Bitcoin Forum

Hello everyone,
So First off I promise i've done everything in my power to fix this with out asking for help. but after a few days and a few cards I need to sleep so I figure I will post this up as I can just no longer go on..

Biostar ta890fxe motherboard
athlonx2 255
2gb samsung ramddr3
2 X MSI 6950 Twin Frozr II
650 watt psu

I have a 68ohm and 100ohm dummy plugs

Basically I cant get more than one gpu to hash without it crashing. Ive tried poclbm gui and command , and  rpc-opencl
Ive tried every combo of drivers i can think of but if someone knows a for sure winner (or something situation specific im missing maybe?) please let me know.
Tried crossfire and no crossfire both in x16 slots (2) and one x16 and one x8
I get a blue crash screen a lot, ati2dvag is the culprit crashing the thing.
I can get a picture to go through each card and get the display settings to recognize it but once i start hashing it will freeze or crash.
It seems to vary with different settings.
GPU Caps Viewer Closes when i try to run a demo on the second card.

I also get failed to load kernel in one of the programs i dont know if this is due to it or not.
It has been days so I cant list all the things ive tried, but im about to take everything back if i cant get it to work by tomorrow. It was tough watching it go from $14 to $29 per btc with both these cards sitting here...
Thanks in advance.
^(0|0)^

UPDATE: FRYS just sells about 80% faulty equipment. unreal amount of shit stuff. do not go there nothing they have that pertains to mining works.

Might be the power supply. Depending CPU utilization is pretty high on 1-2 core low end processors when mining and if you have both 6950 unlocked/overclock with and pau with an efficiency of 80% it's pretty close if not peaking.

I would bet on the PSU as well. 650 is pretty tight for those 2 cards. 750 would probably do, but 1000 or even 1200 would give you more head room.

With the current PSU, you could try a few things to shave some power draw off before you get a new PSU.

- Underclock your CPU and RAM on your motherboard as much as you can. Mining does use these and may people run 45 watt semprons in their mining setups just fine.

- On the GPUs, are you underclocking the GPU memory clock and/or overclocking the core clocks?

Seems like 400-600 on the memory clock is best and of course, pumping up the core clock will draw more power, so you might try underclocking the core of a while to see if it holds stable.

Use MSI Afterburner on windows and aticonfig (or BIOS flashing) on linux to tweak the GPU clocks. Search these forums for good tips on how to use both.

-Finally, what temps are you running at? It could be that the cards are overheating and causing the crash.

It can be done. Dont give up.

Do you have GPU-Z running or anything like that? Whenever I have GPU-Z running or the "video" settings page in CCC my 6950's crash. Try closing all applications and make sure no flash is playing in a web browser. More details on this can be found here:

http://forum.bitcoin.org/index.php?topic=4669.0 (http://forum.bitcoin.org/index.php?topic=4669.0)

and

http://forums.amd.com/forum/messageview.cfm?catid=390&threadid=128404&enterthread=y (http://forums.amd.com/forum/messageview.cfm?catid=390&threadid=128404&enterthread=y)

I have had the worst experience building a computer ever. FRY's Electronics on the west coast has screwed me in a unbelievable way.  I have made 13 trips to their stores. I have gotten more faulty equipment then i could of ever imagined. 2 bad motherboards, a bunch of video cards, everything is missing all components. the service is so bad i cant believe it. I went and took all the components (which have already been exchanged numerous times) back, and got a new mb and 2 new vid cards (not easy finding 1 card good for mining let alone having everyone you find is f**** up). Get home and put it all together and i cant even get a bios boot up off the board. I have 20+ graphics cards and after trying like 10 i inspect the shit out of the MB, only to find a piece of silicon scratched off and a break in the copper connector on back behind ram slots smaller than your pinky nail. Another bad board.
Un fucking real!!!
They cant give you your cash back if its over 500, you have to wait for 10 days then they issue a check from their main office. so your stuck in a loop of terrible equipment that they know is bad and sell anyway. (because only store credit).
Absolutely no one there has ever attempted to do anything technical ever, it is like going to a senior center for help with new technology.
O yah the help i request is, please dude please get me a unopened box man i cant take the missing shit and broken shit...it comes to me begging for a NIB product, and them saying this is all i can find right now handing me something they know wont work all with a attitude.
Please Please boycott frys if at all possible. I will never go there again after im out of this loop. Ive been going there since I can remember too.

Bad Parts:
Biostar TA890FXE MB
MSI 870A Fuzion MB
Diamond 5830 - missing all components, no output
2x MSI 6950 twin frozr II These are absolute shit, there was a guy walking in to return one as i was walking out after returning 2, we talked and we had the same issues. I talked to a girl there and she said that everyone returns them.
-I have 2 Diamond 6870's that arent working but it is most likely due to the short in the msi MB,but i would not be suprised if they are shit too after I get the MB changed for the 14th trip tommorow.
O yeah I returned to 6870's to get these 6870's because get this...I get two boxes exaclty the same, the diamond with the call of duty soldier on the front everything the same...I open it up and there are two different 6870's, different color, size, run temp...both 6870's.fucking different though.This is back on the biostar board, so i get them up and running and ofcourse they are both hashing when i get the blue screen loop crash...over and over every thing i try all suggestions and prior post references had been exhausted. SO i took everything back at  that point.
I had to get this shit off my chest as i have missed like 5 days of mining.there is more too but this is the main jist
Still not sure what to do...
There is no way for me to get my way, i am at their mercy.
Please post your bad experiences with hardware software and companies in the industry...Lets clarify what we should be using and not be using.
Let me be the first to say that We shouldnt be using FRYS
But is new egg or tiger direct better? or do they just send you used molested shit too??
Pissed off would be miner

Mannn that sukkks! yeah frys is good for piddly shit like cables , maybe ram thats factry sealed, etc. But if your in socal, chekout mwave.com or newegg, they are both in city of industry . . you can will-call same day go & get them, its always factory sealed either place ., . frys motherboards are like fukkin hindu 7-11 hotdog roller shit hahaaaaaaaaaaaaaaaaa

Thank you man im not alone, I havent done a build in a couple years so i thought it was me at first but 100% it is them.
O yah i bought a bigger PSU too, a nice one so its not that.

