I'm guessing that having a motherboard with onboard video & using that as your primary monitor solves the problem.  Not an option with me, but now I know.

I figured out what was going on:  I had my two command shells open running phoenix, which was writing a line out to the shell every second or so & scrolling the rest up.  The act of scrolling the text (& therefore updating the screen pixels) was taxing the second card.  Minimizing all windows that had changing contents fixed the problem - both cards running at 100%.

I tried AMD GPU Clock tool - it appears to be incompatible with the 6970.

MSI Afterburner seems to be a much better product that Trixx, however, it still doesn't reduce the clock below 1375.  It allows me to do it, but as soon as I start mining, it reports the speed at 1375.

Also, changing the sliders on Afterburner and/or CCC seems to be a ticklish affair - I will move the slider and one second later it moves back.  Is there a way to make sure that only one of the tools is controlling the cards?  This happens even if I use task manager to make sure that CCC.exe is not running.

See my original post: Antec HCG-900.  Why do you think the power supply is stopping you?  Is it because you only see one adaptor when you use mining tools (or examine them in CCC)?  If so, have you made sure that there's a monitor plugged into both?  if nothing is plugged into your second adaptor, it will be disabled by windows & unavailable for mining.  If you don't have a second adaptor, google "Dummy Plug" for instructions on how to build your own or buy one.

Hi,

My setup:

Intel Q6600
Asus P5K-E (a 16x PCI-E slot & a 4x PCI-E slot)
Antec HCG-900
Sapphire 6970 (x2, no crossfire cable connected)

Driver version 8.85-110419a-118226C-ATI

I'm using Catalyst Control Center (latest drivers) and Trixx.  I've read elsewhere people underclocking their 6970 memory way down to 500, but CCC doesn't let me go any lower than 1375.  In Trixx, I can set the memory to 500, but I don't think this is being honored; using GPU-Z shows the memory speed still at 1375.  Is there another tool I need to use?

After playing around for a while, I've managed to get both cards run at 392 Mhashes/s.  It seems that I don't entirely understand the interaction between Trixx and Catalyst Control Center.  I thought Trixx would control both cards when the "Synchronize cards in Multi-GPU config" was chosen.  However, this seems not to be the case.  The second card is not paying attention to Trixx controls, only CCC controls; yet the first card pays attention to both - they have to be the same in both apps or the performance of the first card suffers.

Also, it seems that mhash rate takes a hit whenever the fan speed changes; having the default "custom" setting in trixx (C=%) resulted in worse performance than just keeping a steady clock speed.

Hi,

I've set up a mining rig:

Intel Q6600
Asus P5K-E (a 16x PCI-E slot & a 4x PCI-E slot)
Antec HCG-900
Sapphire 6970 (x2, no crossfire cable connected)

GPU Clock: 900
Mem Clock: 500
Voltage +10%

phoenix -u http://XXXXXXX -k poclbm DEVICE=1 BFI_INT VECTORS AGGRESSION=11 WORKSIZE=128

When running with just device 0, I get about 410 MHashes/sec; when I run with just device 1, I get 375 MHashes/s.  But with both running, I get 365 Mhashes.  Is the x4 slot dragging me down?  I wouldn't think bus speed would be a limiting factor with all the computation taking place inside the card.

Is anyone else able to get scalable work with 2 6970s?