Tech Support banner

Status
Not open for further replies.
1 - 12 of 12 Posts

·
Registered
Joined
·
8 Posts
Discussion Starter · #1 ·
OK I have this system configuration:

MB ASUS A7N8X-E Deluxe
Proc Barton 2800+ w/ cooler Titan CU5TB
Vid ASUS V9570 GeForce FX5700 with 256MB
2x256MB Kingmax DDR400
HDD Seagate 160GB SATA

My problem started some time ago; while playing opengl/direct3d games, artifacts suddenly appeared on the screen (green/yellow color all over the screen, random times), and any following press of a key (or even waiting 10-20 seconds) would freeze the system, or send it into freeze+stand-by monitor mode. After some time I figured I should lower the CPU multiplier, so I did, and things got good again, for some time. Then it started happening again, so I finally got myself using my 2800+ (2.1 GHz) to 1.1 Ghz (6.5x166). Now, if I try to go into 2800+ again (12.5x166), it artifacts/freezes as soon as I enter the Windows XP boot logo, and sometimes even before that (in BIOS or as soon as the first text appears on the screen). The only thing I can think of is a conflict between my MB and my video card. I tried unplugging any "uneeded for boot" hardware (HDD, floppy, CD-ROM), tried changing my RAM modules (tried with a PQI 256 DDR266), nothing worked. Any suggestion is very apreciated. I hate staying at 1.1 Ghz :)
 

·
Registered
Joined
·
2,171 Posts
That sounds more likely to be a problem with the CPU, motherboard, or power supply than a problem with the video card. The easiest way to tell for sure is to borrow someone else's video card and see if it still has problems. I'd also try running Prime95 at various CPU multipliers to see if your CPU is really running properly. All of your symptoms can be caused by a flakey or overheating CPU and it's one of the easier possible causes to test.
 

·
Registered
Joined
·
8 Posts
Discussion Starter · #3 · (Edited)
Oh yeah, I forgot to mention, I tried the board on another system; works perfectly. Also tried another board on my system, likewise situation (worked without problems). So it clearly is a compatibility issue (my guess is it is between the MB or CPU and the 5700 - the CPU mainly because it works if I lower the multiplier, the fsb stays the same).
 

·
Registered
Joined
·
2,171 Posts
So with the other video card you could return the CPU to 2.1GHz and everything worked properly?

When you increase the CPU multiplier but keep the FSB the same you have the following changes:
1) The power consumption (especially on the 12 volt rail) increases.
2) The CPU gets hotter.
3) The northbridge gets a little hotter. It's running at the same clock rate but it still has to handle more memory accesses in a given time.
4) The communications between the northbridge and the video card GPU are at the same clock rate but there are more of them in a given time so the northbridge and GPU have to work harder.

The first thing I'd try is playing with the AGP port to see if it's causing the problem. Some video card/motherboard combos are pickier than others. I'd go into the BIOS and try disabling fast writes and limit AGP speed to 4X. That probably won't fix it since you've had freezes in the BIOS but it's easy enough to find out. The other thing you could do is raise the AGP signaling voltage. Your BIOS can set it to 1.5 (the default), 1.6, or 1.7. Raising that voltage can improve the stability of the northbridge-GPU communications. 1.6 volts is reasonably safe but I wouldn't leave it at 1.7 volts if it doesn't fix the problem.

If your problems don't happen when the machine is stone cold then you might have some kind of overheating problem. You can tell by running the case with the side off and aiming a deskfan at its innards to see if it improves.

You might also have some kind of power supply problem either in the PSU itself or in one of the voltages generated on the motherboard. When you increase the clock rate of the CPU, the 12 volt rail tends to get noisier. That may affect some video cards more than others. If your problem has gotten worse over time then it can be a symptom of power supply capacitors drying out which makes the supply rails noisier. Unfortunately about all you can do is look for leaking capacitors on the motherboard (see here) and try swapping in another power supply in case it's the PSU. The other video card you tried may be lower power than the 5700 or just less sensitive to noisy supply lines.
 

·
Registered
Joined
·
8 Posts
Discussion Starter · #5 ·
The other video card is a Inno3D Tornado Geforce4 MX 440-SE. I've been running it for two days and I have no problems (using the default 12.5 multiplier). True, it is a lower consumption card, so the noisier railings might not affect it.
Also, I would like to mention that my CPU kinda stays at 50 Celsius, with the case open, idle, and rises up to 70, with the case closed, full load. Most people told me that this is normal. I also think I need better cooling (my current cooler is Titan CU5TB).

Regarding my attempts to fix the problem from BIOS on the FX5700, I tried disabling AGP8X support, FW, lowering the AGP speed to 50MHz, nothing worked. The only thing I didn't do is change the AGP Voltage (which defaults to 1.5).

I will try to do that soon, or try to get a high-end card to test (GF6+).

