For anyone running stock firmware, or not wanting to pay a 2% free forever on a simple mod, below is my tool to change L3+ voltage for free  Its simple to use but requires basic use of ssh and command line. More info in the thread here: https://bitcointalk.org/index.php?topic=3546316.0Since most boards come with ridiculously high voltage for the stock frequency (so bitmain can sell us stuff like L3++), you can reduce the voltage to the lowest setting on most boards, which brings down power consumption from 800 watts to about ~650 per miner. |
|
|
|
So certain members of this forum think its cool to rip off GPL code then do simple mods and profit greatly off it without releasing the code after having plenty of time to do so, therefore I decided to release a simple binary that does the most important function for free on the stock L3+ firmware  Figured lots of L3+ users would appreciate this with low profits and summer incoming. My tests show that most boards will run on stock frequency (384) at the lowest voltage setting (0xfe) reliably at about 160 watts per board at the wall. Source and release binary are here: https://github.com/jstefanop/bitmain-toolsThis binary allows you to easily modify voltages on a per board bases for L3+ miners. It *should* work on all L3+ boards as long as they have the proper PIC version, If the binary complains about a wrong PIC version let me know (it wont do anything unless your board has the version I have tested on). USAGE:Since this need to be run directly from the miner console, below is my attempt at a user friendly instructions for people not familiar with the command line: Download the release binary on GitHub and copy it over to your antminer: scp [Directory you downloaded tool]/set_voltage root@[your miners ip]:/config ssh into your miner and cd to config folder (this is the only folder that is saved on reboot on antminers so we saved the binary here). ssh root@[your miners ip]
cd /config binary accepts two inputs in the format of: ./set_voltage [chain# 1-4] [voltage in hex] bitmains voltage controller can be configured to change the 12v input roughly +/- 1v from 10v, and this is configurable via a hex range of 0x00-0xfe, with the default being set to the middle (0x80). Higher hex values (0x80-0xfe) will LOWER voltage, lower values (0x00-0x7f) will INCREASE voltage from the default. If your not familiar with hex numbering, all you need to know is that they range from 0-9, then a-f For example if you want to slightly decrease your voltage on chain #1 you would input: increments of 0x10 are good starting point to test a sweet spot for each board for a particular frequency. Lowering voltage until you get around 1 HW error per minute is usually a good reference “sweet spot.” So lower voltages from stock in increments of 0x10 are 90, A0, B0, C0, D0, E0, F0 and higher are 0x70, 0x60, 0x50 etc. If you want slightly higher resolution just go increments of 0x08, so lowering voltage in lower increments would look like 0x88, 0x90, 0x98, 0xA0, 0xA8, 0xB0 etc. The tool directly writes the voltages to the pic controllers memory, so all changes are saved even on reboot and power down. If you want to set everything back to stock, you have to set the stock voltage for each chain using: ./set_voltage 1 80
./set_voltage 2 80
./set_voltage 3 80
./set_voltage 4 80 Enjoy! Keep in mind that the "other" solution has take a good few bucks or so a month of profit per L3 miner, so while I like helping out the community when I can especially in cases like this, please consider donating a few dollars per L3 if you end up using this tool, especially if you guys want me to work on similar tools or release more free stuff (like the auto voltage tuner I'm working on)  LTC: LQZpb8AqbggUmsdPKr28DzdNcRP7MJ8kEf BTC: 1LeA29mLjTFCyB5J6Yiq4TFtAck2VDPUZf DISCLAMER: While the tool has basic error checking its not idiot proof, so be careful especially when overvolting. IM NOT RESPONSIBLE IF YOU FRY YOUR BOARD. |
|
|
|
something is just odd about this. let me elaborate.
the E3 is priced at 2000-2400. it mines like a 6 card rig and costs about the same as one
this new z9 is like 12 grand in GPUs but only priced at 2k? if bitmain sold 12k in gpu's they would be asking at least 9-10k for it
sooo what gives? are they just expecting a fork? at current this thing is what? a 45 day ROI which is something i have never seen since i have been in crypto. did this ever exist years back?
