Thank you very much doc!
I have given it a clean start.
I verified that gputool as well as pimp --start 1 was disabled in /root/startup.sh so that was not the problem.
Removed voltage flags from phoenixminer config file.
Rebooted and restarted.
Same thing, crash after just a few minutes.
After dropping the whole thing for a while I thought I would try a different miner.
BANG! that was it! with claymoredual miner and all default it run great! Took rather lot of power but it run.
I let it run like that over night to see so it was stable. It was, it run ok for like 12 houres.
Now it feels like I have a working baseline.
Re-imaged the SSD with PiMP beta 2.17, to get a clean start.
Run pimpup, updated claymoredual, miner nr10. rebooted.
using echo "s 1 1300" > /sys/class/drm/card2/device/pp_od_clk_voltage and echo "m 1 1300" > /sys/class/drm/card2/device/pp_od_clk_voltage followed by echo "c" > /sys/class/drm/card2/device/pp_od_clk_voltage to apply the clock speed for core and mem.
Does it for all the cards.
In that way I could gradually take the speed up to the same as I had under win 10.
But voltage I could not set this way.
Voltage I had to set by altering the frequency / V curve.
GPU[0] : OD_VDDC_CURVE:
GPU[0] : 0: 800MHz @ 800mV
GPU[0] : 1: 1150MHz @ 805mV
GPU[0] : 2: 1550MHz @ 810mV
curve above is made with following comands.
rocm-smi -d 0 --setvc 0 800 800
rocm-smi -d 0 --setvc 1 1150 805
rocm-smi -d 0 --setvc 2 1550 810
then I run echo "c" > /sys/class/drm/card2/device/pp_od_clk_voltage to apply the new values ( not sure I need ).
Work great!
Now it looks like this, GPU[0] is the ASUS card, GPU[1] and GPU[2] Sapphire. (with system mining.)
================================================================================
GPU[0] : OD_SCLK:
GPU[0] : 0: 800Mhz
GPU[0] : 1: 1500Mhz
GPU[0] : OD_MCLK:
GPU[0] : 1: 820MHz
GPU[0] : OD_VDDC_CURVE:
GPU[0] : 0: 800MHz @ 800mV
GPU[0] : 1: 1150MHz @ 805mV
GPU[0] : 2: 1550MHz @ 810mV
GPU[0] : OD_RANGE:
GPU[0] : SCLK: 800Mhz 1820Mhz
GPU[0] : MCLK: 625Mhz 930Mhz
GPU[0] : VDDC_CURVE_SCLK[0]: 800Mhz 1820Mhz
GPU[0] : VDDC_CURVE_VOLT[0]: 800mV 1050mV
GPU[0] : VDDC_CURVE_SCLK[1]: 800Mhz 1820Mhz
GPU[0] : VDDC_CURVE_VOLT[1]: 800mV 1050mV
GPU[0] : VDDC_CURVE_SCLK[2]: 800Mhz 1820Mhz
GPU[0] : VDDC_CURVE_VOLT[2]: 800mV 1050mV
GPU[1] : OD_SCLK:
GPU[1] : 0: 800Mhz
GPU[1] : 1: 1300Mhz
GPU[1] : OD_MCLK:
GPU[1] : 1: 915MHz
GPU[1] : OD_VDDC_CURVE:
GPU[1] : 0: 800MHz @ 800mV
GPU[1] : 1: 1050MHz @ 805mV
GPU[1] : 2: 1350MHz @ 810mV
GPU[1] : OD_RANGE:
GPU[1] : SCLK: 800Mhz 1820Mhz
GPU[1] : MCLK: 625Mhz 930Mhz
GPU[1] : VDDC_CURVE_SCLK[0]: 800Mhz 1820Mhz
GPU[1] : VDDC_CURVE_VOLT[0]: 800mV 1050mV
GPU[1] : VDDC_CURVE_SCLK[1]: 800Mhz 1820Mhz
GPU[1] : VDDC_CURVE_VOLT[1]: 800mV 1050mV
GPU[1] : VDDC_CURVE_SCLK[2]: 800Mhz 1820Mhz
GPU[1] : VDDC_CURVE_VOLT[2]: 800mV 1050mV
GPU[2] : OD_SCLK:
GPU[2] : 0: 800Mhz
GPU[2] : 1: 1300Mhz
GPU[2] : OD_MCLK:
GPU[2] : 1: 915MHz
GPU[2] : OD_VDDC_CURVE:
GPU[2] : 0: 800MHz @ 800mV
GPU[2] : 1: 1050MHz @ 805mV
GPU[2] : 2: 1350MHz @ 810mV
GPU[2] : OD_RANGE:
GPU[2] : SCLK: 800Mhz 1820Mhz
GPU[2] : MCLK: 625Mhz 930Mhz
GPU[2] : VDDC_CURVE_SCLK[0]: 800Mhz 1820Mhz
GPU[2] : VDDC_CURVE_VOLT[0]: 800mV 1050mV
GPU[2] : VDDC_CURVE_SCLK[1]: 800Mhz 1820Mhz
GPU[2] : VDDC_CURVE_VOLT[1]: 800mV 1050mV
GPU[2] : VDDC_CURVE_SCLK[2]: 800Mhz 1820Mhz
GPU[2] : VDDC_CURVE_VOLT[2]: 800mV 1050mV
================================================================================
==============================End of ROCm SMI Log ==============================
========================ROCm System Management Interface========================
================================================================================
GPU[0] : Voltage (mV): 800
GPU[1] : Voltage (mV): 806
GPU[2] : Voltage (mV): 806
================================================================================
==============================End of ROCm SMI Log ==============================
with this settings I get:
ASUS 35.2 Mh/s @ 95 watts
Sapphire 39.3 Mh/s @ 100 watts
It have at the time of writing run stable for 2 hours.
It this runs OK over night again, it remain to put it all into a start script.
Thanks again docdrydenn for reminding how important it is to find/verify that you have a stable baseline to start from. And then not to mix a lot of variables. Stick with one thing at a time.
Fantastic PiMP community helping!
I would love to have All The Miners in miner.farm!
If these devices are USB attached to your rig, and use a miner program running on PIMP, it should work as expected. If it does not, I will need to integrate the mining program API into the agent - easy to do if we have access to a device, but we have a limited budget for the PIMP lab so help from miners in the field is appreciated! If you have a device running on PIMP that doesnt report and are willing to provide developer access for integration, please let us know.
If these devices are standalone, using their own OS and mining program, it is a much different prospect. Depending on the OS and what it will allow, it may be possible to install the standalone agent and have it report to miner.farm, but again we would need access to a running device to properly test and integrate.
Thank you,
Lily