Also, would it be good to upgrade my FX to a 6600/TD/128? (especially since I can upgrade it with 10-15$ difference or less, and supposing I fix my problems)
 

·
Registered
Joined
·
2,171 Posts
Also, I would like to mention that my CPU kinda stays at 50 Celsius, with the case open, idle, and rises up to 70, with the case closed, full load.
50 to 70 isn't unusual but the load temp is a bit high (assuming it's being measured properly). 70 shouldn't cause problems though.

Also, would it be good to upgrade my FX to a 6600/TD/128? (especially since I can upgrade it with 10-15$ difference or less, and supposing I fix my problems)
I'd definitely upgrade from an FX5700 to a 6600 for that price. Both cards have about the same memory bandwidth but the 6600 is a much faster card at running pixel shaders. That makes it quite a lot more useful for future games than an FX5700.


Given that the other video card is a lower power video card than the FX5700 you should probably see if you might consider swapping in another power supply. Your problem could be as simple as overloading the 12 volt rail. You can check the 12 volt rail in the BIOS to make sure it's not too low but the only way to be 100% sure it's not the power supply is to swap in another one.
 

·
Registered
Joined
·
8 Posts
Discussion Starter · #7 ·
How do I know if the voltage rail is too low or too high? What (normal) values should my BIOS display?
 

·
Registered
Joined
·
2,171 Posts
The voltage rails are supposed to be plus or minus 5% except for 3.3 volts which is plus or minus 4%. The only problem with using software to check the voltages is that many motherboards aren't all that accurate at reading the voltages. You'll get more accurate results by reading them with a voltmeter but it's still worth checking the BIOS to see what voltages you have. If the 12 volt rail drops (say .2 volts or more) when your video card kicks into 3D mode then you may have power supply problems. Or if the 12 volt rail reads lower when your CPU is running at full speed then it does when running at the slower speed then that's also a sign of potential problems. There's no way to be 100% sure about power supply problems without swapping in another one but you can sometimes see you're close to your power limit by closely watching the supply lines. The 12 volt rail is the one you should watch because that's where your CPU and high-power video cards get most of their power. Unfortunately there's no good way to see noise on the voltage rails without using an oscilloscope.
 

·
Registered
Joined
·
8 Posts
Discussion Starter · #9 ·
Well, after months of having this problem (and being too lazy to go to warranty), I finally fixed it. After I tried upping the voltage to 1.7 with no success, I was about to pack the computer and send it to warranty (I hate doing that, I don't like the people there). Finally, I decided to investigate one more thing. You told me it could very well be the PSU with the problems, so I thought I'd look to the pin holes in the PSU-to-Mobo cable jack. Seeing everything is in order, I put it back in place. I opened the computer and there, no more problems! No freeze, no artifacts, nothing! I guess the jack wasn't properly inserted in.

Anyway I hope it stays this way, also wanna thank you (UncleMacro) a lot, couldn't have made it without ya :)
 

·
Registered
Joined
·
2,171 Posts
It's good to hear that it's working. Yea, sometimes power supply problems are actually mechanical problems with the connector. They can get a little oxidized or loose and limit the current that it can supply. It's not that common but it happens. I guess I'll add removing and reconnecting the PSU connector to the standard "maybe you have power supply problems" things-to-try list.
 

·
Registered
Joined
·
8 Posts
Discussion Starter · #11 ·
Well the problem came back after a little more than a few hours of 3dmark. I restarted my computer and it again froze when it reached the windows logo screen, and the only way to fix it was to replug the power connector. Thus I have to do this 3-4 times a day if I'm playing 3D games. The worst thing is I get really cranky when it happens when you least expect it.

I documented myself a bit on the web, and found out I should really change my PSU (it's a generic psu, no-name brand, costed about 15$). I found out that low generic PSUs have many problems (overexaggerated power consumption, unreallistic specification - mine says 400W -, high noise in the 12V rail - which could be my problem -, etc). Besides, my 3.3V+5V is < than 160W max power, which I also read on the web that it is quite bad. I am now looking forward to buy a new brand-name PSU, although on a budget score. Some retail shops here sell Apex ALLIED PSUs, which are the cheapest from the brand-named PSUs (others are ThermalTake, Chieftec, Vantec, Aerocool and Silverstone), so I was wondering if it was the way to go (a 350W is about 25$). Also, how much power would I need? Should I go for 400 or is 350W good enough for my specs? (considering that future upgrades are not very demanding - at most another 512 ram and a GF6600 -)
 

·
Registered
Joined
·
2,171 Posts
You can use a power calculator to figure out about how much power you need. Personally I wouldn't use a power supply which cost less than $60 but then I'm a bit paranoid. Add about 30% to 50% to the number returned by that power calculator depending on how safe you'd like to be. Many power supplies don't deliver their claimed ratings. A power supply isn't a good place to skimp. Power supply problems can be very difficult to trace down. It's still possible that something else is causing your problems though. If you have access to another computer it might be worth your time to borrow its power supply and make sure your problems go away. I'd replace a $15 power supply anyway for peace of mind but if you're short on cash, a little more testing might be in order to make 100% sure it's the power supply.
 
1 - 12 of 12 Posts
Status
Not open for further replies.
Top