x1900xtx: rebooting issue resolved?

Hi all, first time posting here.

First of all I'd like to thank all the posters here for their informed and considered input - I wasn't aware of ATI Tool (I've been using ATITrayTools), and this little utility seems to have fixed a major problem I was having with my new Powercolor x1900xtx, namely sudden reboots everytime I ran a 3d game for more than a few minutes (Doom3).

My specs that I'm testing with are:

A64 3700+ San Diego
ASRock Dual939-SATA2
2gb Patriot PC3700 2-3-2-5
Antec NeoHE 550W
Powercolor X1900XTX
1 WD IDE HD
Lite On burner
Windows 2000 w/current SP
No OC'ing

Initially I thought this was a power supply problem; I originally ordered this card along with an Antec NeoHE 500W, since I thought it would be adequate according to stated requirements for the xtx, started having the reboot issue (random reboots from desktop, not even running a game!), and replaced it with a NeoHE 550W since it's one of the Crossfire-certified PSUs ATI lists on their site. The card is alone on it's own 12V rail (this PSU has 3 12V rails @ 18A per). Then I still had this problem after installing Doom3 for testing, but only when the game ran for a few minutes - no more random reboots from the Windows desktop apparently. I read a thread here about heat issues with this card, installed ATI Tool and manually adjusted the fan speed, and voila! I've been able to run Doom3 without any reboot issues (although the game occasionally locks up).

Several posters have mentioned this card seems to have a problem with the automatic fan control, and I would concur with that. It doesn't seem to function correctly using only Catalyst 6.3's and CCC, a tweaking utility like ATI Tool seems obligatory to make it work properly. ATI Tray Tools has a similar fan control in the overclocking options, but didn't seem to work as well as ATI Tool since I was still getting reboots using that utility.

One question I do have at this point: has anyone tried the Thermaltake Tide Water All-In-One VGA liquid cooling module? One post I read was that it was a great idea "implemented badly", but he didn't elaborate on his rationale for that observation. This card obviously requires efficient cooling, and while the stock cooler seems capable of this there is the notorious noise issue that tends to grate on the nerves. I'm not that crazy about the Vantec VF700/900 since these don't direct the airflow out the back of the case like the stock cooler does.

Any suggestions/recommendations? The case this setup will occupy is a full tower with plenty of fans and airflow, and I'm not planning on overclocking any aspect of it.

TIA for any input.

So what your saying is with the stock drivers and ccc the card would overheat and the PC would reboot.

I have to see it that way because you said setting the fan manually cured the reboot problems.

Well, if this is all the case, it would sound like you have a bad card, because the card is not supposed to overheat with stock fan settings.

I'd look into RMA.

BBA said:

So what your saying is with the stock drivers and ccc the card would overheat and the PC would reboot.

I have to see it that way because you said setting the fan manually cured the reboot problems.

Well, if this is all the case, it would sound like you have a bad card, because the card is not supposed to overheat with stock fan settings.

I'd look into RMA.

Click to expand...

Thanks for the reply.

You may be right, since with further testing I've been getting lockups and BSODs in addition to continued reboots, and that after I set the fan speed to run at 100% constantly.

Worked O.K. for a while, and I've been trying to decide if I just need another PSU that has higher amperage on the 12V (like the Silverstone ST56ZF - 560W and 38 amps) or to RMA the damn card. Probably I'll RMA the card for replacement and the Antec NeoHE 550W for the Silverstone just to make sure I've got a PSU with the necessary amperage, as I've been reading some negative things about the NeoHE series. I've noticed that the fan on the card isn't spinning up under heavy load without manually setting it in ATI Tool like it's supposed to, so off it goes.

What a pisser...more shipping costs.

I'd still like to hear opinions on the Thermaltake All-In-One if anyone knows anything about it.

If you were using ATi Tool, make sure you set the voltages properly. I used my XTX for a few months at lower-than-stock voltages for a fair while, but is now unstable at the correct voltages. Either the VRAM can tend to become voltage starved at 2.1 volts, or the voltage control in ATi Tool simply isn't accurate (which it isn't).

Set VGPU voltage to 1.425 or 1.434 volts. Doing so will result in 1.45 volts, which is the correct VGPU voltage for the XTX. Set MVDDC and MVDDQ to +2.1 volts and experiment. My XTX's memory is unstable at the default 2.086 volts, and really require about 2.15 volts. Your card shouldn't be overheating at 60/65% fan speed. Download and configure RivaTuner for use with X1900 cards and record/log the temperature monitor. GPU temperature should (ideally) remain under 75C, while VREG temperatures can occasionally exceed 75C. If all temperatures are inline and you've configured the voltages in ATi Tool properly, and the card still doesn't work, you should absolutely RMA. I'd refund, myself, if that would be possible.

These bootings are known issue with your Asrock mobo and Antec HE PSUs. I´d try with another PSU before RMA your graphic card.

Also there was booting up issues with that mobo before 1.7 bios. Before this bios the machine would boot up/BSOD when visiting a www site with flash animations on it. This could be fixed by setting memory timing from 1T -> 2T. But again 1.7 bios fixed these issues and now 1T works fine, atleast for me.

Thanks for the info guys.

