[Mining OS] SimpleMining.net - Manage Your GPU farm the easy way! (30 days free)

[dzap]

Hi, any way to save logs on claymore (for example)

I get some many reboots on 2 rigs and for sure riser is the problem, but I cant see log and know wich gpu fails.

I saw -dgb and -logfile commands, anyone tried it?

Thanks and thanks tytanick to your hard work for more frequency updates on SM, this is what SM needs for a long time ago.

-logfile /home/miner/name_log.log with -dbg of your choice. Then you can view it over ssh with nano /home/miner/name_log.log, I bet you get the point. Make sure to save log to proper path, as claymore on smos launches on temporary dir that get rewritten on every miner start, hence you would lose your log.

[Bigburci]

Hi guys, can someone point me to the right direction to found out how Smos numbering the GPU-s? I couldn't found any information regarding this topic in the help section so just idea for the Dev to give us, users some more details how smos counting the GPU numbers, because definitely not based on the order of how you plug them into your motherboard.

I got a faulty GPU or riser on a 12x GPU setup which got the same temperature as others so hard to figure out physically which is that one.

Thanks in advance.

[dzap]

Hi guys, can someone point me to the right direction to found out how Smos numbering the GPU-s? I couldn't found any information regarding this topic in the help section so just idea for the Dev to give us, users some more details how smos counting the GPU numbers, because definitely not based on the order of how you plug them into your motherboard.

I got a faulty GPU or riser on a 12x GPU setup which got the same temperature as others so hard to figure out physically which is that one.

Thanks in advance.

just disable it with number on keyboard connected to rig, or over ssh, let it cool down for couple of minutes and you will be able to identify it easily.

As for pice numbers on motherboards, they can be random as well, just take a look for example:

https://imgur.com/a/CTw1p

[User_Shmuser]

For the love of any kind of theoretical omnipotent entity, please add the later release of XMR Stak. It's been out for months.

Also a question: How do I access XMR Stak thread config?

[Bigburci]

just disable it with number on keyboard connected to rig, or over ssh, let it cool down for couple of minutes and you will be able to identify it easily.

As for pice numbers on motherboards, they can be random as well, just take a look for example:

https://imgur.com/a/CTw1p

Thanks for fast reply, but... Is there any other way because I don't use ssh.
On the picture what you linked there, the PCI full slot is the GPU_0. That's mean maybe that in all mining motherboard the full pci slot is the GPU_0? I am using the Biostar TB250-BTC PRO motherboard which got some number markings on the pcb but I think they are not in the same order as smos numbering them.

[Bimmber]

As for pice numbers on motherboards, they can be random as well, just take a look for example:

https://imgur.com/a/CTw1p

Would you have similar picture, but for Asus Z270-A ?

[dzap]

I dont have any more pictures, this was taken from Gravitymaster's guide on SMOS.

Nevertheless the numbers written physically on motherboards dont even match their real physical connection numbers.

Bigburci, as I said before, you simply connect usb keyboard and do it straight on the rig. You can always play with fan settings for that one card, but from my practice it will be hard to determine the card in a forest of noise, the temperature drops to room temp in a matter of 2-3 minutes and is easiest way so far.

Just disable the gpus from 0 to x, where x ur number of gpus one by one and touching them once they cool off and sticky a note beside with their number, so you will know later on. It cant be any simpler...

[Bigburci]

Just disable the gpus from 0 to x, where x ur number of gpus one by one and touching them once they cool off and sticky a note beside with their number, so you will know later on. It cant be any simpler...

Yeah I will try this, thanks for your time.

[dzap]

Just disable the gpus from 0 to x, where x ur number of gpus one by one and touching them once they cool off and sticky a note beside with their number, so you will know later on. It cant be any simpler...

Yeah I will try this, thanks for your time.

But disable with the method I described, because I have a feeling you will just unplug them now, which will mess with your gpu numbers again :-)

[DUNC4N]

power consumption : Tytanick, settings below 900mV seem not to work on RX570/580 cards I have here (Sapphire Nitro, Pulse). Didn't try any else.

