Issues with 2x 6870 rig.

[mazedk]

Hi

Im having some issues with my "re-use" mining rig.

My setup is an AM3+ cpu, gigabyte mobo, 4gb ram, 1200w gold + psu. 2x Powercolor 6870 cards.

Im trying to mine litecoins.

First off, Im still new to cgminer, but cannot get that running properly - just keeps crashing upon startup..

Anyways, I have tried both bitminter (which seems to work okay, but the actual hashrate isnt reporting the same on the website as in the miner..) and guiminer.

Guiminer works well for GPU2, for some time but then it just stops hashing and cpu usage goes to 0%
GPU1 doesnt work at all it seems..

Am I looking at some bad gpu's here? or whats the deal  :s

[1714142435]

1714142435

[1714142435]

1714142435

[1714142435]

1714142435

[1714142435]

1714142435

[1714142435]

1714142435

[mazedk]

the GPU2 runs for about 25 mins un guiminer before it shuts down.

[Trillium]

Software: Please try cgminer with default settings, it should run those cards quite well as well as give you some monitoring of temps. Use GPU-Z simultaneously to monitor the other sensor data for the cards. I recommend on a 6870 no more than 80 degrees C on the core temps. In my experience the VRM temps of the cards that have reference coolers are very low, which is nice, but still worth checking.

Hardware: What you are better off doing is running them at their stock settings (revert back to original BIOS if you've changed it) and then doing some error checking stress testing with a program like FurMark or OCCT GPU tests.

Unfortunately neither of these programs will make your life easy for you because the 2nd card will be hard to test under load even if you have a monitor connected to it as well as your primary display device. My recommendation would be to test the primary card and then swap the positions of the two cards and then retest the other card. It might be possible to test both cards simultaneously with these programs (eg multiple instances) but in my experience it was a headache.

Because it runs for 25+ minutes and then crashes it to me indicates a thermal instability issue ie. it's getting so hot it crashes. Also it's possible you have a problem with your powersupply though 2x 6870's won't even put 400 watts load on it... should be able to handle that.

Lastly... what fan speed are your running??? You'll need to force them to run at fan speeds that are higher than normal, either by using a program like Sapphire Trixx or one of the other vendors alternatives (they work on other vendors cards though). You could also try to setup cgminer or other mining software to control the fanspeed either statically or dynamically. Another option is to edit the BIOS of the card so it has a custom fan profile, although it's safer to use the custom fan profile in a program like Sapphire Trixx than edit BIOS if you are not up to the challenge.

An example of thermal instability on my 6870:

(900 Core, 300 Ram, 1.174 v) with air cooler (reference design): 70+ deg C core temps at 100% load --> Immediate crashes with any overclock more than just +15 MHz

versus

(900 Core, 300 Ram, 1.174 v) with watercooling: 43 to 45 deg C core temps at 100% load --> Easily overclocks to 990 Core and stable for days, not sure how much higher it can get (yet!)

[mazedk]

Hi Trillium.

Thanks a lot for the reply. Im pretty sure it has to do with my OC'ing aswell. I just put them back to stock speed and it (GPU1) is already pushing 92 deg C. after around 10 minutes.. - And im running the fans at 100% speed. GPU2 is running at 78 deg. C. with the same settings.

Im going to try and change the thermal paste tomorrow - got some AS 5 i'll try and put on.


Can you perhaps tell me your string for running cgminer? - im having trouble getting it to run properly - as said, it just crashes upon startup.. :/

[Zanatos666]

What about your airflow? Even with the fans, you have to make sure you are feeding cool air onto the cards, and that their is good exhaust as well.  If you are running fans at 100% and still hitting 90 says to me that air flow isnt optimal and the hot air is staying around the card(s).

[Trillium]

Thats a big difference in temperatures between those cards, if they are the same cooler design I think there must be a thermal paste issue or a severe airflow problem somewhere. Perhaps the heatsink is clogged with dust more than the other one.

