I've done more testing, here are the results.
I've put the 7950 in the riser slot by itself, I get ~400 again.
I've repaired the 6950 and put it back in, and I'm back to my 1Mh/s that I had before (which is what I expected).
In either event I'm not going to worry about it anymore, since I'm back to what I had.
Topic: GPU 0 mines slower (Read 1235 times)
What do you get when you take out the 6950 and leave the 7950 in the riser slot?
Here are the settings from my "base" configuration (before the broken fans on the 6950). The 7950 is GPU1 in this configuration (so it should be using the 2nd set of numbers).
Code:
"intensity" : "18,18",
"vectors" : "1,1",
"worksize" : "256,256",
"kernel" : "scrypt",
"lookup-gap" : "2",
"thread-concurrency" : "8192,22336",
"shaders" : "0,0",
"gpu-engine" : "900,1000",
"gpu-fan" : "90",
"gpu-memclock" : "1375,1400",
"gpu-memdiff" : "0,0",
"gpu-powertune" : "20,20",
"gpu-vddc" : "0.000,0.000",
"temp-cutoff" : "95,95",
"temp-overheat" : "85,85",
"temp-target" : "75,75",
"api-port" : "4028",
"expiry" : "120",
"failover-only" : true,
"gpu-dyninterval" : "7,7",
"gpu-platform" : "0,0",
"gpu-threads" : "2,2",
"log" : "5",
"no-pool-disable" : true,
"per-device-stats" : true,
"queue" : "1",
"scan-time" : "1",
"scrypt" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"kernel-path" : "/usr/local/bin",
"no-submit-stale" : "true",
"verbose" : "false"
The only difference when the 6950 was out was the thread-concurrency"vectors" : "1,1",
"worksize" : "256,256",
"kernel" : "scrypt",
"lookup-gap" : "2",
"thread-concurrency" : "8192,22336",
"shaders" : "0,0",
"gpu-engine" : "900,1000",
"gpu-fan" : "90",
"gpu-memclock" : "1375,1400",
"gpu-memdiff" : "0,0",
"gpu-powertune" : "20,20",
"gpu-vddc" : "0.000,0.000",
"temp-cutoff" : "95,95",
"temp-overheat" : "85,85",
"temp-target" : "75,75",
"api-port" : "4028",
"expiry" : "120",
"failover-only" : true,
"gpu-dyninterval" : "7,7",
"gpu-platform" : "0,0",
"gpu-threads" : "2,2",
"log" : "5",
"no-pool-disable" : true,
"per-device-stats" : true,
"queue" : "1",
"scan-time" : "1",
"scrypt" : true,
"temp-hysteresis" : "3",
"shares" : "0",
"kernel-path" : "/usr/local/bin",
"no-submit-stale" : "true",
"verbose" : "false"
Code:
"thread-concurrency" : "22272",
Interestingly enough though, I put the 6950 back in without fans into the PCI riser slot. I had to turn all the settings for it down so it wouldn't overheat, and the 7950 with the exact same settings is now doing 530 kh/s. Strange... So here's the setup:
1x MSI 6950 (Twin Frozr II) into 16x PCI-E slot
1x XFX 7950 with powered 1x to 16x PCI-E riser
The CPU is one of those APU jobs (A5-Trinity or something).
OS is CentOS 6.4
I'm mining scrypt coins
Rig has been running at 1Mh/s for a while now, the split between them is roughly 620/380 with the advantage going to the 7950. The fans in the MSI Twin Frozr broke (I was being stupid) and I took it off to fix it. While I'm trying to find parts I moved the 7950 to the main board PCI-E slot. I created a new config file and just removed the settings for the 6950, with the exception that I had to lower the thread concurrency slightly.
Now, my hash rate is down to 420ish, and I'm not 100% sure why. I have a theory though, and that is that the main card's resources are being consumed (at least slightly) by running Xserver and/or VNC through the main card. The system is in my garage so I'd prefer to have it "headless", thus the VNC server.