With SMOS, the lowest consumption I can achieve is ~150W at wall.  Asus Prime H270-Plus with January2018 BIOS, CPU G4560, 4GB RAM, single RX580 Sapphire Nitro at this moment but I tested with different cards up to 7 installed, 240V Europe, Corsair TX-850M (should be about 89.5% effective at this load). GPU 1000MHz, VMEM 1900MHz, underclocked BIOS.

MB+CPU+RAM consume 35W idle in Windows and SMOS without any GPU installed  - 35W measured at wall [!] after power supply ineffectivity etc.  With single GPU installed but not mining, Windows consumes about 47W ; SMOS 65W. Windows configured with High Performance profile, no CPU savings, no PCI-e savings in BIOS.


GPU Voltage set in web interface  |  Consumption at wall with SMOS

1100mV   180W
1000mV   170W
900mV   153W
850mV     153W
800mV   153W
750mV     153W

I feel it does NOT work as it should.

Powerstage 0      153W
Powerstage 1      153W
Powerstage 2      153W
Powerstage 3      153W
Powerstage 4      153W
Powerstage 5      153W
Powerstage 6      153W

Powerstage setting has absolutely no effect on power consumption at all. All these tests conducted with 1000MHz GPU, 1900MHz VMEM, 27.2MH/s. This particular card maxes out at 1950MHz memories no matter what voltage I apply, it's a crap.

I know PowerStage settings work in different settings and scenarios - I'm running some rigs with 1100MHz GPUs, 2000+MHz VMEMs and there is big difference in power consumption between PS6 and PS2. Generally I'm using PowerStage2 almost everywhere because PS1 and PS0 do NOT bring any decrease in real consumption / temperatures and everything above PS2 is heavily increasing power consumption.  Because I'm totally stable at PS2, that's what I use ; difference in power consumption between PS6 and PS2 is about 400 Watts with each [!] rig I have.

Here, in this particular case, PowerStage has totally no influence on power consumption.


GPU voltage settings BELOW 900mV  DO NOT have any effect on total consumption in SMOS, while they work in Windows. Decreasing GPU and VMEM in ATI WattMan (newest drivers, Adrenaline v18.2.3 dated 22/Febr/2018) below 900mV works correctly in Windows, HWINFO64 and GPU-Z report lower voltages for both memory and GPU and it is reflected on real wattage measured at wall.  I can get to 750mV GPU and VMEM with 120W total rig consumption at wall and yes, it's stable at 1000/1900MHz.

Lowest I'm able to achieve with SMOS is 152-153W.  I've tried so much...

 
So, the effective delta in power consumption between Windows and SMOS can be 20 Watts per card.  With 10 card rig, that is 200 Watts which really is a lot, especially if you have 10+ rigs.  Will you please look into this ? Any diagnostic output you need ? I can give you SSH access if needed.

And it's not just about power consumption - it's about temperatures, too. Of course Windows10 runs these cards cooler, because it is able to use lower voltage.

I even tried to lower voltages in BIOS, programming 850 (manual) or 65282 even 65281 in appropriate frequencies. Makes no difference in both Windows and SMOS.
Measurements in Windows taken with lowest voltage set as 65282 in BIOS. I'm not saying this is equivalent of 750mV, I'm saying power management in Ubuntu might be "not optimal" and potentially not working below 900mV. Regardless of what is in BIOS, it works in Windows.

Any hints or tips, please throw my way.  Is there anything in Ubuntu like HWINFO64 in Windows ? How can I display parameters of AMD RX570/580, such as frequencies, voltages etc ?

right undervolt does not work, I'm doing 850mv 1150 core, the house is getting high electricity from the wall, it was not like this before

[tytanick]

UPDATE:
- added miner ccminer-tpruvot-v2.2.5-cuda8.0 (for nvidia only)

[Bimmber]

2.2.5? Latest seems to be 2.2.4:

https://github.com/tpruvot/ccminer/releases

Can you add CC miner with SP MOD ?

[thedreamer]

UPDATE:
- added miner ccminer-tpruvot-v2.2.5-cuda8.0 (for nvidia only)

Is it better than claymore?

[tytanick]

2.2.5? Latest seems to be 2.2.4:

https://github.com/tpruvot/ccminer/releases

Can you add CC miner with SP MOD ?

