[ANN] rhminer - an optimized CPU+GPU miner for PascalCoin

[polyminer1]

Official Release
 
rhminer is an optimized miner for RandomHash2 POW algo used by PascalCoin

Features:
Support Intel/AMD 64 bit CPU.
Support stratum and solo mining
Provide simple api.
Works on Windows 7,10 and Ubuntu 18, MacOS.

Pool vs Solo mining
Since solo mining is 45% faster than pool mining, It is best to mine locally at the moment.
Pool mining is supported but no pool exists as of now.

Tested on
CPU: I3, I5, Core2, Xeon, Athlon

Links:
Documentation http://github.com/polyminer1/rhminer
Download on github https://github.com/polyminer1/rhminer/releases/
Download Windows binaries on Mega https://mega.nz/#F!DqpAjCJQ!Q12a_YRlu_CWA92kIglKug
Download Linux binaries on Mega https://mega.nz/#F!Dz4ElAwK!gbWbU4OpmEf6YnOCLIKfSQ


History
Version 2.3 - 3 Apr 2020
*General optimizations from 5% and up, depending on cpu and mem speed
*Default mining cpu count is *All CPU* instead of 1
*Remove requirement of command line options -cpu and -cputhreads
*Network stability improved !

Version 2.2 - Feb 6 2020
New optimizations up to 55% faster.
*Added major optimizations. (55% on Platinuim, 50% on i7 and i5)
*Tested on i5, i7, Epic and Xeon
*Added version in log and submit chrono
cheers Polyminer1

Version 2.2b - Removed pascalpool.org from DevPool list
*Improved pool connection
*code cleanup and minor improvements (linux and macos)

Version 2.1c - Feb 3 2020
** Linux only **
*Fixed miner stalling after 24+ hours
*Fixed miner crashing when resizing terminal
*Fixed screen -r making miner stop
*Fixed terminal randomly closing
*Fixed linux caching error that give out litle more than 2% more speed.
Note: this is a temp version.
Validated by tclark70


Version 2.1 - 22 Jan 2020
* Optimized RandomHash2 with up to 80% more speed. (On some system rhminer 2.1 is up to 250% faster than official miner)
* Fixed linux binary where main thread was exiting after miner startup.
* Linux zip files now contain rhminer as an executable and not a simple file.
* FIxed windows packages and removed extra folders

Version 2.0 - Jan 17 2020
* Added support for RandomHash2
* Dropped support for GPU

Version 1.5.3 - Jul 8 2019
* Added MacOS support and binaries. (Tested on High Sierra and El Capitan)
* Added internal cpu throttling option to slow down mining.
* Fixed: Duplicate shares when pool mining.
* Fixed: EthMan API : sent config.txt was ignored when restarting miner on Linux.

Version 1.5.1 - 14 may 2019
* Support for CUDA 10.1 with drivers 419.x(Windows) and 418.x(Linux)
* Added support for Nvidia Turing architecture (GTX 20xx) Untested.
* Fixed : Cannot Disable gpu mining when using config.txt with gpu binaries.
* Now you can simply empty the value of "gpu" and "gputhreads" to disable gpu mining on gpu binaries. Ex: "gpu":"" and "gputhreads":""
* Little performance boost on linux binaries (CPU and GPU)
* Added missing screensaver source files
* change line ending in many files

Version 1.5 - 2 April 2019
* Added ETHman and AwesomeMiner support (miner_getstat1, miner_restart, miner_reboot, miner_file, control_gpu)
  It is strongly recommented to NOT put rhminer in a loop in a script if you intend to use EthMan or AwseomeMiner.
  The restart option will close the instance and restart a new one. So if rhminer is in a loop, it will start another new instance every time.
* Added Pascal Coin ScreenSaver V1.0 - Windows
* Now starting rhminer without parameters will automatically use config.txt.
* Added support Maxwell architecture 5.3 (Tegra Tegra (Jetson) TX1 / Tegra X1) on windows
* Re-added Pascal architecture 6.0 (Quadro GP100, Tesla P100, DGX-1) that was mistakenly removed in 1.4
* Re-added Pascal architecture 6.2 ( Integrated GPU on the NVIDIA Drive PX2, Tegra (Jetson) TX2) that was mistakenly removed in 1.4


Version 1.4 - 3 march 2019
*Major gpu optimizations:
    Gtx 1060 3gb 280 threads +196%
    Gtx 1070 8gb 384 threads +218%
    gtx 950 2gb 140 threads +226% [linux]
* Fix : On linux, rhminer stall for ever when network is lost (cable unplug or switch/router crashes).
* Fix : On very rare occasions, rhminer would stall at startup.
* Fix : Opensource pool bug where work-ID stuck to the same ID, creating many invalid shares.
* Fix : Duplicate/stale shares was high on cpu with with core count.
* Add -configfile option with a config.xml example
* Add extra option (+) to -cputhreads to enable mining on more threads that system's core/hyperthreads count (ex: on a i5 with 4 core -threadcount +7 will mine on 7 threads instead of 4)