the price to performance just seems out of line
sigh...why do people never learn. Bitmain is not pricing this based on what a single unit on the current network could mine, nor should you ever figure out your earnings on any new miner based on that. As a rule of thumb, take bitmains X new miner, multiply the hashrate by 10,000 units (which is their typical batch size), add that hashrate to the current network and assume thats what you will be earning on day 1. |
|
|
|
If this doomsday device is what the OP suggests. I don't see how GPUs are going to survive this. If the ASIC is a 10 kiloton hiroshima style weapon. This is a 10 megaton hydrogen bomb. And it's going to get dropped right on your GPU rig. I wet the bed last night after reading this thread.  This CPU/GPU/FPGA/ASIC cycle has happened 2-3 times over as some people have pointed out. Except this last cycle everyone forgot about FPGAs because there were more "nerds" back in the early bitcoin days. Making an algorithm FPGA "resistant" is probably easier than ASIC "resistant", and new algorithms will come out with new coins and the cycle repeats ;P Looks like people re-discovered FPGAs this round again since the large ASIC companies have been going straight from GPU->ASIC while ignoring the smaller algorithms. Id be careful in buying these "dev" boards though. They cost 5k because they are not meant for consumers. I could probably build a board with with these xilinix FPGA's for under 2k (and looked into it and been approached about it in the past), and that are designed for crypto in mind. Either way its the same deal as ASICs...of course you'll make 300 bucks a day when there are 5 people mining FPGAs on a coin....as soon as 50 people jump on its 30 bucks a day... Yes I know, people started mining Bitcoin with these FPGA and then suddenly ASIC came and people forgot about them. But now ASICs being released for all of the algos, I think gpu mining has finally come to an end and FPGA will live for a little longer (if it works). I think either NVidia or Amd should release new dedicated mining gpu or we need some revolutionary new algo for gpus. Also can you elaborate how do you make algo FPGA resistant but at the same time it won't be resistant to gpus. I know little bit about FPGAs, my friends owns some spartans and vertix 7. Nothing is resistant...it all comes down to economics. With FPGAs your limited with the number of programable gates the silicon has. The more complicated the algorithm the less "cores" you can fit on the die. If someone wanted to create a GPU algorithm that was not as economical on an FPGA, you would make it so each round requires a ridiculous amount of instructions, or have the coin use a large amount of different algorithms which take up precious FPGA space. You then take the advantage away from FPGAs because 1) it would require alot more effort and time to program the FPGA, and 2) you can't fit as many cores on it, so the instead of it being 10x as fast as a GPU, you can only get away with 2-3x with maybe a slight increase in efficiency. |
|
|
|
If this doomsday device is what the OP suggests. I don't see how GPUs are going to survive this. If the ASIC is a 10 kiloton hiroshima style weapon. This is a 10 megaton hydrogen bomb. And it's going to get dropped right on your GPU rig. I wet the bed last night after reading this thread. This CPU/GPU/FPGA/ASIC cycle has happened 2-3 times over as some people have pointed out. Except this last cycle everyone forgot about FPGAs because there were more "nerds" back in the early bitcoin days. Making an algorithm FPGA "resistant" is probably easier than ASIC "resistant", and new algorithms will come out with new coins and the cycle repeats ;P Looks like people re-discovered FPGAs this round again since the large ASIC companies have been going straight from GPU->ASIC while ignoring the smaller algorithms. Id be careful in buying these "dev" boards though. They cost 5k because they are not meant for consumers. I could probably build a board with with these xilinix FPGA's for under 2k (and looked into it and been approached about it in the past), and that are designed for crypto in mind. Either way its the same deal as ASICs...of course you'll make 300 bucks a day when there are 5 people mining FPGAs on a coin....as soon as 50 people jump on its 30 bucks a day... |
|
|
|
I have a fried Moonlander 2. This one was working perfectly on one of my computers for weeks, and I moved it to another computer that was in a more favorable location. Set up the computer the same way, Windows 7, standard software (bfgminer, etc). However, when I plugged it in, and the fan worked fine, and then I smelled the familiar smell of smoked electronic components. Looks like U2 on the board is fried with a black smoldered hole in the middle of it.
