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

[polyminer1]

on i3 linux version is 25% slower than finminer... only 370HS vs 500HS ( 522HS 24h avg on pool )
I dont know how.  

Some cpu are faster with rhminer, some slower. You have to experiment. But one thing reported so far, rhminer is generally faster on cpu with hyperthreads.

[tlaskows]

Yes, I have had the pleasure to try out finminer.

On Linux.

Amazing how quickly the devs developed exact same SPEED as open source version of polyminer1's.

You know that they took his code without credit, right?

It's a virus.  And it's slow.  And it only supports dagger hashimo and randomhash few days later the fastest randomhash implementation was released.

Amazing coders or...

I looked to their "Company" and yeah their business model is to take opensource code, optimize it and charge for it. Explicitly preventing rightful owners of retributions or/and representation.
They wont stop until someone have the time to mount a case against them.

I sound like an asshole in my posts, but I am trying to defend your work here.
I imagine they took out the dev fee and "added their own".
You should be able to tell by looking at the global HR of randomhash.
That sort of shit should be illegal.  Nice looking website too with fake reviews (finminer).

Cheers.

[polyminer1]

I sound like an asshole in my posts, but I am trying to defend your work here.
I imagine they took out the dev fee and "added their own".
You should be able to tell by looking at the global HR of randomhash.
That sort of shit should be illegal.  Nice looking website too with fake reviews (finminer).

Cheers.

I know, tnx. And yes it it is illegal, it's just not really enforced. You need to mount a good case, then contact GPL foundation or CC0 foundation and they may appoint a lawyer to go further...

[percy_tc]

I have 450 hs/s on a Ryzen 1700x

any tips?

[polyminer1]

I have 450 hs/s on a Ryzen 1700x

any tips?

I would invite you on the PascalCoin's rhminer channel, there is plenty of peoples that can guide you specifically for the ryzen processors.

join here : https://discord.gg/RVcEpF9

Polyminer1

[polyminer1]

@all
rhminer V1.3 is out with major speed improvement (up to 260%).
If you want to enjoy competitive mining speed at low fee, with trusted and safe binaries, please download the latest rhminer on my github(or mega).
https://github.com/polyminer1/rhminer/releases/tag/1.3
https://mega.nz/fm/O24kAAbT

Cheers
Polyminer1

[tlaskows]

@all
rhminer V1.3 is out with major speed improvement (up to 260%).
If you want to enjoy competitive mining speed at low fee, with trusted and safe binaries, please download the latest rhminer on my github(or mega).
https://github.com/polyminer1/rhminer/releases/tag/1.3
https://mega.nz/fm/O24kAAbT

Cheers
Polyminer1

For real?  I was on your discord and you promised a surprise for next week.

Will check it out.   

Cheers.

[polyminer1]

yep.

[tlaskows]

yep.

I ran the miner for few hours.  It's somehow better than nanopool miner.  Same hashrate, but it doesn't kill the CPU and make it run as hot.

The displayed hashrate is lower only on one CPU, but my guess is the dev fee with the other miner has a lot to do with it.

4770k x 2 - same as the *other* miner.
Xeon ES v3 - same.
2700x is a bit slower, but it doesn't boil the chip.

POS core i3 mobile gets 320 H/s.

I have no idea what nanominer does, but it renders the system completely unusable.  With rhminer, I can still use the computers for normal tasks.

Cheers.

[polyminer1]

yep.

I ran the miner for few hours.  It's somehow better than nanopool miner.  Same hashrate, but it doesn't kill the CPU and make it run as hot.

The displayed hashrate is lower only on one CPU, but my guess is the dev fee with the other miner has a lot to do with it.

4770k x 2 - same as the *other* miner.
Xeon ES v3 - same.
2700x is a bit slower, but it doesn't boil the chip.

POS core i3 mobile gets 320 H/s.

I have no idea what nanominer does, but it renders the system completely unusable.  With rhminer, I can still use the computers for normal tasks.

Cheers.

Great. I'm happy it it work for you.
Fox xeon, you can try -sseboost 1, most of the time it add something.

[tlaskows]

yep.

I ran the miner for few hours.  It's somehow better than nanopool miner.  Same hashrate, but it doesn't kill the CPU and make it run as hot.

The displayed hashrate is lower only on one CPU, but my guess is the dev fee with the other miner has a lot to do with it.

4770k x 2 - same as the *other* miner.
Xeon ES v3 - same.
2700x is a bit slower, but it doesn't boil the chip.

POS core i3 mobile gets 320 H/s.

I have no idea what nanominer does, but it renders the system completely unusable.  With rhminer, I can still use the computers for normal tasks.