Version 1.3 - 11 Jan 2019
*Major CPU optimizations. Up to 260% on some CPU
* +155% on Xeon E52665
* +260% on Intel i5 2th gen
* +195% on Intel i5 4th gen
* +250% on Intel Core 2
* New commandline option -sseboost. This option will give up to 10% on some cpu. BUT can also make it slower on other. TEST it before using it, to assert it gives a boost to your cpu.

Version 1.2 - 13 dec 2018
* Major optimization on cuda miner (+32% on gtx 1070 8gb, +37% on gtx 1060 3gb, +40% on gtx 950)
* Simple optimization on Linux. Up to 5% depending on the cpu

Version 1.0 - 30 Nov 2018
* Fixed network error with coinotron and f2pool where miner was in limbo after pool disconnected.
* Fixed miner not starting on cpu without SSe4.1
* Fixed error where miners was rejecting shares due to wrong work package data.
* Fixed 'Exit on too many rejected shares' that was not working properly
* Added SSe detection
* Fixed connection error when coinotron and f2pool randomly closing socket.
* Minor memory optimizations

Dev donations:
Default donation is 1%.
Donation is hard-coded in the binaries downloadable on gitgub.
To disable donation download and compile locally, then use the -disabledevfee option.

Contact:
Discord channel : https://discord.gg/Egz2bdS
Twitter https://twitter.com/polyminer1

[tlaskows]

Following the development.

This miner is already unbelievably fast already on certain systems.

Thank you for your hard work.

[polyminer1]

Mining locally/Solo
To mine locally/solo you'll need the official PascalCoin wallet https://github.com/PascalCoin/PascalCoin/releases
In order to mine locally with rhminer, You need to set a miner name smaller than 26 characters and mine into a private key with encryption type secp256k1
The best way to assure you're mining in a secp256k1 private key is to create one and select it in the option "Always mine with this key".
Do not use the "Use random existing key" option because if one of your key is not a secp256k1 key, the miner will exit when. Plus when there is to much keys in the wallet it gives out errors, sometimes, when submiting nonces
To ensure your miner name is correct, go to Project menu, then Options and set a miner name smaller than 26 characters

Solo mining examples:

For Test net solo mining :  rhminer.exe -v 2 -r 20 -s http://127.0.0.1:4109 -cpu -cputhreads 1 -gpu 0 -gputhreads 100 -extrapayload HelloWorld
For Main net solo mining :  rhminer.exe -v 2 -r 20 -s http://127.0.0.1:4009 -cpu -cputhreads 1 -gpu 0 -gputhreads 100 -extrapayload HelloWorld

NOTE: remove -gpu 0 -gputhreads 100 if you dont have a gpu
Note2: It is not recommended to mine using a laptop.

Gpu mining
To mine using gpu you must provide the gpu numbers and the amount of threads for each gpu.
If you only have one gpu, use -gpu 0 and -gputhreads {amount of threads}
If you have more than one gpus, you can see their number by executing the miner with the list option :

C:>rhminer -list

  rhminer v0.9 beta for CPU and NVIDIA GPUs by polyminer1 (http://github.com/polyminer1)
  NVIDIA CUDA SDK 9.2

  Donations : Pascal account 529692-23
  Donations : Bitcoin address 19GfXGpRJfwcHPx2Nf8wHgMps8Eat1o4Jp

List of gpus and cpus:
GPU0 : GeForce GTX 1060 3GB
GPU1 : GeForce GTX 1060 3GB
GPU2 : GeForce GTX 950 2GB
CPU  : Intel(R) Core(TM) i5-4460 CPU @ 3.20GHz
Then select the ones you want to mine with like this :

rhminer -s http://localhost:4009 -gpu 1,2 -gputhreads 262,102

Ideal CUDA threads count
To find the ideal maximum amount of threads, start with 75% of the memory divided by 8.8.
For a GTX 1060 3GB that is 3000 * 0.75 / 8.8 = 255 threads.
Then run 2 minutes and if everything is stable, raise by say 32 until you get no crashes after 2 min.
To help you in that process, look for the log line that say "CUDA: Using " when the miner starts. It will indicate how much memory you take and how much is left depending on your selected thread count.
ALLWAYS let at lease 150 meg of free memory, for internal OS operations, or you have stability issues.

[tlaskows]

I hope I didn't kill the TESTNET.