What is the component? What could have caused this? Is this ML2 repairable, or is it a lost cause? My other 2 ML2's are working great, but am scared to move them in the event this same thing happens.
Any guidance would be great.
Thanks!
Hmm thats not good. U2 is the memory voltage controller. Sounds like something shorted or overheated. Were you running higher clocks? If only the controller blew the repair would be relatively easy and its a 50 cent component if your good with removing/soldering SMDs. If your in the US and don't mind sending it in shoot me a PM...catastrophic failures like this are super rare so they are good dissecting cases. No, I was running straight defaults at 600 clock speed on litecoinpool.org with 128 difficulty. I am in the US, and will send you a PM. Thanks! Is there something wroung with the USB port on your computer? No, I have been running two other miners and no problem with that port. Only thing I can think of is that a surge took it out. Are there any other components before the voltage controller from the initial usb connection on that circuit? Jstefanop, I sent you a PM a couple of days ago, but I am not sure if you got it. Thanks! Sent you a PM. |
|
|
|
I'm looking to update my raspberry pi from 5.4.2-34 to 5.4.2-38 for the auto restart. Has the GitHub armv6 download link been updated to the newest version? It still installs ver. 5.4.2-34.
thanks!
Maybe that's why bfgminer on my pi will not auto restart. My bfgminer comes up with 5.4.2-34, not 5.4.2-38 when following the instructions from page 1 hmmm, the beta 2 link for armv6 shows build 34? Ill check it and update it. Ill be building new version with latest updates soon anyway. |
|
|
|
Yes point it to my pool in the sig  |
|
|
|
|
This is not new or news...its just an overclocked L3+...
|
|
|
|
|
I need a single L3+ hashboard, and a controller. I can take your bad boards as well as long as they are semi-working (high errors, dead chips etc).
|
|
|
|
I have a fried Moonlander 2. This one was working perfectly on one of my computers for weeks, and I moved it to another computer that was in a more favorable location. Set up the computer the same way, Windows 7, standard software (bfgminer, etc). However, when I plugged it in, and the fan worked fine, and then I smelled the familiar smell of smoked electronic components. Looks like U2 on the board is fried with a black smoldered hole in the middle of it.
What is the component? What could have caused this? Is this ML2 repairable, or is it a lost cause? My other 2 ML2's are working great, but am scared to move them in the event this same thing happens.
Any guidance would be great.
Thanks!
Hmm thats not good. U2 is the memory voltage controller. Sounds like something shorted or overheated. Were you running higher clocks? If only the controller blew the repair would be relatively easy and its a 50 cent component if your good with removing/soldering SMDs. If your in the US and don't mind sending it in shoot me a PM...catastrophic failures like this are super rare so they are good dissecting cases. |
|
|
|
bfgminer.exe --scrypt -o stratum+tcp://eu.multipool.us:3351 -u Moonlander2.1 -p x,d=256 -S MLD:all --set MLD:clock=796
pause Ok so i have it running on litecoinpool and on multipool. Works but i run into the same error. After a while it just stops hashing. If i restart bfg then it works like a champ again. So can someone help me. Is there a way to set bfgminer up to auto restart after an hour. yea I know its not perfect but at this point i just want the dang thing to run nearly 24 hours. I hate waking up to find it pooped out after 30 minutes and sat idle 6 hours. Is there a way to add a piece to this scrypt that will make it auto restart after an hour? Since I have no clue why it keeps stopping. And before we trouble shoot again just assume its all running on the right stuff. yes powered hub. yes 100mbps connection which is stable. Yes way more than adequate cooling. This thing will run 30 minutes and stop or 32 hours and stop. No rhyme or reason i see. If there is a code to auto restart after an hour that would be gold. Any help? Any ideas? This depends on whats going on with your setup. First are you running the latest bfgminer binaries I release that have auto-restart built into the miner? Sounds like your not since restarting bfgminer pretty much does what I already do in the new driver release automatically. i have 5.4.2. isnt that the newest one? i see no restart timer You should see 5.4.2-38 at the top if your running the latest version. |
