Ubuntu mining, optiminer 1.2.0 and Claymore 9.3 results

Here are my results from testing Optiminer 1.2.0 and Claymore 9.3 on a 7 GPU Ubuntu 16.04 Desktop rig. RX480 8GB Nitro Sapphire and MSI Z97 Gaming 5.

Optiminer

Claymore

Mining performance:
Both at -i level 8 on all GPU’s
if I subtract the 2.5% Dev fee from Claymore’s 1693 Sol/s I end up with ~1651 S/s.
Optiminer comes in at 1510 S/s so Claymore is slightly better by 141 S/s or ~9.3%. Optiminer also stated that he has more room for optimization on the RX 480 GPU’s.

Stability:
Claymore is still a beta but Optiminer seems to be fairly stable. Main issue I have with Claymore is the fan speed control has too much Hysteresis, especially for XFX cards I have on another rig (Temps can start oscillating). I found that manually setting the fax speed for each GPU at start in a script works best for both miners. I am looking at using the mining intensity -ttli switch in Claymore to provide more robust temp control (with static fan speed @ 80%).

Both miners can drop a GPU as temps start to rise (or for no apparent reason). Once this happens, on my rigs, even after killing the miner I hang when attempting a soft reboot and it requires a hard reset. Unfortunately, this prevents remote ssh monitoring of the system until I can find a reliable way to soft reboot. I have found staying below ~65C is critical for stability. Claymore does seem to drop GPU’s more frequently, and I have disabled the watchdog as the Miner always hangs when restarting. With the watchdog disabled at least it will keep mining on the remaining GPU’s.

Also: I have not been able to get Ubuntu Server to run reliably with 7 GPU’s for some reason. I am still looking into this as I see no reason to have a desktop running on a headless system. If I find a solution to this I will post.

Update:
I have Claymore running more stable on two rigs. Set static fan speeds in startup script depending on how GPU temps have looked (most @ 80% and few problematic GPU’s @ 88%). I have two XFX card that just get hot so I also have them set for -i 0 and -i 6 respectively. Set the -ttli option to kick in on all 14 cards at 63C. This seems to keep things very stable and no card ever gets above 64C. Have been mining this way most of the day and not a single dropped GPU.

The Nitro’s are running at -i 8 most of the time and only drop down when the room gets a little warm. I need to tweek the XFX cards a bit so they run at a higher -i at night when things are cold.

Important point: Claymore will drop a card lower than -i 0 via the -ttli option. I have seen it drop an additional 20% when a hot card hits 64C. I just need to test how low it will go and if I can set the initial -i higher to cover a larger ambient temperature range (70F to 55F in my house).

So now the next question: IS it better to restart the miner periodically via -r option or to just leave it running long term? I head out of town tomorrow so Ill set one to restart periodically and the other to run continuously to test.

Update:
Death by restart on Claymore.

Seems continuous run would be a better option as the other rig is running fine. Restarting the miner has some level of risk apparently.

Thanks for the sharing, it's really useful for me.

How much PSU are you using for 7 GPU? I'd like to have a try but my PSU is 1000W, does it will work for 7 RX470?

I am using two 1300W PSU's and linking them together with a Add2PSU. Its a little over kill since I don't need 2600W, but I wanted the extra headroom.