How to make my system stable?

ShellShock

ShellShock

ShellShock

Associate
Joined
3 Jul 2004
Posts
179
I rebuilt my system over a year ago but it has never been 100% stable - in games like Medieval Total War 2 and Company of Heroes I sometimes get flickering textures. For example, in MTW2 on the battle map where there is a large city, the buildings and walls will flicker rapidly. This problem does not happen all the time (but it is happening more now), and I don't play games all the time, so I have put up with it, but I have tried a lot of things to try and fix it. I've assumed it is overheating somewhere, but both the CPU and GPU do not report any overheating. The GPU is not overclocked.

Anyway, I'm thinking of upgrading my system and I want to get rid of this problem once and for all. The most important things to me are:

100 % rock solid stable system
Quiet

This is my current spec and how I am thinking of upgrading, highest priority first:

HIS ATI Radeon X1950 Pro - I suspect this is the culprit for my flickering problem. I had a Connect3D ATI Radeon X1800 XT in the same rig before and that also had the flickering problem (so may be the GPU is not the problem - how likely is it that I would have 2 with the same problem?).

I will buy Geforce this time, but I don't know what model. I want something with similar or better performance than my ATI X1950Pro, but most importantly it must be stable, not run too hot (I will not be overclocking) and be relatively quiet. I'm not looking for an Uber rig, just something I can rely on and which will run the current and next crop of games at a reasonable speed and resolution (1680 * 1050).

Window XP - upgrade to Vista to make use of DirectX 10. Not likely to make any difference to the flickering problem?

Motherboard/CPU/RAM. Gigabyte GA_965P_DS3/Intel Core 2 DUO E6300/GeIL 2GB (2x1GB) PC6400 800MHz. I'm worried that the root of the problem is the Mobo overheating somewhere. The RAM has already been replaced once by Overclockers as the previous sticks failed memory tests. If I get a new GPU and the problem still occurs, then I guess I need to replace the Mobo; so I will be getting a new Mobo and probably a more up to date CPU and RAM. If I do this then again I want a system that is rock solid stable and quiet and could not be used to heat a small country.

Perhaps I should be buying an off the shelf rig instead (but this will cost too much)?

Any opinions will be gratefully received.
 
In order to narrow it down you'd need to do a few tests which attempt to isolate certain components.

1) Memtest iso. This will test your memory and report any problems, it does it outside of Windows, so drivers or software can be removed from the equation.

2) Orthos - It's a special version of Prime 95 that will run on all cores of your PC, generally it tests mainly the CPU, although it can do tests over both. The reason to it after the memtest is by that time you should have identified if it is the memory or not.

3) Furmark - This will test (mainly) the graphics card under Windows, if you get crashes, freezes, bluscreens or artefacts then you can be reasonably sure it's a GFX card problem (could also be PSU I suppose).

While running 2 and 3 you should be monitoring your temps for your graphics card and for your processor. I'd suggest getting Speedfan and running that, but make sure you have the correct core temperatures by downloading and running coretemp.

There is no reason to believe than an off the shelf rig would be more, or less likely to overheat or be unstable.

A google search will turn up all that software with no issues, it's all free.

If you have issues or questions keep posting :)
 
Flickering problem? Maybe its your motherboard? If you swapped out graphics cards and got the same problem with each one then it has to be your motherboard. faulty Ram, hard-drive, PSU would not give you that sort of issue, but a faulty motherboard would. A faulty CPU would just give you BSOD's and crashes you see.

Trying buying a cheap AGP/PCI-slot motherboard (depending on what those cards take) and see if the problem still persists.
 
What defender says is valid, but I'd suggest keeping the credit card in the wallet until you try the tests I suggested. Might as well do the free stuff first :)
 
The Halk said:
What defender says is valid, but I'd suggest keeping the credit card in the wallet until you try the tests I suggested. Might as well do the free stuff first :)
Click to expand...

Thank you - your advice is just what I need to get the ball rolling on fixing this problem once and for all. Obviously if I can pinpoint where the fault is that will be most of the battle - at the worst I will be able to replace that component and not face a major upgrade.