I don't run a string for cgminer I use the .exe and the config file

The cards I have run with the default settings that cgminer uses for the GPUs when it finds them the first time. Tweaking them usually makes it worse.

[mazedk]

What about your airflow? Even with the fans, you have to make sure you are feeding cool air onto the cards, and that their is good exhaust as well.  If you are running fans at 100% and still hitting 90 says to me that air flow isnt optimal and the hot air is staying around the card(s).

Airflow.

Crossair Carbonide case, 1x 120mm fan in the side panel pushing in, 2x 120mm in the front pushing in, 2x 120mm in the top pushing out.

I got 2x gentle typhoon laying around, but those are too loud even for the place i got this box sitting..

[Zanatos666]

What about your airflow? Even with the fans, you have to make sure you are feeding cool air onto the cards, and that their is good exhaust as well.  If you are running fans at 100% and still hitting 90 says to me that air flow isnt optimal and the hot air is staying around the card(s).

Airflow.

Crossair Carbonide case, 1x 120mm fan in the side panel pushing in, 2x 120mm in the front pushing in, 2x 120mm in the top pushing out.

I got 2x gentle typhoon laying around, but those are too loud even for the place i got this box sitting..

Dont forget the back exhaust but sounds pretty good.  I think Trillium may have a good suggestion.  Take out the card running hot, take off the heatsink, make sure its clear, and maybe re-paste (with something better than the stock paste) and see if that helps.  I have a 5830 running with a 92 Delta (cut the stock fan off) and after I cleaned and re-pasted with AS 5 My temps, even below my desk, are mid to high 50's.

[Trillium]

I have a 5830 running with a 92 Delta (cut the stock fan off) and after I cleaned and re-pasted with AS 5 My temps, even below my desk, are mid to high 50's.

Sure that's degrees F, right?

By 'Delta' do you mean ambient<-->load temps? Just checking.

[Zanatos666]

I have a 5830 running with a 92 Delta (cut the stock fan off) and after I cleaned and re-pasted with AS 5 My temps, even below my desk, are mid to high 50's.

Sure that's degrees F, right?

By 'Delta' do you mean ambient<-->load temps? Just checking.

Delta as in 92mm fan. http://www.newegg.com/Product/Product.aspx?Item=N82E16835213009

And 50's as in C

[FiiNALiZE]

I have a 5830 running with a 92 Delta (cut the stock fan off) and after I cleaned and re-pasted with AS 5 My temps, even below my desk, are mid to high 50's.

Sure that's degrees F, right?

By 'Delta' do you mean ambient<-->load temps? Just checking.

I'm pretty sure that's celsius.

 

[Trillium]

I have a 5830 running with a 92 Delta (cut the stock fan off) and after I cleaned and re-pasted with AS 5 My temps, even below my desk, are mid to high 50's.

Sure that's degrees F, right?

By 'Delta' do you mean ambient<-->load temps? Just checking.

I'm pretty sure that's celsius.

 

Yeah. Because of my science background where 'delta' means 'difference between' I thought he was saying that his temps under load are 92 F more than ambient (it would have to have been F because if it was C no card could survive unless he lived in the arctic or something).

[mazedk]

Okay, so changed the thermal paste today.. AS5.. and removed dust from the fans/ribbons - with compressed air.. that didnt really do any good 

Now the temps are as follows:

GPU1 93 degrees C
GPU2 86 degrees C
- this is AFTER i underclocked both of them to 700core/525mem - lowest mem clock in msi afterburner for this model

I changed the top fans to push more cold air into the case though.

I know that AS5 is supposed to have a 200 hour "burn in" phase.. but doesnt this seem to be wrong? - im wondering if i put too little paste on the gpu maybe?

[Trillium]

Okay, so changed the thermal paste today.. AS5.. and removed dust from the fans/ribbons - with compressed air.. that didnt really do any good  

Now the temps are as follows:

GPU1 93 degrees C
GPU2 86 degrees C
- this is AFTER i underclocked both of them to 700core/525mem - lowest mem clock in msi afterburner for this model

I changed the top fans to push more cold air into the case though.

