How do i config GPU to scrypt mining and a FPGA to SHA mining?

easy as, how do i config scrypt mining device 0 and SHA mining a device 1.

I have a computer that had two 7870 XTs in it, later I have added one 7770 and after a while replaced it with regular 7870, after I bought some more riser I have added 7770 again.

So right now I have four cards in following order:

PCI-E x1: 7870
PCI-E x16: 7870 XT
PCI-E x1: 7770
PCI-E x8: 7870 XT

Everything works fine, except that GPUs in cgminer and their details are all mixed up, for example one of my 7870 XTs shows temps from 7770 etc.

When I run cgminer -ndevs I get:


So you can clearly see that cgminer says that GPU3 is Tahiti (7870 XT) and then later displays that GPU 3 is 7770, can it be somehow reordered and saved to a config file, so sensors and auto-gpu would work?

Hi all,

I have been working trying to configure a guys rig with cgminer on Linux. I have configured everything as normal as I have cgminer working fine on 5 of my own machines with HD 7950 cards. I have also configured cgminer to work with 7970 cards with no problems.

However I keep having issues with this guys setup. Im installing on Ubuntu 12.04 LTS. The guy has several HD5870 cards and one HD5970 card

aticonfig detects all the cards fine e.g
* 0. 01:00.0 ATI Radeon HD 5800 Series
  1. 0a:00.0 ATI Radeon HD 5800 Series
  2. 09:00.0 ATI Radeon HD 5800 Series
  3. 06:00.0 ATI Radeon HD 5800 Series
  4. 05:00.0 ATI Radeon HD 5800 Series
  5. 04:00.0 ATI Radeon HD 5900 Series

The problem is no matter what Catalyst Driver I install I cannot get cgminer to start without throwing out errors.

Currently I get the error.
root@rig:/home/rig/cgminer-3.0.0-x86_64-built# ./cgminer -n
No protocol specified
No protocol specified
 [2013-05-20 23:18:05] CL Platform 0 vendor: Advanced Micro Devices, Inc.                    
 [2013-05-20 23:18:05] CL Platform 0 name: AMD Accelerated Parallel Processing                    
 [2013-05-20 23:18:05] CL Platform 0 version: OpenCL 1.2 AMD-APP (1016.4)                    
 [2013-05-20 23:18:05] Error -1: Getting Device IDs (num)                    
 [2013-05-20 23:18:05] clDevicesNum returned error, no GPUs usable                    
 [2013-05-20 23:18:05] 0 GPU devices max detected                    
 [2013-05-20 23:18:05] USB all: found 7 devices - listing known devices  

I have now had him remove the 5900 card as I thought it may be down to the number of cards he has hanging off this motherboard. I can see in lspci he has lots of Intel stuff so presume its an Intel based motherboard.

Has anyone any experience of this error?. I have tried the latest Catalyst Drivers for Linux for 64 bit including Beta and still cant get this working. I dropped the driver down to AMD Catalyst Driver 11.3 as I read that series 5xxx AMD Cards need the older drivers to work. Does anyone have working 5970,5870 cards and resolved this error?

Any advice is much appreciated to resolve this issue.

Thanks for the info. I have finally resolved this

I found a file .Xauthority that was owned by root in the home folder. This was causing issues.
sudo rm .Xauthority
reboot -h now
ssh back on the box
DISPLAY=localhost:10.0 ; export DISPLAY
Run up cgminer and everything is working fine :-)

What OS are you running?
I guess you could try the new code - it's working but not very well tested yet - and I need to chase up someone with a Cairnsmore1 to do that final part of it.
There's a branch I'm working on in the main git with the code as it stands at the moment - it handles '4' of the '5':
ICA (Icarus), BLT (Black Arrow Lancelot), LLT (Normal Lancelot - same as BLT), AMU (Asicminer USB) but not the CMR (Cairnsmore1) yet - well, to be correct I've no idea if it handles the CMR, but it shouldn't - I guess I should PM someone about that now ...

A screen grab (~15hrs run)

Ah windows, no I've not tested it on windows yet - so I've no idea if it even works
(code often needs minor or major changes after it's tested on windows ...)

For the time being you could:
Run cgminer in a script loop (so when it exits it simply starts again) and
Run some secondary program to check the API and send a quit if needed

I dont know if its my rigs or what but 3.1.1 is horrible for my gpu's on both rigs.
I get random HW errors on different GPU's after restarting cgminer, even after having it rebuild the kernel files. Random HW errors...
It takes like 3-7 restarts of cgminer to not get HW errors...

I run 7 x64 w/ 11.12 drivers on my 5xxx rig and 13.4 drivers on my 7xxx rig, same phenomenon.

I revert back to cgminer 3.0.1 and rebuild the kernel files, changing no settings ... everything works fine. No unexpected HW errors...

anyone else experience this randomness?

Are you scrypt mining? Earlier scrypt kernels would unintentionally hide hardware errors at high diffs...

Right, so if your scrypt settings are fucked, you'll generate almost nothing but loads of fucked data right from the start. Classic scenario would be high diff mining, such as solo, with high intensity and not being remotely aware that it was generating crap on previous versions, just wondering why you never find a block.