Up to now I've haven't had a definitive set of tests to run. So thanks again.
 
So far I have run memtest and it came up with 0 Errors.

I'm now running Orthos, with the "Blend - stree CPU and RAM" test.

Core Temp has identified my CPU as:

Intel Core 2 Duo E6300 (Conroe-2M)
Freq: 2800.04 Mhz (400.01 x 7.0)
Platform: LGA 775
VID: 1.3250v

Temperature Readings (while running Orthos):

Tj. Max: 85 C
Core #0: 65 C
Core #1: 64 C


SpeedFan at the same time shows:

Found Intel(R) Core(TM)2 CPU 6300 @ 1.86GHz

Fan1: 2500 RPM
Fan2: 1439 RPM

Temp1: 34C
Temp2: 56C
Temp3: -2C
HD0: 29C
HD2: 30C
Core0: 64C
Core1: 64C

Vcore1: 1.42V
Vcore2: 2.06V
+3.3V: 3.33V
+5V: 4.84V
+12V: 2.11V
-12V: -16.97V
-5V: -8.53V
+5V: 5,13V
Vbat: 3.04V

I get similar figures for the "Small FFTs - stress CPU" test.

I don't know if these temperature readings/fan speeds are abnormal or not.
 
I've been running Furmark too. With the windowed burn test at 1280 x 1024 the ATI CCC temperature slowly crept up to 70C over 4 minutes. I could not see any artifacts in the Furmark display. I don't know if this means anything...Also I ticked the box for Furmark to log the GPU temperature but I can't see where it saves this info.
 
Last edited:
Pea0n said:
Silly question, but are your drivers up to date?

- Pea0n
Click to expand...

Not at all - I updated them recently:

Code: 
Driver Packaging Version	8.552-081028a-070226C-ATI	
Catalyst® Version	08.11	
Provider	ATI Technologies Inc.	
2D Driver Version	6.14.10.6879	
2D Driver File Path	System/CurrentControlSet/Control/Video/{1A3D2BD6-12FF-49FB-B326-577F27342121}/0000	
Direct3D Version	6.14.10.0621	
OpenGL Version	6.14.10.8201	
Catalyst® Control Center Version	2008.1028.2134.36851	
AIW/VIVO WDM Driver Version	6.14.10.6238	
AIW/VIVO WDM SP Driver Version	6.14.10.6238

I've also updated my mobo BIOS and chipset drivers. Plus I removed my aging SoundBlaster PCI card (I changed to the audio on the mobo) and disconnected my floppy drive and one of my CD drives. Made no difference. I've tried running with the side off the system case - still the same. It has been really cold recently in my gaming room but even so I still get the flickering problem (it is so bad sometimes the game is unplayable).

I have just run Furmark fullscreen (1680 x 1080) in Xtreme Burning Mode for 5 minutes. The GPU temperature in CCC went up to 70C, with no artifacts that I could see.
 
The CPU temp readings are a high, not extremely high, but high. You may want to remove your cooler and reseat it, it might not be fitted properly. If you're not overclocking/overvolting you shouldn't be as high as that.

You seem to have eliminated memory, and graphics cards as the problems.

The bad news is you didn't easily find one component was faulty, so it leaves some less precise options.

PSU - When everything is going together too much power is being drawn, causing the PSU to wobble.

Heat - When things get too hot your PC is becoming unstable.

Speedfan is incorrectly reporting the voltages from your board. Sadly I can't tell you exactly how to fix this, but have you downloaded a config file for Speedfan for your specific board? That sometimes fixes the temps/voltages. An alternative is to see if your motherboard manufacturer has a program that does the same thing, if they do then this should be accurate.

As for temps, what kind of CPU temps are you seeing after load? I'm meaning 2-3 minutes after Orthos has been running for long enough to reach the maximum CPU temperature, what do the cores come down to. If they are only coming down to 40 or so then it indicates either poor airflow in the case (hence high ambient temps) or a poorly seated/thermal pasted heatsink.
 
