CGMiner - Disparate graphic cards

[erpbridge]

I had a 6870 in my normal desktop, been mining LTC with CGminer for a few weeks now. Just got a couple 7950's to use in an old motherboard which I just replaced (Core2 Quad, 8GB DDR2 Ram). Found out my old motherboard only has 1 PCIE 16, but it has two more PCIE 1x on it. I have some powered x1 to x16 adaptors coming in, but they'll be at least half a week.

My normal desktop has enough PCIE x16 to run three cards, but due to fan location in the case, power supply restrictions, etc, I can only fit one of the 7950 cards (longer form) and the 6870. I'll run the other 7950 in the mining rig until I can get the cable.

All cards are Gigabyte.

Question is:
-The first question, which pretty much overrides anything else... Can you mine realistically with two different model ATI cards? A 6870 and a 7950?
-In CGMiner, how do you tell which card is GPU0, and which is GPU1?
-Also, in CGminer, it seems that no matter which way I set the conf file, the cards seem to match to the lowest common hash rate. The 7950 SHOULD be mining well above the 6870 (which would clearly answer question 1), but instead, they're both equalizing to within 20 or so hash rate of each other. Is this normal? Is it a limitation of the PCIe bus, where cards only talk at the slowest speed on the bus? Or is it something where I should choose a different mining software, or maybe does CGminer require opening two sessions, one saying use GPU0, the other saying use GPU1?

All these become a moot point when I get the powered 1x-16x cables later this week, but sometime down the road I will want to upgrade the card in my desktop (for gaming) and re-use it's 300-ish kh (or 300 MH) old 6870 in a miner... which just might make that miner a mixed model machine again.

Thanks,
Steve

[centove]

Question is:
-The first question, which pretty much overrides anything else... Can you mine realistically with two different model ATI cards? A 6870 and a 7950?

I may be wrong, but I think I saw somewhere that though it will work, it's not recommended to mix the cards like that, you should keep the 7xxx cards and the 6xxx cards separate. (I can't find the reference right this second)

-In CGMiner, how do you tell which card is GPU0, and which is GPU1?

cgminer -ndevs should tell you what cards it's seeing and which one is which.
 

-Also, in CGminer, it seems that no matter which way I set the conf file, the cards seem to match to the lowest common hash rate. The 7950 SHOULD be mining well above the 6870 (which would clearly answer question 1), but instead, they're both equalizing to within 20 or so hash rate of each other. Is this normal? Is it a limitation of the PCIe bus, where cards only talk at the slowest speed on the bus? Or is it something where I should choose a different mining software, or maybe does CGminer require opening two sessions, one saying use GPU0, the other saying use GPU1?

See above.

[goemon888]

Question is:
-The first question, which pretty much overrides anything else... Can you mine realistically with two different model ATI cards? A 6870 and a 7950?

I may be wrong, but I think I saw somewhere that though it will work, it's not recommended to mix the cards like that, you should keep the 7xxx cards and the 6xxx cards separate. (I can't find the reference right this second)

then a 7970 and a 7870 can be live fine together? and even a A10-5800K?

[centove]

Question is:
-The first question, which pretty much overrides anything else... Can you mine realistically with two different model ATI cards? A 6870 and a 7950?

I may be wrong, but I think I saw somewhere that though it will work, it's not recommended to mix the cards like that, you should keep the 7xxx cards and the 6xxx cards separate. (I can't find the reference right this second)

then a 7970 and a 7870 can be live fine together? and even a A10-5800K?

I don't have those exact cards... but this is what I have:
 [2013-04-06 19:35:38] CL Platform 0 vendor: Advanced Micro Devices, Inc.                    
 [2013-04-06 19:35:38] CL Platform 0 name: AMD Accelerated Parallel Processing                    
 [2013-04-06 19:35:38] CL Platform 0 version: OpenCL 1.2 AMD-APP (1113.2)                    
 [2013-04-06 19:35:38] Platform 0 devices: 4                    
 [2013-04-06 19:35:38]    0   Scrapper                    
 [2013-04-06 19:35:38]    1   Pitcairn                    
 [2013-04-06 19:35:38]    2   Pitcairn                    
 [2013-04-06 19:35:38]    3   Capeverde                    
 [2013-04-06 19:35:38] GPU 0 AMD Radeon HD 7540D hardware monitoring enabled                    
 [2013-04-06 19:35:38] Failed to ADL_Overdrive5_FanSpeedInfo_Get                    
 [2013-04-06 19:35:38] GPU 1 AMD Radeon HD 7800 Series  hardware monitoring enabled                    
 [2013-04-06 19:35:38] GPU 2 AMD Radeon HD 7800 Series  hardware monitoring enabled                    
 [2013-04-06 19:35:38] GPU 3 AMD Radeon HD 7700 Series  hardware monitoring enabled                    
 [2013-04-06 19:35:38] 4 GPU devices max detected                    

and a snap shot of the performance:
--------------------------------------------------------------------------------
 [P]ool management [G]PU management (S)ettings [D]isplay options [Q]uit
 OCL 0:  12.0C         |  42.9/ 42.9/ 63.8Mh/s | A: 7 R:0 HW:0 U: 0.89/m
 OCL 1:  75.0C 1670RPM | 316.5/314.7/319.0Mh/s | A:35 R:0 HW:0 U: 4.46/m
 OCL 2:  75.0C 1954RPM | 316.7/314.4/282.5Mh/s | A:31 R:0 HW:0 U: 3.95/m
 OCL 3:  73.0C  59%    | 147.4/142.3/136.7Mh/s | A:15 R:0 HW:0 U: 1.91/m
--------------------------------------------------------------------------------

and it runs like that for days...

[biged214]

Hey guys,
I run a 7950 and a 6770 just fine with cgminer. I use two difference instances for each card with their own settings and they both work perfectly. Hopes this helps 

[grue]

each card's clocks can be set independently of each other. there is no reason to run separate instances for each card. for example, to configure clocks for multiple gpu, you can use
"--gpu-engine 900" to set 900mhz for every card, or "--gpu-engine 900,850,875" to set 900mhz for card0, 850mhz for card1, 875 for card2, etc.

[biged214]

each card's clocks can be set independently of each other. there is no reason to run separate instances for each card. for example, to configure clocks for multiple gpu, you can use
"--gpu-engine 900" to set 900mhz for every card, or "--gpu-engine 900,850,875" to set 900mhz for card0, 850mhz for card1, 875 for card2, etc.

^ WIN! Thanks for the info