Right on man dont worry yeah i catch myself buying impulse shit at frys too. But really in the end its always bette to get it from a place that doesnt have their own shrink-wrap machine in the back if you catch my drift. You know that shiny saran-wrap box with the 'reduced price' sticker . . its like %3 off the regular price for somethign that a ritalin tard 14yr old gamer couldnt get to work in 5mins and threw it against the wall and crammed it bakk in the box

650 W psu is not enough for 2 6950s mining.
My unlocked 6950's use  ~640w at the power socket. i strongly doubt a 650w psu has 590-600w on the 12v rails used for the gpus.

My Seasonic X-650 has 648 watt on the 12V  ;D

Hey XER I agree that could be a potential issue thats why I got a nice 800w, i wrote that somewhere above, but i dont blame you it was a long ramble lol. but yah that is not the prob FRYS really sux on a new level.

And this is why I haven't bought anything from Fry's in a long time. They don't have any good stuff anymore. Most of the generic to low end things. Higher end quality products are hard to come by at the Fry's in the Bay Area

Took back the motherboard, and managed to get a factory sealed (so they say) Mb
MSI 870U-G55
Im going to open it up right now and inspect it, then do the build. I will report back later. I hope this works so I can forget FRYS exists and just order from new egg.

Many people are misinformed about wattage. It is NOT the wattage that you should be looking at on a psu. It's the amperage on the 12v rails people should be looking at. Depending on the psu, a 650w will have no problem powering a pair of 6970 in a non overclocked system. For example, there is a certain psu out there that has 72 amps combined on the 12v rails and was tested to output over 700w. And then there are 650 watt psu's out there that won't even power two 6950 :D. But anyway if I were you I would return everything to Frys and order from the egg just for piece of mind. They have great products backed by great service.

I got this all too work, but my hash rate is jumping around like crazy, and it seems to overdo it then reset itself with vpu...any suggestions?

Sounds like it's not getting enough power then. Does it still happen if you only use one card at a time?

It still happens with just one card...
PSU is TR2 RX 750W (Thermaltake)
I am pretty sure it is enough.
I guess it is possible that it is just more bad cards from frys??

I'm running an overclocked and shader-unlocked 6950, along with an overclocked 5770 on an Antec 650w PSU. I doubt it's a matter of insufficient power. I also can't see having such high failure rates. Have you considered faulty wiring in your house?

LOL funny you should ask, I am for sure that it is not that...
I know my electrical pretty well, at least on large scales like psu and up. I dabble in the raising of everyones favorite plant under quite a bit of wattage.
I am going to reinstall windows, new drivers, start over.
Suxxx guys thx for your help. I have a feeling this time that it isnt hardware related, which is all I really care about now

This is a total shot in the dark, but if you are running a few kw of lights is it possible that the ballasts create too much noise on the house wiring or possibly mini brownout situations that are minor enough that most equipment rolls through fine but the video cards are showing a sensitivity to? Is it possible you're seeing errors happen when the lights cycle on or off? If you had even a tiny UPS you might experiment with putting it in front of the mining system.

Ok so I found a setting to increase the frequency of the PCIe Bus in the BIOS settings. I upped it to around 110, turned off a few settings that appear to restrict the power for "green" purposes, put both cards (which are now diamond 6870's) in, default settings with -v -w128 and I have 15 minutes of constant hash power...I think I may have done L and G.

Note: Prior to the above solution, I reinstalled all MB drivers, and ATI drivers Etc....they were still crashing.

Thank you all for brainstorming I hope that the pcie frequency helps some people...
I guess I should state that you can F your board up by raising this too high... That being said, the default was 100 i turned it to 110 and the max it will let me is 190 (not that i have tried that, i havent tried above 110). If anyone knows more abot this or can confirm that this solved my problem then please let us know. Also is it possible to raise the performance of our gpu's by increasing this frequency.
Thanks

If it was going to do anything, upping the pcie freq by 10% would be more likely to make it less stable instead of more. Can't rule out it resolving some freak issue, but if I were to bet I'd be betting on something else you changed having resolved the problem. You could test it quickly by setting the pcie clock back to 100mhz and mining again to see if the problem comes back. Increasing the pcie frequency won't increase the cards hashing power.

Ok does anyone know the ideal setting for this then??
two diamond 6870's

Don't change the PCI-E frequency, that won't help.


Have you tested your ram?  Bad ram will cause all kinds of strange problems that are hard to pinpoint.

Download ORTHOS (multithreaded version) and run the "blend" test (maybe two instances if you have a quad-core)

I know people here will suggest running memtest86, but that's a real pain in the ass.


Also, please tell everyone here the amperage on your 12v rails

i have to wonder if the motherboard is sitting on a static bag that has those black strips on it which may be causing shorts
or somehow it is mounted in a case causing shorts. but then i did not read every post in detail and just a guess. i have seen
people do things like this before is all.