Thanks for the advice again. Your comments above the relatively high CPU temperature has prompted me to go into the BIOS and load the "optimised defaults". Previously I had manually set the memory timings and overclocked the CPU. Up to now I had kind of assumed the problem was with the GPU so I had not seriously considered whether the CPU could be the problem.

Anyway, using the BIOS defaults the CPU temperature is about 10C lower, before and after stress (Orthos).

A test in MTW2 has not shown the flickering problem - but this will need a lot more testing.

For now this is a definite thumbs up.
 
Well I spoke too soon - the flickering problem is as bad as ever. This is with no overclocking in the CPU or GPU. I need to confirm the CPU core temps when the problem happens; I do know that the temps are generally 10C lower than before after booting Windows - about 32C.

I also have another symptom: when I was playing MTW2 last night and went into a battle, the buildings/walls were flickering like crazy. The more buildings they were it seems the more likely the problem is to occur. Also, sometimes the screen went completely black for a few seconds, and then the battle reloaded itself and I could continue playing from the same point. I'm wondering if this is the VPU Recover feature in the ATI CCC? That is the GPU had reset itself? If so, why would it do this?
 
I'd try getting completely rid of your graphics (and chipset drivers if that doesn't work) and installing them again. Sadly I can't remember the name of the program I used last to do it. I do remember it's free, so there's no need to pay for it.

Essentially what you do is go to control panel and uninstall the ATI drivers, and then reboot to safe mode. When you're in safe mode you run the cleaner program, which removes all traces of the drivers. This means Windows doesn't end up installing little bits of old drivers along with new ones.

Then reboot normally, and install the latest drivers.
 
Thanks for your advice (again).

I did what you said - using Driver Sweeper, and Driver Cleaner Pro. I also installed the very latest 8.12 drivers.

And...so far so good. The problem has not re-occurred (would have expected to see by now). I'm still not overclocking and core temps are maxing at 47C (GPU is also mid 40s).

I need to give this a lot more play testing before I will be convinced the problem has been fixed, but so far so good.
 
I can't figure this out. I played a couple of hours earlier today, and the system was OK, but now after a hour or so it has been flickering badly again.

So I guess it is not due to bad drivers...
 
ShellShock said:
I can't figure this out. I played a couple of hours earlier today, and the system was OK, but now after a hour or so it has been flickering badly again.

So I guess it is not due to bad drivers...
Click to expand...

*&£^!

I'm trying to work out what the problem could be then.

Your CPU temps are fine. Your GPU temps are cold.

Is it just the one game having issues?

Have you tried reading a book and running 3DMark on a loop for a few hours, every so often coming into the room to see if there's stuff going wrong.

Speedfan for me reports correctly on some but not all voltages.

+12V (Which is the main power draw)
+5V (I know some stuff uses this)
Vcore (Processor voltage)
+3.3V (No idea off hand what uses this)

It might be an idea to start speedfan running, and then play a game until it starts flickering, quit the game and get back to speedfan quickly to see if there's been significant fluctuation in any of the voltages.

Clutching at straws to be honest though...

AtiTool I think should work for your graphics card - AtiTool hasn't been updated in a while, so it doesn't work for me. Grab a copy of that and have it artefact test for a while, see if it comes up with anything. I believe it's a better test than Furmark, but again for me it doesn't work.
 
The next test I'm doing is running with only one stick of RAM, as suggested by Pea0n. The memtest should have already eliminated that as a problem but who knows? It will take me a day or so to test the various RAM combinations.

I also get the same problem in Company of Heroes, so I don't think it is game related.

I'm just going to have to be methodical about this and test one thing at a time. The alternative, which I've comtemplated, is to get out the credit card. The advantage of buying an off the shelf system is that if it does have a problem like this, then it is someone else's problem to fix. But that is a very expensive solution.

Really, it should not be this difficult. I bought all these components at the same time from overclockers as being good quality, well matched and up to date, but I've had this problem since the beginning. I did upgrade the GPU later to the X1950 thinking that would solve the problem, but no dice.
 
You must log in or register to reply here.
Back
Top Bottom