I ran few old computers overnight.  The AVG HR stabilised.

I use Linux, but those instructions are the same.  I just put the miner parameters in a small bash script.

Is there a way or need the remove the appended .rhminer. in an instance?  I don't mind it, just not sure if the miner name length has anything to do with it.

Cheers.

[polyminer1]

I hope I didn't kill the TESTNET.

I ran few old computers overnight.  The AVG HR stabilised.

I use Linux, but those instructions are the same.  I just put the miner parameters in a small bash script.

Is there a way or need the remove the appended .rhminer. in an instance?  I don't mind it, just not sure if the miner name length has anything to do with it.

Cheers.

Yeah next release I will fix that, "rhminer" will be there if there is room after extrapayload

[tlaskows]

I hope I didn't kill the TESTNET.

I ran few old computers overnight.  The AVG HR stabilised.

I use Linux, but those instructions are the same.  I just put the miner parameters in a small bash script.

Is there a way or need the remove the appended .rhminer. in an instance?  I don't mind it, just not sure if the miner name length has anything to do with it.

Cheers.

Yeah next release I will fix that, "rhminer" will be there if there is room after extrapayload

Thanks.  It doesn't bother me.  It's just that the extra payload gets overwritten with stuff.
Other devs for miners had their names/nicks hardcoded.

I looked at the source code.  Didn't bother recompiling it with a different name.   

Cheers.

~Thomas

[tlaskows]

@polyminer1

Are those "Ryzen 1800X @ 4GHz, 3200MHz" HRs posted on your GitHub tested personally?

I have a Ryzen 2700x and it's a really odd CPU.  I get around the same HR as that 1800x, no matter what settings.

It seems to require special DDR4 RAM.  I pulled half the RAM from my Xeon workstation.  It works, but no way it will go to 3200.  1866 max.
Anything higher and it will not post.

The Ryzen is the worst tech/CPU I have dealt with in 18 years.  My opinion.  I don't get the point of marketing it as an unlocked chip when it will not go over 4.1 GH/z without melting down.  My old 4770ks hit 4.5 easily from stock of 3.5.

Thanks.

[polyminer1]

@polyminer1

Are those "Ryzen 1800X @ 4GHz, 3200MHz" HRs posted on your GitHub tested personally?

I have a Ryzen 2700x and it's a really odd CPU.  I get around the same HR as that 1800x, no matter what settings.

It seems to require special DDR4 RAM.  I pulled half the RAM from my Xeon workstation.  It works, but no way it will go to 3200.  1866 max.
Anything higher and it will not post.

The Ryzen is the worst tech/CPU I have dealt with in 18 years.  My opinion.  I don't get the point of marketing it as an unlocked chip when it will not go over 4.1 GH/z without melting down.  My old 4770ks hit 4.5 easily from stock of 3.5.

Thanks.

No these numbers where given by the community.
Best things is to wait for more of these and compare later on.
Ram speed is also limited by the mobo, so it may be that.
About Ryzens, they are "slower" than intel's when comparing with equal ram speed and cpu speed. I heard 30% slower. But their 40% cheaper, so in the end it's worth it.

[tlaskows]

@polyminer1

Are those "Ryzen 1800X @ 4GHz, 3200MHz" HRs posted on your GitHub tested personally?

I have a Ryzen 2700x and it's a really odd CPU.  I get around the same HR as that 1800x, no matter what settings.

It seems to require special DDR4 RAM.  I pulled half the RAM from my Xeon workstation.  It works, but no way it will go to 3200.  1866 max.
Anything higher and it will not post.

The Ryzen is the worst tech/CPU I have dealt with in 18 years.  My opinion.  I don't get the point of marketing it as an unlocked chip when it will not go over 4.1 GH/z without melting down.  My old 4770ks hit 4.5 easily from stock of 3.5.

Thanks.

No these numbers where given by the community.
Best things is to wait for more of these and compare later on.
Ram speed is also limited by the mobo, so it may be that.
About Ryzens, they are "slower" than intel's when comparing with equal ram speed and cpu speed. I heard 30% slower. But their 40% cheaper, so in the end it's worth it.

Thank you.

I have some numbers too.  Clock for clock, AMD is still way slower and less compatible, runs hot... (this is my first AMD CPU since the Athlon XP 1800+).
I have been using Intel CPUs only.  I thought I would give AMD another chance.

BTW.  I switched to Linux on that 2700x.  Still around the same HR.  ~1360 average @ 4.0GHz all cores.

[tlaskows]

@Polyminer1

I got this error just today.  The miner ran fine for 3 days.

