Topic: Bytom Mining - ASIC algorithm now cracked by GPU miners? Big profits? - page 16. (Read 17547 times)

Correct 5gpus and 2 instances was best case scenario for me. 1050ti and 1060 3gb are the best as they are both cheap and as the miner is not optimised the big gpus don't have any hashrate advantage. I think with a professional miner, this problem will be solved.

Just the news I needed to start my day. Thanks.

B3 will go down in history as the ASIC killed by the smallest of gpus. Let's mine the hell out of this coin before Bitmain releases the real deal.

This girl deserves a follow, a retweet, and a good ****.

The ohgodapill for Ethereum has done wonders for boosting the profitablity of our GDDR5X GPUs and now with this miner coming out soon, this female developer has been greatly contributing to the mining scene. Won't be surprised if she gets job offers soon (not that she needs them). Congratulations!

https://twitter.com/OhGodAGirl/status/1000258062979092481

So you mean 5gpus running two instances is the best choice in your case?
I read the whole thread and see that 1050ti is the best p/p vga for this bytom mining, right?

Thanks to gregory021998 for helping me get it running.

On G4400, 4 gb ram and 12*1050ti HiveOS rig, I found out that I can't run more then 10 gpus. With 10 gpus I can only run a single instance. Hash rate - 600-750 h/s.

Only with 5 gpus I can run two instances. Hash rate approx 300-400 h/s per instance. So hash rate of 5 gpu is same as hash rate of 10. I can only run 3 instances with 3 gpus only. Hashrate was 125-140 h/s per instance i.e. 125-140 h/s per 1050ti. Cpu load was high with 3 instances. No OC.

I will test it again when I get an i5 cpu.

There is a lot of hidden potential with GPU Bytom mining and if the miners are optimised, I am sure it will be the most profitable big mcap coin. Sharing is caring guys, so please share if you discover something about Bytom gpu mining.

EDIT : After some messing around, I am able to run all 12 1050ti on single instance. Hash rate 600-800 h/s. Can't run second instance with G4400, it freezes. Neither I am able to increase hashrate with more instances like some of the guys here. My hashrates gets divided if I open more instances so Hash rate remains same (or even less). It puts a LOT of load on CPU. So yeah powerful CPU is needed.

You're right, my CPU usage is about 200% with a Pentium G4400 3.3.Ghz! Don't think miners usualy have better CPU...

Not to my knowledge someone correct me if that’s wrong though

Here is a little guide to run this on UBUNTU

I did a lot of trouble shot just to decide, it is not for me (all my 6 gpus nvidia rigs have crappy cpu, and I think that the pool is stealing from us or the miner is(and hate the fact that the "-url" doesn't do shit)).

So first things first, here is my  run.sh, just little mods (to not kill other process, and to see the log on the term).

#!/bin/bash

cd $(dirname $0)

if [ ! -f address.txt ];then
echo -e "\n file address.txt not exists! \n"
exit
fi

SMI=nvidia-smi

ADDR=$(cat address.txt | head -1 )

DRV=$( $SMI -h |grep Interface | awk -Fv '{print $2}' | cut -d. -f1 )
CARDS=$( $SMI -L | wc -l )

WK=$( /sbin/ifconfig  eth0|grep "inet addr"|awk '{print $2}'|awk -F. '{print $3"x"$4}' )

if [ $DRV -lt 387 ];then
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:cuda8
else
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:cuda9
fi

echo "Driver = $DRV , CARD COUNT=$CARDS , WK=${WK}"


cd btm-miner
./miner -user ${ADDR}.${WK}


Take care with the address, is the most common mistake, you can´t mine directly to most of the exchanges (because the address type, i don't know how does that really works), easiest fix, just download the wallet and create one.


Now, I did everything on UBUNTU 16.04 LTS (this works for 18.04, but the oc method doesn't, someone with more knowledge should  fix that very easy)

Download and fresh install UBUNTU 16.04, click all the extra installations.

Next you need to install nvidia drivers, I used 387  version, so please use the same.

just type this on the terminal and accept everything

$ sudo add-apt-repository ppa:graphics-drivers/ppa
$ sudo apt update
$ sudo apt install nvidia-387

RESTART the system

now you have to see all your gpus if you type "nvidia-smi"

Now download the miner and extract it.