Cheers.

Great. I'm happy it it work for you.
Fox xeon, you can try -sseboost 1, most of the time it add something.

I get a steady 3400 H/s from that old Xeon w 24 threads (quad channel DDR4).  That's faster than the 2700x.

sseboost 1 forces sse2 from what I read in the help file?
I tried it on few CPUs.  It slowed down the miner.
My oldest CPU is from 2012.
memboost does make a pretty big difference - hyper-threading enabled.

Either way, it's better and faster in most cases than nanos**t closed source.  I'll run it for few weeks.

I even have it running on an Intel Atom x7 for the kicks.

Cheers.

[biospb]

polyminer1

first of all, Happy new year!  

second, you did it! on linux your faster 30% than *other* miner. on windows - about the same. (was 50% slower).
so something is wrong with windows binary. it can be much better. can you try to build it not with MSVC?

[tlaskows]

polyminer1

first of all, Happy new year!  

second, you did it! on linux your faster 30% than *other* miner. on windows - about the same. (was 50% slower).
so something is wrong with windows binary. it can be much better. can you try to build it not with MSVC?

The *other* miner was slower on Windows too.  It's not the miner.  Use Linux.
Windows is a disaster of an OS.  Bloated and uses your PC's resources in the background.

Look at the geekbench scores for any given CPU.  Linux is always on top of the list.  I compared it many times.  Same hardware, dual boot.

https://browser.geekbench.com/v4/cpu/multicore?dir=desc&page=1

I only use Linux now.  Windows pissed me off few too many times.

Cheers.

[polyminer1]

polyminer1

first of all, Happy new year!  

second, you did it! on linux your faster 30% than *other* miner. on windows - about the same. (was 50% slower).
so something is wrong with windows binary. it can be much better. can you try to build it not with MSVC?

The *other* miner was slower on Windows too.  It's not the miner.  Use Linux.
Windows is a disaster of an OS.  Bloated and uses your PC's resources in the background.

Look at the geekbench scores for any given CPU.  Linux is always on top of the list.  I compared it many times.  Same hardware, dual boot.

https://browser.geekbench.com/v4/cpu/multicore?dir=desc&page=1

I only use Linux now.  Windows pissed me off few too many times.

Cheers.

Happy new year to you to mate !

I'll look into it. VS compiler is somehow limited with regards of sse intrinsic optimizations.
I've been told you can install Windows Subsystem for Linux so you can run rhminer linux binaries on win10. It's said to be much faster.
I did not tried yet toh.

Cheers.

[biospb]

yes, I know. but win 10 is rarely used.
some devs build win binaries on ubuntu with cross compile - example

btw, there are simd asm implementations for some algos (sha2* from Intel! and etc).

and one more thing. in RandomHashCPUMiner.cpp there is check for diff

Code:

if (RH_swap_u32(*work) <= packageData->m_target)  
and so on with a lot of swaps

maybe it is better to swap target once and compare to non swapped work?

[eBit]

polyminer1,

CUDA call error in cudaSetDevice(m_deviceID) : CUDA driver version is insufficient for CUDA runtime version

Do I need to install the Cuda Toolkit for this on Windows 7?  Does it matter if it's v9.2 or v10?

[eBit]

Never mind. I just noticed a GPU can't match a cpu on this algorithm.

[tlaskows]

Never mind. I just noticed a GPU can't match a cpu on this algorithm.

Yes it can, but a lower end CPU.

300 H/s.  That's the same speed as a newer mobile almost anything dual core.  core i3 for example.

You need CUDA v 9.2 I believe.  I used it on Linux and the

Code:

root@mac-mini:~# apt install nvidia-cuda-toolkit

was required.

Hashcat uses CUDA/OpenCL.  The miner failed to start until I got hashcat to work.

[redmonski]

Here are some performance posted in discord - credit to tclarke.                                                               =================
CPU:ryzen threadripper 1950X @ 3.4 Ghz
Threads:32 
OS: Ubuntu
Memory: Quad DDR4 3200
Option: sseboost 1
Hashrate:5378 h/s

CPU:ryzen 2700 @ 3.4 GHz   
Threads: 16
OS: Ubuntu   
Memory: Dual DDR4 3200
Option: sseboost 1
Hashrate: 2856 h/s

[redmonski]

PascalCoin has 2 new open source mining pools.

https://pool.pascalpool.org/
https://pasc.c3pool.com/

Please consider supporting these two open source pools, 0% fee.

For update, PascalCoin is now using a new algo RandomHash - cpu friendly algo.

Happy mining!