Code:

Net   13:09:18   Error: Error. Deamon/Wallet miner name is to long. Set a name under 26 caracters.
NOTE, if this error persist, just restart the demaon/wallet.

All my instances (2 or so) exited.

I did not change any settings.

I noticed a number of odd transaction on TESTNET.  Wrong miner rewards, etc...

It probably crashed, or maybe it's still buggy.

[polyminer1]

@Polyminer1

I got this error just today.  The miner ran fine for 3 days.

Code:

Net   13:09:18   Error: Error. Deamon/Wallet miner name is to long. Set a name under 26 caracters.
NOTE, if this error persist, just restart the demaon/wallet.

All my instances (2 or so) exited.

I did not change any settings.

I noticed a number of odd transaction on TESTNET.  Wrong miner rewards, etc...

It probably crashed, or maybe it's still buggy.

See the guide for solomoning here https://github.com/polyminer1/rhminer#mining-locallysolo
Basicaly, you have to restart the wallet once you change the name of your miner.

[tlaskows]

@Polyminer1

I got this error just today.  The miner ran fine for 3 days.

Code:

Net   13:09:18   Error: Error. Deamon/Wallet miner name is to long. Set a name under 26 caracters.
NOTE, if this error persist, just restart the demaon/wallet.

All my instances (2 or so) exited.

I did not change any settings.

I noticed a number of odd transaction on TESTNET.  Wrong miner rewards, etc...

It probably crashed, or maybe it's still buggy.

See the guide for solomoning here https://github.com/polyminer1/rhminer#mining-locallysolo
Basicaly, you have to restart the wallet once you change the name of your miner.

I restarted the wallet and it works again.
I do not recall changing the name of my node.  It's possible that I did couple days ago.

Thanks.

[chup]

Official Release
 

Dev donations:
Default donation is 1%.
Donation is hard-coded in the binaries downloadable on gitgub.
To disable donation download and compile locally, then use the -disabledevfee option.

Can You explain this devfee duration:

[thedreamer]

Is it normal for rhminer running under windows to never have any accepted shares sent when mining locally?

I am running the latest Pascal Wallet on the same PC locally, up to date, run the RHminer, no errors, it gets the work as I see it in the logs, but
the miner always shows '0' accepted or rejected shares. The Server gets about 900h/s and is CPU only.

[polyminer1]

Official Release
 

Dev donations:
Default donation is 1%.
Donation is hard-coded in the binaries downloadable on gitgub.
To disable donation download and compile locally, then use the -disabledevfee option.

Can You explain this devfee duration:

https://i.imgur.com/fkuxdoA.png

That is pretty fuckedup and scary. It's not normal at all. Do you had a log ? (-lofilename)
But so you knwo I was hit by an attack yesturday about 12H ago for 8h straight. May be due to that.
But this log scares me alot, i'll add awatchdog in the miner so this never happen again.

[Truthchanter]

Mine keeps crashing after 20 minutes or so i think.. I have the error message in the log I will get it soon

EDIT: heres the crash error in the log

Net   10:42:45   Error. Assert '(m_fullHeader.size() == PascalHeaderSize)'

[tlaskows]

Mine keeps crashing after 20 minutes or so i think.. I have the error message in the log I will get it soon

I am mining in a pool.

Several times I got invalid response end of line character.
Every few hours, not 20 minutes.

I just restart it.  Polyminer did mention keep it running in a loop or increase the -r times.

Cheers.

[chup]

That is pretty fuckedup and scary. It's not normal at all. Do you had a log ? (-lofilename)
But so you knwo I was hit by an attack yesturday about 12H ago for 8h straight. May be due to that.
But this log scares me alot, i'll add awatchdog in the miner so this never happen again.

Not scary at all. But memorable. If there is no auto logging, than I do not have logs. BTW I think my provider is changing the IP quite often. At my home twice a day.

[mrpcrp2]

Good day dev.
Try to use this miner some hours on diffrent rigs, unfortunately well working only on 750Ti and 1060 3Gb rigs .
On 1060 6gb and 1070 working only on a half of card memory - 262 threads max for 1060 6gb and 340 threads for 1070 , increasing amount of threads immediately crash miner after start with Cuda unknown error message
All rigs with latest nvidia drivers.

[biospb]

miner not properly handle connection looses and stop working after few hours.
It stops to receive new jobs after ~3h on most rigs and works without any profit until manual restart. (windows and linux)

also it can not set cpu affinity and priority on linux. very bad. does not report cpu temp...
I tried it on gpus and as it was mentioned above - it sees only half memory on 1060 6gb.
also it completely hanged rig with 13 gpus. but if it works - it would give only 1,5kH. not too much


I think I will add someday randomhash to cpuminer-opt and be happy)

anyway - thank you for your work.