I know that AS5 is supposed to have a 200 hour "burn in" phase.. but doesnt this seem to be wrong? - im wondering if i put too little paste on the gpu maybe?

Underclocking the speeds will not affect heat output as much as lowering the voltage... you should not have to lower anything though! Surely not and get those temps.

The burn in phase will not alter the temps you get by too much, in my experience, it would be only 3-4 degrees C I'd say.

I think you might have a problem with what you think is the 'correct' quantity of thermal paste to use, either too much or too little will result in terrible temps. That or you are not seating the heatsinks back on tho chip right.

As for cleaning the heatsink and fan well the heatsink is just a chunk of metal there is nothing stopping you from washing it with some water and detergent until it's shiney new. I dry mine with a powerful room fan afterwards.

As for cleaning the fan then I'd probably recommend you just use an old tooth/paint brush to remove most of the dust. It is possible to actually wash the entire fan (yes, in water...) to get is 'clean like new'. The full process is:

- Submerge and scrub fan with a brush in some warm water along with a kitchen washing-up detergent. You want to buy the cheapest variety you can because it will contain less additives. You want to avoid anything that has things like Aloe-vera extract and moisturizers etc
- Remove the clean fan and rinse very thoroughly with running water. Then submerge it in clean water and shake it vigorously until it stops forming bubbles, and then clean it once more for good measure.
- Start drying the fan: Shake the fan in different orientations to remove as much water as possible. If you have a air compressor now use it to remove even more water.
- Importantly, completely dry the fan now using a powerful room fan, portable air conditioner exhaust, fan-heater, or hair dryer. I have a portable airconditioner and usually place the parts to be dried inside the hot air exhaust tube for an hour or two.
- Importantly, after you think it has dried enough, leave it a little longer, to be really sure it's dry.
- Importantly, re-lubricate the bearing of the motor with some light machine oil from a hardware store. It is seriously useful stuff to have and it costs $2

^ I have used this method on probably hundreds of computer fans of many types over the years and I am still to have one fail.

[Zanatos666]

Okay, so changed the thermal paste today.. AS5.. and removed dust from the fans/ribbons - with compressed air.. that didnt really do any good 

Now the temps are as follows:

GPU1 93 degrees C
GPU2 86 degrees C
- this is AFTER i underclocked both of them to 700core/525mem - lowest mem clock in msi afterburner for this model

I changed the top fans to push more cold air into the case though.

I know that AS5 is supposed to have a 200 hour "burn in" phase.. but doesnt this seem to be wrong? - im wondering if i put too little paste on the gpu maybe?

You change your top fans to PUSH AIR IN?  Not how the cases are designed.  Those top fans are suppose to blow heat out, heat rises.  Side panel and front fans bring cool air in, fans on cards suck the cool air up onto the cards, and hot air goes out the back and out the top.  If your fans are blowing air DOWN onto your cards from the top of the case, reverse them.

[Trillium]

Or just leave the side of the case off for now. Or maybe I have missed this guys problem... he might just have outrageously crap case ventilation.

I had assumed in all my comments he was running this WITH THE SIDE OF THE CASE OFF...

[Zanatos666]

Or just leave the side of the case off for now. Or maybe I have missed this guys problem... he might just have outrageously crap case ventilation.

I had assumed in all my comments he was running this WITH THE SIDE OF THE CASE OFF...

He stated before that he has 1 x 120mm on the side, 2 x 120mm in the front, and 2 x 120mm on the top, so I was assuming the side panel was on, hence the 1 x 120mm on the side.

[mazedk]

Having them push in was a thought about putting more airflow towards the back of the case - where the gpu's are.

Anyways i think its a matter of the amount of thermal paste - im thinking too little. will see when i remove the heatsinks tomorrow.

The sidepanel is on yeah, with a 120mm fan blowing cold air into the case.