I'm replacing the PSU and the card both. The Silverstone is a better match per specs (and which hopefully won't present any problems with my board), and since I have a thirty day window to return the card to Newegg, and I still have questions about whether it's functioning correctly, to be on the safe side I'm returning it for replacement. I've talked to Powercolor on the phone once already, and frankly I don't want to have to deal with them again unless absolutely necessary. Newegg has given me a $25 credit to cover the shipping costs I'm incurring, and waived the restocking fee for the Antec PSU so that's very cool...thanks Newegg. I suppose that's why in some cases it's better to handle RMA's over the phone rather than online.

Re: ASRock booting issues, wasn't aware of issues with this board with the HE series, glad to know about that. My board came with the BIOS that addressed the cold boot issue, so no worries there. Also glad to get the info about voltage settings in ATI Tool, thanks.

18A is not enough on 12V if you are running off of only 1 of the 3 rails. I believe the X1900xtx uses up to 22A @ load. I have never had a problem with Cat 6.3 and CCC, and I used ATI's fan service without issue. As mentioned before, maybe your card overheated and exhibited issues, but with the speed that the reboot happened that doesn't seem to be the case. This sounds exactly like a PSU issue. I actually had a similar issue with my OCZ 520 Powerstream when I used the dual molex->single PCI-e adapter and plugged in only 1 molex connector from the PSU. I had frequent reboots in 3d, and occasional reboots just on Windows idle. When I switched to the dedicated PCI-E connector whic the powerstream has 2 of, problems all went away. I also tweaked my 12v rail to 12.28v (never drops below 12v on load), and the 3.3v to 3.31 (was dropping like .07 below 3.3 before).

pakotlar said:

18A is not enough on 12V if you are running off of only 1 of the 3 rails. I believe the X1900xtx uses up to 22A @ load. I have never had a problem with Cat 6.3 and CCC, and I used ATI's fan service without issue. As mentioned before, maybe your card overheated and exhibited issues, but with the speed that the reboot happened that doesn't seem to be the case. This sounds exactly like a PSU issue. I actually had a similar issue with my OCZ 520 Powerstream when I used the dual molex->single PCI-e adapter and plugged in only 1 molex connector from the PSU. I had frequent reboots in 3d, and occasional reboots just on Windows idle. When I switched to the dedicated PCI-E connector whic the powerstream has 2 of, problems all went away. I also tweaked my 12v rail to 12.28v (never drops below 12v on load), and the 3.3v to 3.31 (was dropping like .07 below 3.3 before).

Click to expand...

I think it was a combination of the fan not running correctly AND an insufficient PSU. Despite the stated minimum spec of 30a on a 12v rail, the Antec NeoHE 550W I was using is on ATI's official Crossfire-certified PSU list for the x1900 series cards. Wonder why. What did they do, run the damn thing for 10 minutes and declare it "certified"??? When I spoke to an ATI tech, he said that PSU should power the card fine...the Antec guy I talked to said to make sure the card was on its own 12v rail, which I did, and which the Antec 500 and 550 each have 3 of. (17a and 18a respectively). Antec doesn't even make a PSU that's rated at 30a on a single 12v rail. The Silverstone I'm getting is a single 12v rated at 38a, so with that and the replacement card (which hopefully nothing is wrong with out of the box), I imagine the previous problems will disappear (knock on wood).

I've also read of people that had no problems powering this card with an Antec Truepower II 550W, and that one isn't up to the 30a spec either.

ATI could have been clearer on the specs. They state the requirements, yet a PSU that doesn't meet them is on their certified list...keeping in mind Crossfire means two of these cards drawing power from a single PSU... and their tech support doesn't appear entirely informed about the actual power requirements of this card, at least the guy I spoke to wasn't. Powercolor tech support was just as misinformed, not to mention practically unintelligible.

batai37 said:

I think it was a combination of the fan not running correctly AND an insufficient PSU. Despite the stated minimum spec of 30a on a 12v rail, the Antec NeoHE 550W I was using is on ATI's official Crossfire-certified PSU list for the x1900 series cards. Wonder why. What did they do, run the damn thing for 10 minutes and declare it "certified"??? When I spoke to an ATI tech, he said that PSU should power the card fine...the Antec guy I talked to said to make sure the card was on its own 12v rail, which I did, and which the Antec 500 and 550 each have 3 of. (17a and 18a respectively). Antec doesn't even make a PSU that's rated at 30a on a single 12v rail. The Silverstone I'm getting is a single 12v rated at 38a, so with that and the replacement card (which hopefully nothing is wrong with out of the box), I imagine the previous problems will disappear (knock on wood).

I've also read of people that had no problems powering this card with an Antec Truepower II 550W, and that one isn't up to the 30a spec either.

ATI could have been clearer on the specs. They state the requirements, yet a PSU that doesn't meet them is on their certified list...keeping in mind Crossfire means two of these cards drawing power from a single PSU... and their tech support doesn't appear entirely informed about the actual power requirements of this card, at least the guy I spoke to wasn't. Powercolor tech support was just as misinformed, not to mention practically unintelligible.

Click to expand...

Lol, yeah that seems very misleading. Antec is a great PSU brand, but the amperage rating on the rails seem too low. Wonder why ATI recommended them when their own specs recommend a much higher amp rating on 12v. I imagine that the only way that you would be ok is to use the dual-molex adapter and run off of 2 of the 3 rails. Although I have also hear problems with dual/triple rail PSU's in terms of stability vs. single massive rails. Dunno tho, the only thing I can tell you with absolute certainty is that I'm not an electrical engineer lol. For what it is worth the x1800xt was the first PC component that has caused a PSU related issue for me. With Quad-SLI I can only imagine what a pain in the ass it must be to supply steady, low-noise current.