|
|
|
bfgminer.exe --scrypt -o stratum+tcp://eu.multipool.us:3351 -u Moonlander2.1 -p x,d=256 -S MLD:all --set MLD:clock=796
pause Ok so i have it running on litecoinpool and on multipool. Works but i run into the same error. After a while it just stops hashing. If i restart bfg then it works like a champ again. So can someone help me. Is there a way to set bfgminer up to auto restart after an hour. yea I know its not perfect but at this point i just want the dang thing to run nearly 24 hours. I hate waking up to find it pooped out after 30 minutes and sat idle 6 hours. Is there a way to add a piece to this scrypt that will make it auto restart after an hour? Since I have no clue why it keeps stopping. And before we trouble shoot again just assume its all running on the right stuff. yes powered hub. yes 100mbps connection which is stable. Yes way more than adequate cooling. This thing will run 30 minutes and stop or 32 hours and stop. No rhyme or reason i see. If there is a code to auto restart after an hour that would be gold. Any help? Any ideas? This depends on whats going on with your setup. First are you running the latest bfgminer binaries I release that have auto-restart built into the miner? Sounds like your not since restarting bfgminer pretty much does what I already do in the new driver release automatically. |
|
|
|
|
Haha wow, not only is Bitmain pumping out ASICs for every algorithm out there, they are now also CREATING coins that they can sell ASICs to in the future.
|
|
|
|
THIS IS THE SUPPORT THREAD: Keep this thread on topic! ..... BFGMINER 5.4.2 Instructions I have built a native bfgminer driver binaries with support all major systems and architectures. Please follow the GitHub release link below, which has binaries for each system attached. bfgminer download: https://github.com/jstefanop/bfgminer/releases/tag/bfgminer-5.4.2-futurebit2-beta2You will also need to install the latest version of Silicon Labs VPC drivers for Win/Mac OS link below. If you downloaded this driver below, make sure you update to the latest version, as the Moonlander 2 uses a new UART chip and the old driver will not work will with it. UART VPC Driver: https://www.silabs.com/products/development-tools/software/usb-to-uart-bridge-vcp-driversMAC OS 10.13 HIGH SIERRA USERS: Sililabs drivers will NOT work with the newest version of OS X, and you cannot run the stick with this version until Sililabs provides a driver update! EDIT: Workaround here https://bitcointalk.org/index.php?topic=2420357.msg25153407#msg25153407Once you have installed the drivers and extracted bfgminer binary for your system just plug in the miner to a USB port and double click on the Start_Moonlander2 file on your Mac/PC. The driver will auto-detect the board and start hashing at the default frequency (for linux you need to execute the .sh file in terminal with sudo command i.e. sudo start_moonlander2.sh). Just edit that same file with a text editor to change pools, add bfgminer options, and change frequency. You can also use the -c flag and load it with your own config file. ...... Hello Jstefanop, I need to run my Moonlanders by a live 32bit Linux running on an old Pentium III based PC, yes, very old. This old PC has not a HD, it boots from a memory stick. It is a live distribution but it is persistent so I can install software and it will be saved as it was with an HD. CGminer for other ASICs (Gridseed) worked very well but here I see there is not a 32 bit BFGminer 5.4.2 How can I set it up? Should I compile it from source files? I only pre-compile 32bit version for windows. Linux you would have to do a native compile on your system. Its very easy for linux, and you can actually use that install scrypt posted above by one of the community members. Its designed for Pi's, but should work fine for your linux system as well. |
|
|
|
I ordered 2 Moonlander 2 and have them hashing. The problem is that no matter what I do one of the sticks have an error rate of 5-10%. I've tried modifying clock settings, different USB ports, different computers, powered USB hubs, etc.
One stick is <.5% error rate and the other is 5-10% consistently. I've gone as low at 600MHz and as high as 796 each with similar results.
I've purchased directly from the developer on eBay. Not sure if I have a defective unit or I need to tweak the pots maybe?