Please copy and paste the .sh I put before, and paste your own address on the address.txt (upool pays with no registration)

now move on your terminal to the carpet of your sh and do

$ sudo chmod +x run.sh

now runit with

$ ./run.sh

you should see all working now to oc (this part doesnt work on 18.04 )


First you have to make a virutal screen for each gpu easiest way to do

$ sudo nvidia-xconfig -a --cool-bits=28 --allow-empty-initial-configuration

We have to check that was correct

go to

$ sudo nano /etc/X11/xorg.conf

now search for the word Coolbits, you should see this

Option "Coolbits" "28"

not this

#Option "Coolbits" "28"

if you have the hastag please erase ti on each gpu (you will have listed all your gpus there)
And dont move anything more!

ctr+x for exit, and save it (read the foot of the screen )

Now do this (change permissions so at startup the configuration doesn't reset)

$ sudo chmod 444 /etc/X11/xorg.conf && sudo chattr +i /etc/X11/xorg.conf

RESTART the system

now you have to make a new script, crankit.sh (or any other name you want)

#!/bin/bash
 
# Script needs to run as sudo for nvidia-smi settings to take effect.
[ "$UID" -eq 0 ] || exec sudo bash "$0" "$@"
 
# Setting a terminal variable called memoryOffset
 
# Since all my cards are the same, I'm happy with using the same Memory Transfer Rate Offset I use this conf to 1050 ti
memoryOffset="700"
cpuOffset="100"
 
# Enable nvidia-smi settings so they are persistent the whole time the system is on.
nvidia-smi -pm 1
 
# Set the power limit for each card (note this value is in watts, not percent! CHANGE THIS TO THE NUMBER OF GPUS YOU HAVE

nvidia-smi -i 0,1,2,3,4,5,6,7,8 -pl 53
 
## Apply overclocking settings to each GPU
nvidia-settings -a [gpu:0]/GpuPowerMizerMode=1
nvidia-settings -a [gpu:0]/GPUMemoryTransferRateOffset[2]=$memoryOffset
nvidia-settings -a '[gpu:0]/GPUGraphicsClockOffset[2]=$cpuOfset


nvidia-settings -a [gpu:1]/GpuPowerMizerMode=1
nvidia-settings -a [gpu:1]/GPUMemoryTransferRateOffset[2]=$memoryOffset
nvidia-settings -a '[gpu:1]/GPUGraphicsClockOffset[2]=$cpuOffset




Add another and change the number on gpu:x for each of your gpus

to run the script you have to chmod again

$ sudo chmod +x crankit.sh

then

$ ./crankit.sh

And you are done, just change the variable on blue and run it and you will be oc your nvidia GPUS

Any tips (eth)

0x0EE0873De03961208F955B4F38d7b953204De042

How much ram do you have? cpu?

Hey, use two instances instead of one. (need 4 gb of ram at least and a good cpu).
Instead of using run.sh, use this https://pastebin.com/gR2e1bp5

Create miner.sh and miner2.sh (with the script above) and run them in screen

Change the path of the miner, the address and the worker

Check your CPU usage, if max, then you need a better CPU to drive the cards.

Mine is stuck at "New job received from stratum server" for last 20 minutes.
What should I do ?

I have not registered at Upool or anything

All i did so far : download and extract in HiveOS- change wall address- Run *run.sh     this what it's showing
https://ibb.co/kwVLqo

https://ibb.co/g0cN38
 
https://ibb.co/iKpDAo  

i restarted the rig and now it is stuck at last gpu with "No work ready" message.

Edit : restarted again. Now again stuck at "New job received from stratum server"
Please help.

I don't understand: I only make 650 H/s with 6x1060!
edit: 720 with OC (+100/+700).
edit2: very strange, only use between 35 and 60W per card and load is between 0 and 60%!
edit3: pools sucks or miner sucks, I don't know but it sucks

Pool use new address not erc20

Your wallet address is wrong...

I will be reading it shortly also ...

#crysx

very insightful! read this thread all the way

Well, regarding the 4 x 1060 the hash seems pretty low to me.
If you want pm me , i would like to run some test , we can chat on telegram...
Are you using more than 4 gb of ram? with an i5 it's impossible to do an avg of 400 h/s, you should be at least to 600-700 h/s

Latest mining stats summary

Antminer b3

https://ibb.co/nayeFo

GTX 1060 rig x4. At 250w with i5 quad

https://ibb.co/hwwwT8

Considering the pc rig takes less juice I would say they are pretty close in total hash rate per watt.

that is realistic earnings for that amount of hash -- you can compare w/ stock antminer b3 here - https://goo.gl/v86eKG