Topic: EDIT: SOLVED Double 7970 setup help (Read 998 times)

Cant help you here sorry.

Ok, the problem got solved. It was the driver. Did sudo apt-get remove --purge fglrx* ... reboot ... sudo sh ./amd-catalyst-13.4-linux-x86.x86_64.run ... reboot ... VOILA! Stable 720+ khs on both GPUs.
Thanks for the help.

Yes, i have tried both.
At the moment the board runs with one GPU and the other is on different board, both are doing 650+ khs

Hi,
Have you tried switching the GPU's in place?
Have you also tried running 1 gpu at a time to test if all of the settings are ok?

I don't think it is the intensity problem, I suspect it has something to do with the mobo or the way PCIEs work on this specific mobo. When i stick the GPUs in separate mobos they work like charm using exactly the same conf. If i raise the intensity above 13 the hashrate drops quickly. For some 7970s the I 13 is the maximum. I plan to buy another mobo soon and then I can try out again.

Here's what I did to solve this...

I uninstalled my ATI drivers (I used the fglrx repo previously)

* (re) installed AMD APP SDK v2.7
* installed the LATEST AMD ATI drivers [ amd-catalyst-13.4-linux-x86.x86_64.zip ]
* downloaded AMD ADL SDL v5.0 (then extracted the .zip)
* downloaded cgminer v3.0.1 source (not BUILT)
* copied ADL/include/*.h to cgminer v301 ADL_SDK dir
* compiled according to the README, then make, then make install

I do not know if it was due to the updated AMD driver or compiling cgminer - but now I get this ~820Mh/s with about 400Mh/s each, without OC!! (screenshot coming)

hmm interesting.. before I even used v3.0.1 I had a script in my home dir that called v3.0.0 and it works fine.. therefore, cgminer version (at least v3.0.0) was not the cause

I'm getting accurate FAN sppeds and monitoring - perhaps it was the ADL? I'm confused now

edit - sorry now I'm confused. your screenshot, gudmunsn, shows Kh/s mine are Mh/s - and your total is 1.xMh/s mine is ~400Mh/s:

I get it now, you were posting script hash rates while I was focused on BTC


thanks for all the advice!

below..



Wow, appreciate you posting that screenshot - I'll def try higher intensity settings. Can you post a SS with hitting the "G" option to show the clock/mem speeds that cgminer is reporting

edit - forgot to add, not using the cfx cable
edit 2 - better yet, would you post your cgminer params? I would greatly appreciate seeing what other settings you might have that differ.



since the volatage is almost half difference between them, I assumed PSU as well as MannyEric - I did plan to run (3)78xx but might need larger PSU - also will the change in PCI bus impact? (16x8x8x when you have 3x PCIe vs 16x16x right now)


I'll increase logging - I grep'd syslog and did not see anything for "sick"


and while we are here, got any advice on litecoin/minerd settings? I'm only seeing 11kh/s per GPU

Are they OCed?

Does cgminer say something like SICK GPU?

Could you post some cgminer logs?

Psu should be fine, I run 2x7950 off a decent 650w with plenty left over. As for hash rate whack that intensity up to 18.

Also, I don't think you need to use the crossfire cable. Not sure if that will help anything, but I'm not using one.

I think the problem you guys are having is intensity. I have mine up a lot higher than yours (13 and 9). This is mining w/ 3 7950's btw.

I'm having a similar problem. I'm still searching to resolve, I'll update here if I find anything.

Here's what I have and what I've tried -

double 7870 GE setup
Asus 750W PSU
ASUS Sabertooth 990fx RA.2
Xubuntu 12.10
cgminer 3.0.0



Latest AMD/ATI drivers

I get the same GPU output in terms of GPU#1 stuck at 300/150:


my question really is, is this cgminer config issue or drivers? almost tempted to dualboot/install windows with guiminer and test.

Hi.
It's my first multigpu rig so I'm having some issues with cgminer doing scrypt.

Here's the output. It seems to me that the hashrate is spreaded along both GPUs and actually only 1 GPU is working? Total hashrate is 658.5 kh/s as would be OK for 1 GPU. As you can see the 2nd GPU has low clocks and low voltage, although in cgminer the configuration is the same for both. I'm suspecting lack of power, I have SeaSonic M12II series Modular 650W 85+ PSU. Maybe it's not enough? If this is the case how much should i have?
MoBo is MSI 970A-G46 and CPU Sempron 145, have 2 4GB DDR3 dimms. Both GPUs are Sapphire HD7970 OC.
Any help would be much appreciated. Please have a look on the cgminer conf as well below. Thank you.

GPU 0: 329.4 / 329.4 Kh/s | A:2  R:0  HW:0  U:1.94/m  I:13
65.0 C  F: 70% (2638 RPM)  E: 950 MHz  M: 1425 Mhz  V: 1.175V  A: 99% P: 0%
Last initialised: [2013-04-10 14:16:29]
Intensity: 13
Thread 0: 164.7 Kh/s Enabled ALIVE
Thread 1: 164.5 Kh/s Enabled ALIVE

GPU 1: 330.4 / 329.1 Kh/s | A:6  R:1  HW:0  U:5.82/m  I:13
32.0 C  F: 70% (2724 RPM)  E: 300 MHz  M: 150 Mhz  V: 0.850V  A: 0% P: 0%
Last initialised: [2013-04-10 14:16:29]
Intensity: 13
Thread 2: 165.0 Kh/s Enabled ALIVE
Thread 3: 165.5 Kh/s Enabled ALIVE


"intensity" : "13,13",
"vectors" : "1,1",
"worksize" : "256,256",
"kernel" : "scrypt,scrypt",
"lookup-gap" : "0,0",
"thread-concurrency" : "8192,8192",
"shaders" : "2048,2048",
"gpu-engine" : "950,950",
"gpu-fan" : "0-70,0-70",
"gpu-memclock" : "1425,1425",
"gpu-memdiff" : "0,0",
"gpu-powertune" : "0,0",
"gpu-vddc" : "0.000,0.000",
"temp-cutoff" : "95,95",
"temp-overheat" : "85,85",
"temp-target" : "75,75",
"api-port" : "4028",
"expiry" : "120",
"gpu-dyninterval" : "7",
"gpu-platform" : "0",
"gpu-threads" : "2",
"hotplug" : "5",
"log" : "5",
"no-pool-disable" : true,
"queue" : "1",
"scan-time" : "60",
"scrypt" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"kernel-path" : "/usr/local/bin"
}