Interested to hear any ideas on how to troubleshoot this.
Try raising the memory and core voltage. Also some units do show higher error rate, but does not actually affect the hashrate (i.e. the uart controller sometimes spits out garbage, but all the "real" shares still come out properly). Are you also seeing a corresponding 5-10% drop in hashrate? So the plot thickens (well not really... I think I just have a bad device). These devices shipped with the following voltages Core: 0.714VDC Mem: 0.771VDC. I changed them to their supposed defaults listed in the first post and was able to get it to Core: 0.751 Mem: 0.854. Now the error rate is consistent with the other 2 devices, except I constantly get these errors after a few hours: [2018-02-03 10:45:24] FutureBit Write error: Permission denied [2018-02-03 10:45:24] MLD 1: ASIC has stopped hashing, attempting to restart [2018-02-03 10:45:24] FutureBit Write error: Permission denied I've dropped the frequency from 756 (where the others are running happily) to 600 and all steps in between. I've been troubleshooting this for days and keep ending with the same result no matter which USB port or hub I use. I've run this thing solo, with the other two, basically every combination I've tried gets these errors. Not sure what else to try at this point. what hub and what pc or rasp pi? I want to stress that my setup is at 832 and just runs and runs and runs and runs https://www.youtube.com/watch?v=d3rqTUNt5XII am having the same issue, was there ever a solution? [2018-02-03 10:45:24] FutureBit Write error: Permission denied This error is because the miner temporarily lost contact with the USB port, this happens on cheaper hubs...especially if you have the sticks vertical mounted. The vibrations from the fans can cause the port to lose contact with the pins inside the USB connector. Unfortunately a USB disconnect is not recoverable and it would need to be unplugged/replugged. |
|
|
|
BITMAIN killed this little company .. Definitely not dead, I’ll have something quite interesting in a few weeks |
|
|
|
...
Nice looking setup. Personally, I would only run 5 off that hub so you can run them full speed. If you only have on one hub then you could plug you Pi and extra fans into the hub too because they don't need that much power. Have you tried to get bfgminer to run at startup? I am new to pi and Linux in general and have not been able to get it to work.
Nah I dont really need to been using pi's for years and they seldom crash, I use ssh and tmux to control it. that theres my trusty old pi2 with a broken hdmi out I'm sure you know what ssh is tmux may be a mystery to you it allows you to run a program through ssh and it'll keep running even after you disconnect. its similair to screen but a bit more advanced/configurable Yes, I tried screen but it was not getting me anywhere. Besides I just want it to autostart every time the pi boots up. make sure screen is installed (sudo apt install screen) create a file in your home directory with nano like this: inside this file put something like this cd /home/YOUR_PI_ACCOUNT_NAME/YOUR_BFGMINER_FOLDER_NAME
screen -dmS miner bash -c './start_moonlander.sh; exec bash' then turn startup script to executible and add this to crontab go down after all the ## and add this line @reboot /home/YOUR_PI_ACCOUNT_NAME/startup.sh then once rebooted just do and you should see bfgminer running. Press control then A then D at the same time to exit screen without terminating process. That should do it |
|
|
|
Hmmm. Lots of confusing signals here. For starters Intel already makes GPU's, just not high end ones. But pretty much 90% (or maybe even more) of the mobo integrated GPU's are Intel. The exception, AMD CPU boards with AMD integrated GPU's are arguably even worse. I bought an AMD CPU\GPU integrated PC for work and even though the GPU model was decent as a stand-alone (it's a few years old now but back when I got it, the stand-alone card was decent), in the integrated version that GPU sucks really bad. I regretted it immediately.
Also, the recent Intel\AMD partnership with the integrated AMD GPU has been regarded everywhere as a huge success, even this article talks about it in very positive terms. Why would Intel go in the opposite direction after that?
Unless your talking about AMDs APU's from years ago, AMDs current APUs they just released with vega cores blows anything Intel has out of the water. |
|
|
|
Wow, bastards took out nice polymer caps for aluminum ones. That is saving them like 25 bucks per board. |
|
|
|
|
Show Posts
