New build, and some weird issues I'm trying to figure out...

[Slaugh]

Hi there,

I just finished rebuilding my system which was watercooled in the past with a Zalman LQ-1000 (more details in this thread). The only changes I made were the following:

• 1) Switched to a Fractal Define R4 case, moved the 2 Fractal stock fans to the front and set them at 5V, the installed a NoiseBlocker NB-BlackSilentPro PK-1 140mm at the back, which is connected to a motherboard fan header @ 12V.
• 2) Put a Prolimatech Megahalems on my i7 870 and attached a NoiseBlocker NB-Multiframe M12-S1, connected to the CPU fan header @ 12V.
• 3) Put a brand new Seasonic 520W fanless power supply in the Fractal Define R4 (the bottom intake fan directly blows on it so it stays cool)
• 4) Removed all the drive cages and put my existing hard drive (Western Digital Black 1TB) in my existing NoVibes III to kill its vibration noise. No DVD-ROM is installed, I have an external USB one that I can use if needed. The HDD is installed in the first 5.25 bay and never reaches 40C in this case.
• 5) Installed a brand new Asus GTX 670 DirectCU II (I had a Radeon HD 5850 before)
• 6) Completely wiped all existing ATI drivers, and installed the latest one from nVidia. Also installed the latest release of Asus GPUTweak. I also use Speccy and DTEMP to monitor the temperatures. Speedfan is also installed in case I need it, but not launched at startup.

Now here are the issues I have:



1) Strange GPU crash, then booting troubles

The first day, I spent around 3 hours without any trouble... I installed the latest nVidia drivers, cleaned the ATI ones, installed GPU Tweak, and launched Unigine Heaven to test the new video card, and did other simple tasks with the system. There was no problem so far, the temperatures were very good (HDD, CPU, chipset and GPU), and everything was working fine...

The next day, I retried Unigine Heaven again for less than 15 minutes, the GPU never went above 58C during this test, and the CPU was hovering around 45C. The hard drive mounted in the NoVibes III was around 28C. I then tried to launch Steam and a few seconds later, the screen went totally black, my monitor went to sleep and I had to do a hard reset.

When I restarted Windows, the animated logo on the "Starting Windows" screen froze, and I had to press the power button for 4 seconds to turn off the system. I retried booting Windows a second time, and it froze again, but at a different stage on the same screen... Then I tried to launch the Windows startup repair tool (F8), it properly loaded the files, but it froze when the green progress bar appeared...

I first thought that the hard drive could be the culprit, and I launched UBCD and made a quick test with WD Data Liveguard utility. No error was reported... I rebooted the system and this time, Windows was starting without any trouble. I checked my minidump folder and saw that a BSOD was generated exactly when I got the black screen (In my case, Windows is configured to not reboot in case of system crashes). I analysed the BSOD and it was caused by the nVidia driver (VIDEO_TDR_ERROR). I left the system turned off for a few hours because I had no more time to investigate.

In the evening, I tried to start and reboot Windows a few times in a row without any problem, then I made a full media scan with WD Data Liveguard utility, and no error was reported. I also checked all S.M.A.R.T. attributes and they're all good. So I'm pretty sure it's not the hard drive right now.

Could it be a defective video card? Or is the Seasonic 520W the culprit? It could also be a driver issue, but I'm not so sure. The drivers are loaded and initialized when starting Windows, which could explain why the system was hanging at the "Starting Windows" screen (the animated logo never completed its full animation and the system completely froze). ...But no driver is loaded when you launch the startup recovery from the boot menu (F8), so I'm more thinking of a hardware issue... This video on Youtube shows exactly what was happening 2 times in a row earlier this day when I was trying to boot Windows.



2) GPU fan noise:

Most of the time, the GPU fans were at 10% speed and quiet, but when I stressed the GTX 670 with Furmark for a few minutes this evening, it reached 77C with fans at 54%. I noticed that the fans were making a strange noise, some kind of oscillating noise, like if one of the fan was unbalanced. Is this a normal issue with an Asus GTX 670 DirectCU II? The noise is not very loud, but it's annoying. Could the upper front fan (@ 5V) is causing too much turbulence?



3) Fractal fan noise:
At one time, one of the two fractal stock fans, installed on the front, started to rattle. I had to put the front fan at 12V, then lower them to 5V again to stop the rattling noise... I'll probably change these fans in the future. Does anybody had any problem with the stock fans?


