Problems with Radeon X600

I've got this problem with my new PC (don't wonder the lousy video adapter, I won the whole system in lottery...).

The problem is my X600 stops responding quite often, resulting in black screen and (after that) driver stating that the VPU has been recovered and reset. This has happened few times during normal desktop usage, but more often it occurs while playing.

I have tried Atitool to test the core and the memory. Core can be overclocked from 400MHz to over 500MHz without errors, but memory will have problems when overclocked as little as 5MHz (from 290 to 295MHz). At the default frequencies the card has performed fine when tested with Atitool, though I have only tested it for about 20 minutes.

This problem occurs whether or not the card is overclocked.

Could it be a driver related issue, or is some physical defect more likely? What to do?

System specs: P4 3.2GHz + Intel D915PGNL motherboard, 2 * 512MB DDR PC-400, Radeon X600 Pro 256MB, WXP Home SP2 (all updated), DX9.0c, Ati Catalyst 5.6 drivers

The memory having problems with such miniscule overfrequency suggests there might be a very marginal spec memory chip. Even though it tests OK at stock frequency, if it is very close to its limit it could be tipped over the edge by small PSU voltage fluctuations or temperature.

Do you know if the memory chips on the card have heatsinks attached? It could be worth physically checking the card to ensure coolers are fitted correctly, fan(s) operating correctly, and feeling memory chips with a fingertip to see how hot they are. (Touch the metal of the case to discharge static electricity before touching any components inside.) Temporarily playing a fan onto the video card could also be a check to see if it is temperature related, if the VPU reseting stops or reduces that would give a lead.

Also if the PSU is a generic type, its quality may be suspect, any opportunity to try a good quality PSU if this is the situation?

Thanks for the answer.

Quote:

---

Originally Posted by Platypus

Do you know if the memory chips on the card have heatsinks attached?

---

I took a peek inside, but I just thought I'd make sure, so I took two images, if you just care to have a look at them. I apologize for them being so blurry, but I guess they're clear enough for the purpose.
Above
Below

I thought Radeon X-series was PCI-based (while they have, or at least mine has that PCIe notation after the name), but doesn't that one sit in the AGP bus? (It's in the black socket on top of the white ones, which I assume are PCIs.)

So, the core wouldn't be so small that both it and the memory chips would be located under the fan? So there's no cooling for the memory chips. Or am I wrong?

I just wonder if the vendor can be held responsible. It's summer now, but not too hot, room temperature being perhaps 25 degrees Celsius. Does anyone have any experience with vendors or manufacturers supplying products with inadequate cooling?

Blurry, you're not joking! :)

Your memory chips don't have heatsinks, which is not unusual on a basic spec card. And it is a PCI-E (PCI Express) card - the black socket it is in is a 16x PCI-E, the two short black ones are 1x , and the four white slots are standard PCI.

But now I've had a chance to get back after work, :) , I do wonder if your card may be underclocked to compensate for a sub-spec memory chip. The standard ATI spec for the card seems to be 400MHz GPU/300MHz Memory, like the one, similar to yours, pictured here:

http://www.connect3d.com/products/pcie_x600pro.htm

290 MHz is an unusual frequency, although it's not uncommon for a 256MB card to have lower spec memory than the equivalent 128M model, to avoid too great an increase in cost. It should still be stable at it's rated speed though.

If this is the cause of your problem, you'd need to take it up with the provider of the system. Even if the system was won as a prize, if it has new components they should carry a warranty from their manufacturer. Although getting that warranty may be more involved than if it was purchased through a store.

Thanks again.

I've got one brief question yet. I've been running some other 3D apps (tests and games), some of them for several hours, and so far Grand Theft Auto San Andreas has been the only one to cause these crashes. Desktop crashing hasn't happened in days when running on default clock frequencies (although it happenes quite easily when the memory is just slightly overclocked). So could it be that it isn't memory chip related problem after all? Just odd that one certain game can cause the video adapter to crash (I'd figure it would be some software crashing if it were game/application related, but instead the adapter crashes, or so the driver says).

Interesting, personally I suspect drivers cover their own backsides and blame everyone but themselves... :)

If it seems to be down to one game causing the problem, some things to try could be:

Check game developer's site for patches in case it's a known problem.

Turn off game features one by one to see if one is provoking the fault.

It's hard to push a 256M card, maybe the game is the only one that pulls in enough texture & level data to start to use a weak memory chip if there is one. Try setting the "AGP Aperture" in the BIOS setup to the smallest setting available & see if this provokes more crashing.

Quote:

---

Originally Posted by Platypus

It's hard to push a 256M card, maybe the game is the only one that pulls in enough texture & level data to start to use a weak memory chip if there is one.

---

Could be, but this game is graphically very modest when compared to many others (ported from PS2).

Quote:

---

Try setting the "AGP Aperture" in the BIOS setup to the smallest setting available & see if this provokes more crashing.

---

What would be corresponding setting for PCI-E-card, as X600 belongs to such series? I think. I actually know nothing more about that. Or is there anything like that for PCI-E-cards?

Just thought I'd let you know, if anyone else happens to come across this problem, that it seems to be a driver/game related problem. I've seen other people, who have not experienced trouble with other software, reporting exactly same symptoms with Catalyst 5.6 and GTA San Andreas.

Its not unusual for "A" particular game and "A" particular set of drivers to have an issue
Pretty normal for that to happen

Quote:

---

Originally Posted by AScannerDarkly

What would be corresponding setting for PCI-E-card?

---

Aha, silly me... At the moment I can't actually answer that, I haven't yet looked into PCI-E cards' access to system memory. If the game is not data-intensive, the suggestion is probably redundant anyway.

Where once you had an agp gart (graphics address relocation table) , you now have a pci-e gart & all you changed was how it was joined up, but everything else is pretty much the same, principles wise.

So if PCI-E has a similar memory mapping capacity to AGP is there an equivalent aperture setting in the motherboard BIOS?