My question is, does anyone else experience this? Is there anything I can do to remedy it? What about using the APU to try and run XServer/VNC as opposed to the Main GPU? Is that possible? EDIT - Is there any reason to suspect the powered riser vs the main PCI slot is making a difference?
Thanks in advance for the advise.
1x MSI 6950 (Twin Frozr II) into 16x PCI-E slot
1x XFX 7950 with powered 1x to 16x PCI-E riser
The CPU is one of those APU jobs (A5-Trinity or something).
OS is CentOS 6.4
I'm mining scrypt coins
Rig has been running at 1Mh/s for a while now, the split between them is roughly 620/380 with the advantage going to the 7950. The fans in the MSI Twin Frozr broke (I was being stupid) and I took it off to fix it. While I'm trying to find parts I moved the 7950 to the main board PCI-E slot. I created a new config file and just removed the settings for the 6950, with the exception that I had to lower the thread concurrency slightly.
Now, my hash rate is down to 420ish, and I'm not 100% sure why. I have a theory though, and that is that the main card's resources are being consumed (at least slightly) by running Xserver and/or VNC through the main card. The system is in my garage so I'd prefer to have it "headless", thus the VNC server.
My question is, does anyone else experience this? Is there anything I can do to remedy it? What about using the APU to try and run XServer/VNC as opposed to the Main GPU? Is that possible? EDIT - Is there any reason to suspect the powered riser vs the main PCI slot is making a difference?
Thanks in advance for the advise.
Doubt it's VNC causing the problems since I do not have those issues. Check your 7950 settings again... what are they?
So here's the setup:
1x MSI 6950 (Twin Frozr II) into 16x PCI-E slot
1x XFX 7950 with powered 1x to 16x PCI-E riser
The CPU is one of those APU jobs (A5-Trinity or something).
OS is CentOS 6.4
I'm mining scrypt coins
Rig has been running at 1Mh/s for a while now, the split between them is roughly 620/380 with the advantage going to the 7950. The fans in the MSI Twin Frozr broke (I was being stupid) and I took it off to fix it. While I'm trying to find parts I moved the 7950 to the main board PCI-E slot. I created a new config file and just removed the settings for the 6950, with the exception that I had to lower the thread concurrency slightly.
Now, my hash rate is down to 420ish, and I'm not 100% sure why. I have a theory though, and that is that the main card's resources are being consumed (at least slightly) by running Xserver and/or VNC through the main card. The system is in my garage so I'd prefer to have it "headless", thus the VNC server.
My question is, does anyone else experience this? Is there anything I can do to remedy it? What about using the APU to try and run XServer/VNC as opposed to the Main GPU? Is that possible? EDIT - Is there any reason to suspect the powered riser vs the main PCI slot is making a difference?
Thanks in advance for the advise.
1x MSI 6950 (Twin Frozr II) into 16x PCI-E slot
1x XFX 7950 with powered 1x to 16x PCI-E riser
The CPU is one of those APU jobs (A5-Trinity or something).
OS is CentOS 6.4
I'm mining scrypt coins
Rig has been running at 1Mh/s for a while now, the split between them is roughly 620/380 with the advantage going to the 7950. The fans in the MSI Twin Frozr broke (I was being stupid) and I took it off to fix it. While I'm trying to find parts I moved the 7950 to the main board PCI-E slot. I created a new config file and just removed the settings for the 6950, with the exception that I had to lower the thread concurrency slightly.
Now, my hash rate is down to 420ish, and I'm not 100% sure why. I have a theory though, and that is that the main card's resources are being consumed (at least slightly) by running Xserver and/or VNC through the main card. The system is in my garage so I'd prefer to have it "headless", thus the VNC server.
My question is, does anyone else experience this? Is there anything I can do to remedy it? What about using the APU to try and run XServer/VNC as opposed to the Main GPU? Is that possible? EDIT - Is there any reason to suspect the powered riser vs the main PCI slot is making a difference?
Thanks in advance for the advise.
Jump to: