Bitcoin Forum

I have tried to search for multi GPU but the search pulls up everything but.  How do I mine with my 2 R9 280x cards?

currently using ethminer -G -F http://eth-us.suprnova.cc:3001/

for single GPU but how do I run for both GPUs?

etminer --help

For nvidia it's --cuda-devices 0 1 2 etc for AMD it's probably something similar.

Ok that --help command helped..  so it appears if I got it correct to be ..

ethminer -G -t 2 -F http://eth-us.suprnova.cc:3001/worker/<estimated MH/s)

--opencl-device 0,1

tried that and it failed.. reported rpc error and only GPU Miner 0 was actually showing in the workload.

ethminer -G --opencl-device 0,1 -F http://eth-us.suprnova.cc:3001/

--cuda-devices only work without commas (0 1 2 instead of 0,1,2) so it might be the case for opencl as well, I'm not sure.

Hmm, weird..
I used ethminer as well and my bat looks like:

ethminer.exe -G -F http://yourpoolhere
--opencl-device 0,1

The only difference is that i have the device specification one line underneath the rest. Try doing that.

Thanks that seems to work seems the --opencl-device 0,1 has to be at least at the end.  Now I seen my first failure. Is that normal when mining on more than one card or should I increase my voltage a bit?

Sapphire Vapor-X R9 280X   875/1250/1125
XFX Double D R9 280X         875/1250/1125

Awesome. Glad it's working for you  :)

I've always left my cards at stock settings. If your looking to squeeze a little bit more hash out you could add the "--cl-global-work 16384" parameter.

Thanks for that tip so I imagine I would add that right after the --opencl-device 0,1 --cl-global-work 16384

Getting some really good hash rate on the easy pool on Suprnova bouncing between 44-67MH/s so not bad at all.

Yep. I dont remember the percentage increase exactly but it should give you a few more mh. Its somethig similar to raising the intensity on cgminer and similar software.

Keep in mind that the pools calculate your hash by share rate per certain amount of time. So you will see crazy swings in your displayed hashrate from time to time.

With 875 MHz core frequency, you can reduce the core voltage from 1125mV to 1000mV or lower. That will reduce the power consumption.

Tried to go down got the BSOD.  It appears XFX Double D does not like to run below 1.125 volts someone mentioned here and looks to be correct..

hi guys

having similar problem

getting json-rpo all the time

i have bought 6 gpus 3+3 sepparate   R9 380x
so 3 gpus per rig.
using windows 10, latest amd drivers, ethminer 0.9.41 genoil 1.0.4b3

bat file
setx GPU_MAX_ALLOC_PERCENT 99
setx GPU_USE_SYNC_OBJECTS 1
setx GPU_MAX_HEAP_SIZE 100
ethminer.exe --farm-recheck 100 -F http://eth-eu.dwarfpool.com:80/address/Worker1 -G --opencl-platform 0 --cl-global-work 8192

one card works fine
two...they work but sometimes not
Three , not at all. json error

thank you

Make sure you have the power supply to support the cards, and that you have good drivers?
I have a Gold-cert 850W PSU in one system. It's close to the limit but it works fine. (1000W unit on the shelf in case it blows :) )
You won't make it happen with a marginal PSU without going into your GPU utility and underclocking or reducing the power. (to a point where you may as well run one card, maybe two)

But no, you don't need anything special at all for multi-GPU.
If you're trying with crossfire enabled, turn it off and try. (Though mine works OK with it on as well)

I use the opencl ethminer 9.4.1.
My batch file for suprnova, for one of my rigs:
ethminer -G -F http://eth-us.suprnova.cc:3000/username.workername/93

That's it! Period.  All four of the cards in that rig come online with that simple command.
Adjust your user/worker name and hashrate to match reality of course. (though the hashrate being high/low won't keep it from 'working' in any way.)

In playing around, I had better 'luck' with my 4-card rig when tweaking the command a bit.
ethminer -G -F http://eth-us.suprnova.cc:3000/username.workername/93 --cl-global-work 8192 --cl-local-work 256 --no-precompute --farm-recheck 333
If I start getting excessive errors I drop the --no-precompute and let it generate a new DAG but it doesn't happen often.

But yes... it will "run" all four cards without the extra stuff.
NO conf file, nothing special required.

I have another with two R9 380X using the same basic command.
Windows 10 64-bit

Psu is ok.
i have been able to run the miner for a while without json error
but dwarfpool didnt notice even thou my rig was working

now
everithing is back to  old. json error and thats it

Perhaps try underclocking the cards a bit, or increasing the voltage.
Especially if you're only seeing this with all three of them running.
If you're positive the psu is fine(ie you have checked it out under load), I'm out of other ideas at this point, unless your cards are just flaky or something.
These things normally 'just work' with no issues at all.
Running latest stable versions of ethminer, video card driver, etc...?

If you can't nail it down, you might try the cards in another system.
Or take one out, buy another card and build another 2-card system, if it only errors with 3 cards.
With so many variables and so little information given, I can't think about it any more without my head hurting. Sorry :)
Good luck.

my psu
EVGA SuperNOVA 850 B2

i have been able to run it for a while on full but then usual json appeared

H81 chipset MB
4GB RAM
32G SSD
connected over wifi

latest crimson amd drivers
i havent made win update yet since it has been downloaded from microsoft website 3 days ago

quite a headache :D

yes i tried and it work without comma, all my 970 are hashing for amd there are other option, like -t2 or -g

Wifi?

Json are connection issues.

See where i'm going with this?

Try a cable for starters..

Or if that isn't possible, at least use powerlines...

Good luck!

140MHs
more coming soon :D
thanks guys

i have considered wifi as one of the reasons but my router is situated really bad
made some testing but have to find another place...

due to BBC Winter is coming but thanks to GPUs i consider this as not woth of mentioning  :D

If you buy the Asrock H81 or H61, you can put 6 GPU into 1 rig, that will reduce the cost of the system, ie one motherboard instead of 2.

I'm trying to get my 5 card 280x rig running but ive no idea if its going to work or not. I've used the bat file which cdub kindly posted above. I've run it and it is still generating the DAG file after over 1 hour. Is this correct?

It got up to 100% earlier,then it started again, much slower and I'm seeing all sorts of lines such as

gpu miner1 work loop....
Pause 9858ms
gpu miner 1 stopping 11812ms
gpu miner 1 workloop....
main start worker
 and so on...

should I be seeing this.

Sorry for the noob questions but I'm new to Ethereum...

Edit: it has started mining :)

If it is first time you start mining, the etherminer will generate two DAG files. So it will take a while to start mining.