I need some advice here before asking for a RMA, the only new components are the power supply and the video card... I'm having a hard time to figure out what's causing the crash... Both came from different places (power supply from Newegg and video card/case from NCIX)...


Thanks in advance.
Slaugh

EDIT: Oops... The BSOD was VIDEO_TDR_ERROR, not VIDEO_TDL_ERROR.

[edh]

I would not suspect the PSU. NVIDIA graphics cards with supplemental power connectors do test the power connection on startup. If the voltage is too low on a Windows system then when you get to the desktop you get a warning window and it suggest the power connector might be loose as well as downclocking the card. I have an older machine which does this on cold starts sometimes as the PSU voltage lines seem to take a little bit of time to come up to voltage when cold. If you leave the connector off completely then this message will be displayed but the system should not crash. Likewise if the voltages were too low for the motherboard it would most likely not start either.

It could be a case of something doesn't play well with another piece of hardware. If you have multiple 16x PCI-E sockets then you could try moving the graphics card to another slot and repeat.

Do you still have the old hardware that you replaced to test against?

[Slaugh]

It could be a case of something doesn't play well with another piece of hardware. If you have multiple 16x PCI-E sockets then you could try moving the graphics card to another slot and repeat.

Do you still have the old hardware that you replaced to test against?

I still have the Radeon, but it has a waterblock installed on it right now. The previous card was an Asus Radeon HD 5850 TOP DirectCU, and I still have the original cooler somewhere in the closet, but I'm not sure I got the original RAM/mosfet heatsinks and everything to put the card back to its original state. I bought that card 2 or 3 years ago, just before the HD 6000 serie was released, so it's been a while. The motherboard has 3 PCI-Express 2.x slots (the first two are x16 and the third one is x8). The GTX 670 uses PCI-Express v3 specifications, but it's fully backward compatible to PCI-Express v2, so I don't think it's an issue for me.

Yesterday in the evening, I was able to play a bit with the game "Worms Reloaded" (not very heavy on the GPU I confess), and I saw some curious micro stuttering (small pause during two graphic frames) during the game, which I've never encountered with the previous card. I did some search on the Internet about all the issues I got so far, and some had reported problems with the nVidia HDMI audio driver. I'm not using HDMI, only a dual-link DVI with my Samsung 2233RZ (22", 120 Hz, 1680x1050). The audio was perfect during the game. I'm using the PCI-E x1 audio card that comes with my Asus Maximus III Formula motheroard. I can try to disable the HDMI audio drivers, but I'm still skeptical. Like I said, the system froze while starting the startup recovery repair tool, where no nVidia drivers are loaded at all (only generic drivers for VGA, USB, PS/2, etc. are used).

An intermittent VIDEO_TDR_ERROR (which means that the GPU has crashed and was unable to recover) can be a PITA to figure out... It can be a faulty driver, a bad video card, an incompatible hardware, a failing PSU, a software-related issue, etc. I've checked the RAM with memtest, the HDD with Data Lifeguard, stressed the card with FurMark, inspected all capacitors, double-checked that all ATI drivers were gone, made several other tests to reproduce the initial crash, but the system never wanted to crash again. Since it's intermittent, It's pretty hard to pinpoint the source of the problem. I'm pretty sure it's not an overheating issue, the system was almost idle when the initial crash occured (I was only surfing on the Internet, and Steam, GPUTweak and Speccy were running in the background). The only thing I didn't do was to flash the BIOS with the latest one from the Asus website, which is something I normally do as a last resort... I have a feeling it's something else, probably the video card itself or a faulty Seasonic PSU. Putting the card in a different slot is like a shot in the dark, as the problem only last for a few minutes and never came back after... Hmmmm... Could it be my UPS which was acting up? It's an old Back-UPS 720VA...

I'm still debating if I should just exchange the card, or get a refund and put a good aftermarket cooler on my existing Radeon. I purchased the card on december 23th, but I had to wait about 2 weeks to receive it by mail, then I got sick for a few days... so I just rebuilt the system a few days ago and had trouble on day two. If I want to return it, I have about 3 or 4 days to get a RMA number from NCIX (30 days exchange/refund). I'll send a message to NCIX explaining all that... Right now, trying to figure out what's going on is driving me nuts, so I'll stop for the day!

Any clue about all this? Defective video card? defective/uneven PSU? bad UPS? Or something else...

Thanks in advance
Slaugh