My VDDCs don't match on 5870 CF setup

[kenkickr]

I was doing some benchmarking yesterday after getting my 5870's setup for crossfire and tried to do some OC'ing but with either MSI or Trixx the system would lockup/crash. Did the ULPS reg fix and started to do some OC'ing. Pushed them both up to 900 on the cores and left ram alone before doing some testing and within secounds I'd get a BSOD about the ati driver...keep in mind I OC'd my original @ 1Ghz core and 1.3Ghz on ram. Ran both cards at stock after resetting the Water block on the 2nd card again and noticed the VDDC's on the 2nd(new) card are kinda off of what the original card showed. I think something is up with the 2nd card but wanted to run this by you guys to see if anyone has had this before.

P.S. Also noticed one of the R23's are really loose. Couldn't this be the culprit!?

 

[Swamp Monster]

R23? what did you mean with that, SMD resistor on card?

 

[kenkickr]

It's back by the VRM's so I guess it's a resistor.

 

[Swamp Monster]

It could be the problem if it's "loose". Can you post a pic please? You can solder it back, if it is not damaged, or RMA the card.

 

[kenkickr]

I'm at work now but I think then I'm just going to RMA it. I hope I can since I bought it off a guy on Ebay. I thought the card + block was a hell of a deal...Now I know why!?

The craziest thing is I played Dead Space 2 for an hour and a half without issues last night. Once I OC though it won't last 10seconds.

 

[Swamp Monster]

One more questoin - is it small or big? I attached pics for comparison. Please post your pic, when you can anyway.
*edit* Which one is similar to yours?

 

[kenkickr]

The Magic R19 chip is similar in size.

 

[Swamp Monster]

Then it is coil, used to deliver power to GPU. If it is not soldered properly, then on bigger load it could loose contact and causing errors.
Is it "loose" like:
a) bad soldering, you can move soldering pins, or see crack in soldering contact
b) it's upper part is moving, but soldering contacts stand firmly?

Case B could not cause errors, because it is only casing for coil itself, If it is not firmly glued to coil, then it can move, but there is nothing wrong with that.

 

[kenkickr]

The black part is really loose but the solder points are fine.

 

[Swamp Monster]

Then it should not be causing problems. You need to look for something else on card, heating of different components(maybe this coil is slightly hotter than others), or it is software problems. But pic of that coil would still be helpful.
VDDC may be different, because it is decision of manufacturer (BIOS is just different), but I am not sure about that.

 

[kenkickr]

I've flashed both cards(Sapphire reference) to the same bios from the newer card. In the screen shot my old card is the top one and the bottom is the new one.

 

[Swamp Monster]

I guess if biose's is the same then it can be software- driver, something to do with crossfire. That can be easily checked What if you pull out other card and use only second? what GPU-Z will show in VDDC then? Also change placement of cards and check GPU-Z.

Just noticed in screenshot that GPU-Z shows very much info about your card. Even VDDC Current and temperature of every phase. I wish I had so much control of my card. Temperatures looks good and that is why I think this is software fault.

Maybe you accidentaly pencil-modded your second card? jokes aside, check if it was pencil modded (modded wrong) when you received it from ebay.

 

[kenkickr]

Really sucks to try each card cause I'm gonna go through water like crazy, already went through half a gallon, but I wanna be sure I send the card for a reason. I'll try some this weekend and post back.

 

[Swamp Monster]

It's me again Before taking cards out you can at first look in the second GPU bios what is 1.125V used for. Is it for 2d clock or maybe power saving or something. Then you will know at what bios setting your second card is stuck. Compare both bioses again - If that 1.125V cannot be found anywhere in bios, only then it is hardware fault.

 

[kenkickr]

When in 2d/power saving it sits @ 0.95v like the other card. When I turn on Furmark Multi-GPU then one goes to 1.16v and the other to 1.12v. It was doing this before I did the ULPS reg fix for crossfire.

 

[Swamp Monster]

I got that, but I mean't load bios in RBE and check the voltage table if that 1.125V exists in bios at all, or there is only 1.1625V. If it not exists, then it could be that second card was unsuccessfully pencil modded before and some graphite is still on some component.

 

[kenkickr]

When I received the card it looked like the owner was a heavy smoker so I grabbed the alcohol and gave it a bath before putting it in the system. I'll look again but I am going to try each card individually to see what they can do.

P.S. Thanks for being suggestive and helpful.

 

[kenkickr]

Update:

The 1.125 reading is actually coming from my original card and 1.165 is the new card. I did some more individual testing with both cards and the 2nd(new) card clearly is having problems...tons of electrical noise when stressing. Trying to setup an RMA right now. Thanks again for helping me out SwampM

 

[Swamp Monster]

Always happy to help! I guess I will be more careful too, when shopping @Ebay.