5970+7950 BAMT for Litecoin problem/question

[limburatorul]

Hey everyone,

Since I can't do anything without posting in the Newbies section on the forum I thought I should ask for a little help.
 
I just made my first mining rig. Started with a 5970 and later added a 7950 to the configuration. The problem is that I can't overclock the 7950. I mean I can oc it but it does nothing to the hash rate. Still 52 Kh/s. No matter what I do.
The 7950 is in the main PCI-e, followed by the 5970 in the x8 PCI-e port.
I got a Gigabyte Z77X UD3H mobo and on the specs it stated that the x16 port shares the bandwith with the x8. So when the x8 port is populated the x16 one will operate as a x8 one.
The 5970 works great in the x8 port. Got the same hash rate as I got when it was in the x16 port.
Should I move this in the x4 port?
Anyone experienced this while configuring the mining rigs?

Thanks

[jdarta]

hello,

I've a similar rig and I'm also having some problems with both GPUs. not the same problem as you, but maybe someone can help us both.

Cooler Master HAF XB
Cooler Master V1000 Modulaire - 1000W 80+ GOLD
Intel Pentium G630
MSI Z77A-G45
Sapphire Radeon HD 7950 with Boost - 3 Go
Sapphire Radeon HD 5970
2x2Gb DDR2 PC3-10600
64Gb Sandisk SSD
running Win 7 Pro x64
CGWatcher 1.3.3.1
CGMiner 3.7.2
Catalyst 19.2
AMD-APP-SDK-v2.9

This is my cgminer.conf:

Code:

{
"pools" : [
   {
      "url" : "http://tag.cryptopools.com:3334",
      "user" : "x",
      "pass" : "x"
   }
]
,
"intensity" : "19,18,18",
"vectors" : "1,1,1",
"worksize" : "256,128,128",
"kernel" : "scrypt,scrypt,scrypt",
"lookup-gap" : "2,2,2",
"thread-concurrency" : "22400,8000,8000",
"shaders" : "0,0,0",
"gpu-fan" : "50-90,50-90,50-90",
"gpu-engine" : "1120,785,785",
"gpu-memclock" : "1475,1150,1150",
"gpu-memdiff" : "0,0,0",
"gpu-powertune" : "20,20,20",
"gpu-vddc" : "0.000,0.000,0.000",
"temp-cutoff" : "90,90,90",
"temp-overheat" : "80,80,80",
"temp-target" : "72,72,72",
"api-port" : "4028",
"auto-fan" : false,
"expiry" : "30",
"failover-only" : true,
"gpu-dyninterval" : "7",
"gpu-platform" : "0",
"gpu-threads" : "1",
"hotplug" : "0",
"log" : "5",
"no-pool-disable" : true,
"no-submit-stale" : false,
"queue" : "1",
"scan-time" : "5",
"scrypt" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"kernel-path" : "/usr/local/bin"
}

With HD 7950 i get ~650KH/s stable with this settings. I think I found the sweet spot without OC (other that cgminer.conf).
With HD 5970 i get with 1st GPU ~210KH/s and with 2nd GPU ~300KH/s

in theory I should be able to get at least 350KH/s per core with HD 5970.

At this moment i get ~1,12MH/s shouldn't I get at least 1,3MH/s with both HD 7950 and HD 5970?

I've tryed many differents combinations of Intensity, memclock, gpuclock and TC. but anything over 790MHz (GPU clock) crashes the driver even with intensity of 13.

With this config I've managed to run the rig for 24h non stop, today I've tested new configs but after 10min HD 5970 goes sick.

Another thing that I find strange it's that both GPUs of HD 5970 don't produce the same hashrate. The 1st core it's always slower that the second.

I've HD5970 in the first PCI 16x and the 7950 in the 2nd PCI 16X.

Did you tried to OC the HD 7950 alone, without the 5970 installed?
are you using cgminer 3.1.0? I had to use the last cgminer that supports scrypt (3.7.2), cgminer 3.1 wouldn't start with everything by default and Intensity 13.
Changed for cgminer 3.7.2 and all went smoothly, just had to add the -T parameter to cgminer.

http://i42.tinypic.com/2gwga5h.png
http://i39.tinypic.com/ehaxvo.jpg
http://i43.tinypic.com/2ezpr94.png
http://i42.tinypic.com/22ew0k.jpg


hope this helped.