Gfx card - Return it or not?

I bought a 2nd hand GeCube Radeon X1950 Pro AGP card from fleabay on Friday (from a powerseller). I received it Saturday. It looks brand new apart from a tiny bit of dust on the fan blades. I installed it into my system and connected 2 molex connectors from two seperate lines from my PSU. Fired it up and was shocked to see a glitchy bios full of artifacts, but I left it on and let it boot up. Resolution was reset to 640 x 480 so it was hard to see especially with a screen full of artifacts (short horizontal lines mostly all over the screen) but I had already prepared the Radeon drivers and I could see well enough to load them up so I did that after installing them I rebooted. Upon rebooting, the screen was free of artifacts in the BIOS and in Windows so all was well.

So I booted up Trackmania Nations Forever and put all settings on high. Loaded the game and started racing away. Fantastic framerates, beautiful detail... then after approx 15 mins the system randomly froze mid race. The sound that was playing started looping then the screen was black and the "no signal" message appeared on screen. I had to hard reboot.

Now after about 24 hours of extensive testing, I've found that this happens in pretty much every single game I play and the crashing is completely random. It might be 15 seconds into the game and other times it might go for as long as an hour (even in the same game). I've tried different resolutions, detail levels, underclocking the card with ATI Tool, upping the voltage to the AGP slot, different drivers (Omega ones currently installed)... can't seem to shift the problem.

I've read in other forums that it might be due my PSU not being up to the job. I have a 400W Enermax Liberty which was supposed to be able to hand SLI Gfx card setups... I don't know what ampage the rails have but the manual for the X1950 Pro states that it recommends a 450W PSU with a minimum of 30A supplied to the card... Would my PSU being underpowered lead to this sort of instability?

Or maybe the card is overheating? The fan is certainly very loud. However, I've underclocked it by as much as 300Mhz in ATI Tool and I've run the artifact scanner for upto 30mins with no results. It's worth noting that I've also not seen any signs of glitches in games at all even during a crash before I have to reboot it.

I'm loath to send the card back as I'm pleased as punch with the performance and don't want to be without a card for ages... especially if it can be fixed by a PSU upgrade. Your thoughts? (will be much appreciated)

PSU shouldn't be an issue. I'd go with the obvious issue being the card itself. Are the POST screens completely free of artefacts now?

Yes, as is everything else. No artifacts anywhere now.

I forgot to mention. I was using a GeForce2 MX 440 AGP card in the system before I installed the X1950 Pro. Wasn't having any problems with that (although it barely ran any games anyway so I didn't push it at all)

It would be worthwhile testing with another PSU first off, X1950 cards are power hungry beasts, got one myself!

Mind you artifacts are never a good sign, could be something wrong with the video memory.

I would definately start by upping the PSU to a 500 watt or above though, just for future proofing the PC, more of an investment regardless. As for the card possibly try it in another machine with a higher watt PSU and see what results that brings.

*Sounds* like your card may have been overheating right enough, as underclocking it helped. Perhaps someone could recommend a temperature monitoring tool?

A 400W Enermax Liberty should easily be capable of powering any X1950, let alone the least greedy Pro version. I'm not a fan of Enermax after the below-the-belt tactics they used on some budget boxes a couple of year back, but the Liberty's are supposed to be solid.

If you could test another PSU or similar card it would help a lot, but I'd be ruling out the PSU at the moment.

When graphics memory fails, in my experience, the system doesn't lock up, but merely show the artefacts you mentioned earlier. Hard locks are usually the fault of an overheating or faulty GPU/other component on the card.

Catalyst Control Center tells you the temperature, so perhaps open it up, open a game, run around for 10 minutes and see what the temperature is immediately on exiting. Anything below 80c and you can probably rule out temperature as the cause. Rivatuner can record temperature over a period of time too.

underclocking didn't really make a difference. It still crashes just as often. ATI Tool doesn't find any artifacts even left at stock clock rates...

