Shagittarius wrote:

Finally got a 6800 Ultra AGP for a good price. Tried to hook it up and it warns me it doesn't have enough power. I have an Antec True550W I know these can run on quality 350w supplies and the Antec is supposed to be a good PSU.

My question is what is more likely, that the PSU is old and can't maintain voltage across the 2 molex connectors, or that one of the molex connectors on the card is bad?

Also which is the rail Im most interested in when powering this card?

I would check the connectors to the GPU first, clean if need be.

As for which voltage rail, most GPU's require a lot of power from the +12vdc rail.

Didn't think that a 6800 AGP or PCIe required a whole lot of power.

Check the back of the card for anything that has broken off, there is a small diode and if it breaks the card won't have power.

I have seen this error as a flaky one-off under Windows XP when the OS was in a bad state. See if it reproduces.

Ensure that the two Molex connectors are fed by separate wires from the PSU with no significant competing loads on them. Even if you have a single-rail PSU, it can't hurt. If it still reproduces, try another PSU or check out related threads in Nvidia forums (nvnews, GeForce forums).

As a probably unrelated issue, if you haven't done so already, it's a good idea to remove the cover from the heat sink (5 or so microscopic screws) and blow out the dust bunnies. Blowing air into the fan isn't sufficient once it's really choked up.

MatureTech wrote:

I have seen this error as a flaky one-off under Windows XP when the OS was in a bad state. See if it reproduces.

Ensure that the two Molex connectors are fed by separate wires from the PSU with no significant competing loads on them. Even if you have a single-rail PSU, it can't hurt. If it still reproduces, try another PSU or check out related threads in Nvidia forums (nvnews, GeForce forums).

As a probably unrelated issue, if you haven't done so already, it's a good idea to remove the cover from the heat sink (5 or so microscopic screws) and blow out the dust bunnies. Blowing air into the fan isn't sufficient once it's really choked up.

I actually took a little time to look into the PSU but only through the bios, however the bios is reporting that the 12v rail is only delivering 10.89V...So I ordered another PSU which Im waiting for now. The 5900 Ultra didnt seem to have a problem with this PSU but perhaps the 6800 Ultra actually requires more power.

Also I tried every different config of wires without anything on them and consistently got the same message. Once in a while the board itself would beep at startup, but usually it remained silent.

The PSU should be here any day and I'll report back then.

FWIW, 6800 Ultra power estimates

It is the same core as 6800 GT run faster and hotter (and using more power).

That voltage is way too low for an Ultra, get a better psu and the issue should be gone. Anything below 11v on the 12v source you can run into a lot of problems including instability or worse.

nforce4max wrote:

That voltage is way too low for an Ultra, get a better psu and the issue should be gone. Anything below 11v on the 12v source you can run into a lot of problems including instability or worse.

Just got the new PSU which tested working with a tester, and this one gives me the same warning message that the 6800 Ultra is not getting enough power, though it appears to underclock it less. Could I have gotten a second bad PSU? The 12V rail is reported at 11.68 on this one.

As a side note I'm going to drop this project anyways...Since the 6800 seemed to be downclocked but still a tiny bit faster than the 5900 Ultra I was going to use it but then I found out that Nvidia 45.23 drivers don't recognize the 6800 Ultra, so there goes some important compatibility for this machine. Looks like I'm sticking with the 5900.

Shagittarius wrote:

nforce4max wrote:

That voltage is way too low for an Ultra, get a better psu and the issue should be gone. Anything below 11v on the 12v source you can run into a lot of problems including instability or worse.

Just got the new PSU which tested working with a tester, and this one gives me the same warning message that the 6800 Ultra is not getting enough power, though it appears to underclock it less. Could I have gotten a second bad PSU? The 12V rail is reported at 11.68 on this one.

As a side note I'm going to drop this project anyways...Since the 6800 seemed to be downclocked but still a tiny bit faster than the 5900 Ultra I was going to use it but then I found out that Nvidia 45.23 drivers don't recognize the 6800 Ultra, so there goes some important compatibility for this machine. Looks like I'm sticking with the 5900.

It is good that you got a new unit but that voltage however much better is still on the low side. Check the card again for anything missing, there is a diode or resistor that when missing the molex connectors on the card become useless. I had this happen with a 7900GTX Duo (rare) and all it would do after is beep. If anything else it could be the card being in old age.

nforce4max wrote:

Shagittarius wrote:

nforce4max wrote:

That voltage is way too low for an Ultra, get a better psu and the issue should be gone. Anything below 11v on the 12v source you can run into a lot of problems including instability or worse.

Just got the new PSU which tested working with a tester, and this one gives me the same warning message that the 6800 Ultra is not getting enough power, though it appears to underclock it less. Could I have gotten a second bad PSU? The 12V rail is reported at 11.68 on this one.

As a side note I'm going to drop this project anyways...Since the 6800 seemed to be downclocked but still a tiny bit faster than the 5900 Ultra I was going to use it but then I found out that Nvidia 45.23 drivers don't recognize the 6800 Ultra, so there goes some important compatibility for this machine. Looks like I'm sticking with the 5900.

It is good that you got a new unit but that voltage however much better is still on the low side. Check the card again for anything missing, there is a diode or resistor that when missing the molex connectors on the card become useless. I had this happen with a 7900GTX Duo (rare) and all it would do after is beep. If anything else it could be the card being in old age.

I checked the card and I didnt see anything that looked like it was missing, I saw what I believe to be a single diode with a trace leading to the middle pin of the lower molex connector. It's funny that the new PSU allowed it to clock itself higher than the old PSU, so maybe it's still the PSU, but like I said I'm giving up because I really wanted a card that works with the 45.23 drivers. Thanks for all your help though.

Shame on me for not asking about the drivers. Although it does sound like the first PSU was particularly weak, the use of incompatible drivers now seems the most likely root cause of the power warning. I have never used the Detonator drivers for anything. Final version to support a 6-series GPU and W98 is 81.98 (after which official support for W98 was dropped).

MatureTech wrote:

Shame on me for not asking about the drivers. Although it does sound like the first PSU was particularly weak, the use of incompatible drivers now seems the most likely root cause of the power warning. I have never used the Detonator drivers for anything. Final version to support a 6-series GPU and W98 is 81.98 (after which official support for W98 was dropped).

Yeah the first set I tired was 81.98 and those gave me the warning, then I went back and tried a couple older drivers including the detonator drivers which are incompatible.

The detonator drivers are important some games have texture problems if you go beyond 45.23, No One Lives Forever for example, all of the characters are blank white and most of the environment textures are missing or distorted at least as far as Windows 98 drivers go.