voltages on reboot after such a fault i note the CPU at 47C and all voltages within limits.

Zanth · « **Reply #9 on:** November 06, 2003, 09:02:30 AM »

What power supply do you have? If memory serves that card requires a whole bunch of power (even has its own molex connector).

I was having a different sort of CTD (see thread "Chronic CTD")
, but in my case it was RAM voltage. I solved my problem by slightly adjusting my RAM voltage up.

P.S. (I had also run a memory test program 12 hours without error previous or I would have looked at this sooner. I am now not so confident of memtest programs. I am going to rely on hardware test equipment in the future.)

Tilt · « **Reply #10 on:** November 06, 2003, 10:51:09 AM »

Quote

Originally posted by Zanth
What power supply do you have? If memory serves that card requires a whole bunch of power (even has its own molex connector).

It all plugs into the PCI..........dunno what a molex connector is

I must admit its only 250W.........but as I say the voltages seem OK

I will up the ram voltage to ensure it is never minus however.