It is compiled version from source files (today) and it sasy 2.2.5

Can you show me some source code what you mean exactly ?
Also why you want this over tpruvot ? - i need some arguments

[Bimmber]

CC miner SP mod has been modified to mine some coins with XX% (yes, double digit increase) better performance. Difference with default cc miner is huge.

[thedreamer]

CC miner SP mod has been modified to mine some coins with XX% (yes, double digit increase) better performance. Difference with default cc miner is huge.

No ETH support?

[jombolo]

Can you please add bminer to your SMOS? https://bitcointalk.org/index.php?topic=2519271.0

I have a problem with dstm miner in your SMOS. Using 12 x P106 mining cards, they all work good for about 2-3 hours, after that time, the hashrate is starting to decline from 3700+ Sol/s to 2000 Sol/s for unknown reason. When using ewbf miners, the speed is lower and the machine is restarting more frequently, but the miner stays at 3500+ Sol/s. bminer should be better alternative to both of these miner programs, and I found out that it gives higher performance and also better stability.

I'm currently mining with dstm on SMoS through flypool, but everyone seems to agree that bminer is a slightly faster. i cannot get the config line to work for flypool on SMoS. can somone provide an example of the complete config line

[chartor]

UPDATE:
- added miner ccminer-tpruvot-v2.2.5-cuda8.0 (for nvidia only)

OMG!!! The best!

[theminingbrother]

hello everybody,

I am desperate in finding a solution to my problem. I have several rigs with Nvidia 1070 cards that work perfectly and can be easily overclocked via simple-mining website. The problem is that when I try to overclock my AMD 580 and 570 cards instead, I get no result apart from changing the fan speed. So changing the power-stage and undervolting do not affect power consumption.

I need to solve the problem. I bought 800 units of XFX 580 XXX 8GB with hynyx and samsung memory and although I get 29mhs/s the rig power consumption is 1200W per rig of 6 GPUs. And this is so bad.

0.2 Ethereum to whomever first solve the problem

Try reducing your core clock. Undervolting in linux is not perfect. I get least power consumption in windows followed by ethos and then simplemining. Windows is hard to manage with multiple rig and ethos gives a slightly less hash rate. The only way to reduce power consumption is to reduce core clock. Reduce the core clock till the gain made by reduction in power consumption is equal to loss incur by hash rate reduction.

hello bro and thank you for the answer

the fact is that I can't manage to get under 950W by playing with power stage and core clock and core undervolting without dropping to 16mhs and below.....
can you suggest me a setting that could work with my XFX RX 580 XXX 8GB?
my goal is to go under 700w per 28mhs

Only windows can give you 700w per 28mhs. I am running 6 rx 570 4 gb in simplemining at 900 and 7 rx 570 4 gb in windows at 880.

I managed to go on 950 w with 38.5 mhs but given the cost of electricity where I live, it is not a sufficient improvement.

What about modifying the bios? The fact is that having 800 units of these GPUs (I am not kidding, 800 units) I need to be able to manage the whole thing without having to restart the rig from a crash once every day (maybe less, maybe more, it's windows). Can you imagine doing this for 130 rigs? Is there something I don't get or know that can solve the issue?

You can do few things

1) Modify the voltage, mem and core clocks in bios only.
2) Then try in simplemining, ethos and last in windows.
3) There is one guide on internet on how to setup windows for mining. I have some of my rigs running continuously for months on window without any issue. Always remember to modify the voltage, mem and core clocks in bios, when using window so that loading is very fast after crash.  I am converting all my AMD rigs from simplemining to windows due to excessive heat issue (summer is coming). Nvdia rigs will remain with simplemining.

hello Dhino
I bought a 128gb SSD and installed an original copy of windows. I installed the amd drivers for mining and as you stated I get an higher hash rate and less power consumption! that is great.

I was asking myself if there is any linux based system that can be similar or equal to windows in terms of hash rates and power consumption balance for mining with AMD cards... simplemining is not so effective in respect of windows when it comes to efficency...

AMD released in the last October the drivers for mining in linux https://support.amd.com/en-us/kb-articles/Pages/AMDGPU-Pro-Beta-Mining-Driver-for-Linux-Release-Notes.aspx do they work properly?

[jombolo]

UPDATE:
- added miner ccminer-tpruvot-v2.2.5-cuda8.0 (for nvidia only)

OMG!!! The best!

What is tpruvot best used for?