Re: [ANN] T-Rex 0.8.2 NVIDIA GPU miner with web control monitoring page
Did a test with my rig, 8 x 1070 under Linux Ubuntu LTE 18.04 Cuda 10.
Trex 0.8.2 first 4 cards intensity 21 / gpu +50 / men -400 / 170W -pl Brought me to 113 Raven/24h on a miner
Trex 0.8.2 last 4 cards intensity 22 / gpu +50 / men -400 / 170W -pl Brought me to 98 Raven/24h on a miner.
Great past 24h mining results I must say, blocks flying around
but getting a lower result on a higher intensity makes you think and get nuts. What to do with your settings on a X16r also?
GUI on rig was usable, for graph I do use the internal Intel of course. nvidia-smi shows some spare on the pl of 170W sometimes it jumps a little over it on both settings. GPU use most of the time 100% temps around 69 C and fans running 60%
What bothers me is the intensity, there is just so little usable info for it on a wild algorithm like the x16r and is difficult to prove as well. A day later it react smooth and then really aggressive again. But that?s probably also the reason it backs of the ASIC world of course.
For now I ?am happy and collecting, looking forward on shares of other experiences.
PS! switched now to t-rex 0.8.3 on Linux Ubuntu LTE 18.04 Cuda 10, till now it runs fine for 12 hours again.
Was it on the same rig on the same day or one day after the other.Trex 0.8.2 first 4 cards intensity 21 / gpu +50 / men -400 / 170W -pl Brought me to 113 Raven/24h on a miner
Trex 0.8.2 last 4 cards intensity 22 / gpu +50 / men -400 / 170W -pl Brought me to 98 Raven/24h on a miner.
Great past 24h mining results I must say, blocks flying around
but getting a lower result on a higher intensity makes you think and get nuts. What to do with your settings on a X16r also?GUI on rig was usable, for graph I do use the internal Intel of course. nvidia-smi shows some spare on the pl of 170W sometimes it jumps a little over it on both settings. GPU use most of the time 100% temps around 69 C and fans running 60%
What bothers me is the intensity, there is just so little usable info for it on a wild algorithm like the x16r and is difficult to prove as well. A day later it react smooth and then really aggressive again. But that?s probably also the reason it backs of the ASIC world of course.
For now I ?am happy and collecting, looking forward on shares of other experiences.
PS! switched now to t-rex 0.8.3 on Linux Ubuntu LTE 18.04 Cuda 10, till now it runs fine for 12 hours again.
With the difficulty changing everyday you can only compare on the same day with two rigs doing exactly the same hashrate on the same intensity. Then you can do a parallel test with different intensity.
Otherwise, you are comparing oranges to apples
Same miner with 8 cards split in 2 times 4 running parallel on same pool.
I know it's hard to get reliable results only the difficulty switching differs over the past 24h, that's my hole point of sharing it here to get useful info how to setup things well.