I'd certainly be insterested in checking the temps, I haven't looked properly but I don't think there are any with the drivers or tools that I've downloaded so far.

I wish I could just try a higher wattage PSU but unfortunately don't know anyone near me who'll try it - less taking to a pc shop and paying a fee. I guess it's a small price to pay for peace of mind...

You may have missed it in my post above as you were posting at the same time, but I mentioned a couple of means of checking temperature. :)

High wattage isn't the issue, it's amps which are important for graphics cards. A 330W Seasonic should be able to run the spec in your "My System" with relative comfort.

I can't really recommend it given what's been said, but if you do purchase another PSU, for the love of God don't get a cheapy EZCool or something. Get a renowned name. Seasonic or Corsair. You seem clocked up on the importance of PSUs (given you have an Enermax already), but it's something that can never be said enough. :)

Thanks, yes did miss that post.

I hear what you're saying, I didn't really intend to buy a purely higher wattage PSU, it's just that I tend to find that the higher the wattage the higher the ampage on the 12v rails (although I'm no expert, far from it!). I certainly wouldn't fork out on a cheapo one. As you rightly said, definately one of the most important factors in a PC.

Well I'll certainly look at temps before I part with any money. I will probably have the card tested properly by my local pc shop to rule out my setup in the event that temp checking doesn't answer the problem.

But if I do find extremely high temps (in excess of 80 degrees), should I invest in an arctic cooler or some such aftermarket cooler? I had good results with this on my last card (Radeon 9800 pro) and they are near silent which is always nice compared to the current 747 engine which is currently on the X1950 Pro! Or would it be wise to return it since it may now be damaged?

Also, just to clarify, I have not had any artifacts at all since 2nd boot with the card in (after rebooting with the drivers installed). I put the artifacting down to the fact that I'd gone from a GeForce 2MX 440 to an 8x AGP ATI card - although a bit unusual for it to glitch during POST, I figured it wasn't the card being faulty since it's never done it since...

Heres a mad one for you which is easy enough to try.

You mentioned that you are still on AGP. Go into the system bios on startup and manually change the AGP to 4x

When I still had my 9600 Pro a few years back if I had the board set to 8x AGP it would lock up the system when playing games. Has something to do with the ATI chips not liking certain chipsets at the time. Might be something similar with this one.

Also worth doing is getting Driver Sweeper or Cleaner from Guru3D to remove all the last remaining bits of the Nvidia drivers.

Thanks, I'll give it a try.

Just as a bit of reference my Radeon X1950 Pro card (pcie) under load hits around 65 degrees celcius with stock cooler. You could try downloading Riva Tuner (Also from Guru3D) and up the fan speed to about 80%

Definetly a overheating problem had the same exact issue with my old 9800 pro years back when playing games on max settings it would just buckle but play fine at lower in the end it blew the card a return is highly advisable unless you fancy putting a custom fan on the card yourself. If it was a PSU it wouldn't carry on running it would probably just BSOD and other hardware would be having issues. If it was wattage of PSU it wouldn't even start.

This card sounds like its overheating.
Its massively unlikely to be the PSU. It has 20+ amps IIRC on each 12V. Thats more than enough for that card.

Return it....its faulty. If its a Powerseller then they have to accept the return under DSR as they must be registered as a business now to be a powerseller.

Right, I've contacted the seller and he's agreed a refund.

I had a play with it last night and couldn't for the life of me find anyway of monitoring the temps on it. Catalyst control center doesn't seem to have it and ATI Tool doesn't have it listed either. It says something about fan and memory timings changes not supported by this chipset... oh well, it's going back anyway :-(

Got my refund and managed to find an absolute bargain replacement: an ATI Radeon X1950Pro 512mb AGP - Basically the same card but with double memory and slightly higher clocks. From the reviews I've read this card fairs a bit better and it's fan is quieter, controllable and temp monitoring works on it... managed to get it for £39.99 from flea bay - well chuffed.

Now have the card installed and it's working perfectly, no crashing, artifacts or anything.

Thank you all very much for you advice - well pleased...