[ANN] ccminer 2.3 - opensource - GPL (tpruvot)

[santan]

I was not able to compile with 19 gpu's, i've got a lot of errors. 16 gpu's version compiled without any troubles.
I htink tpruvot must check this situation. 

You need to modify 2 more files...... check here....  https://github.com/tpruvot/ccminer/pull/54
It build without any error for me..... But i don't have 19 GPU rig to test....
Please try and let us know.

[bensam1231]

I'VE NOTICED THAT INTENSITY SETTINGS DON'T WORK--

In the most recent versions of CCminer (2.2+), intensity settings appear to be ignored, and a default setting is applied.  Other users have posted about not being able to disable auto-tuning on algos like Scrypt pr CryptoNight.  

Frankly, default or auto-tune settings are often the best, but I have noticed that some older tuning parameters are ignored.       --scryptr

I have only been wrestling with ccminer for the last 4 hours, and can tell you for certain, every combination of INTENSITY parameter I threw at it, was ignored.  The default settings are NOT the fastest for my cards.  I have also been spending this whole time trying to learn this crazy "launch-config" parameter, digging through documents to find which of the 1,000 different technical specifications ccminer needs for me to accept.  It's parameter list is confusing to say the least.

Intensity is hardcoded in implementations of many algos (for ccminer 2.2.4).
It's a good thing to avoid you posting on youtube that your rig is on fire

I DON'T POST ON YOUTUBE--

But I do enjoy custom-tuning my rigs.  Perhaps Epsylon3 could "star" (mark with asterisk) the algorithms affected in his documentation. Then, make a note of explanation.  

Scrypt, Scrypt-N, BoolBerry, CryptoNight all had difficult-to-master tuning parameters.  Even NeoScrypt could be trouble as the code changed with optimization.  Spending an hour on un-necessary, non-effective tuning could be avoided with a few documentation notes.       --scryptr

Agreed, also intensity will NOT cause hardware failures. If you have a system that starts on fire as long as the GPU is operating within spec you've done something wrong. Mining software has never once been responsible for any miner fires, it's always people using inferior hardware or halfhazard setups, like daisy chaining splitters. Hardware will fail regardless as manufacturers usually don't test them super rigorously, which is why one of the secrets of mining is knowing the right brands to buy.

I find it hard to believe Epsylon would intentionally hardcode in a intensity value unless it's to artificially reduce performance of his miner. Tweaking intensity has been a staple of good mining for a long time.

[cyberdeath]

I was not able to compile with 19 gpu's, i've got a lot of errors. 16 gpu's version compiled without any troubles.
I htink tpruvot must check this situation. 

You need to modify 2 more files...... check here....  https://github.com/tpruvot/ccminer/pull/54
It build without any error for me..... But i don't have 19 GPU rig to test....
Please try and let us know.

Thanks, just compiled x32 x64 versions without any troubles.
But haven't tested yet< because i have removed cards from rig.

[scryptr]

Guys, is this legit? https://github.com/MSFTserver/ccminer/releases

It seems to be released by tpruvot but on MSFTserver repository... and description doesn't seem to be from him. What do you guys think?

I CLONED THE REPOSITORY AND IT COMPILED--

Afterwards, I mined the new RavenCOIN (RVN) algorithm, x16r, on "yiimp.eu:3666".  The binary for the wallet-qt  synchronized rapidly on my Win7 x64 system, and received coins on payout.

CCminer 2.2.5 compiled on my Ubuntu 14.04 system under CUDA 8 (with both current patches).  Intensity is adjustable; I had to lower it to 16 for CCminer 2.2.5 to run on my 6x 750ti rig.  The default intensity is 19.

The notes read that CCminer 2.2.5 is a CUDA 9 creation.  There is no CUDA 9 toolkit for Ubuntu 14.04.  My 750ti rig kept crashing with a memory error in "line 41".  I may post a screen capture of the error later.

Does anybody know how to squeeze CUDA 9 onto Ubuntu 14.04?  Is there some code I could change in "line 41"?  Thanks...      --scryptr

[cyberdeath]

I was not able to compile with 19 gpu's, i've got a lot of errors. 16 gpu's version compiled without any troubles.
I htink tpruvot must check this situation.  

You need to modify 2 more files...... check here....  https://github.com/tpruvot/ccminer/pull/54
It build without any error for me..... But i don't have 19 GPU rig to test....
Please try and let us know.

Thanks, just compiled x32 x64 versions without any troubles.
But haven't tested yet< because i have removed cards from rig.

Ccminer is very unstable with 19 gpu's and die very fast on lyra2z and skein algo, so ccminer is not usable with 19gpu's. I think only tpruvot can find solution(((
Strange thing, but on both algos miner dies on gpu0 but in case when i am using 16 gpu config, there is no troubles with gpu0 on ccminer x64 ver and x32 ver have different gpu troubels. In my situration gpu0 is p104-100

GPU #0: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #1: SM 6.1 Gigabyte GTX 1080 Ti @ 1633 MHz (MEM 5505)
GPU #2: SM 6.1 ASUS GTX 1060 3GB @ 1785 MHz (MEM 4004)
GPU #3: SM 6.1 MSI GTX 1080 @ 1823 MHz (MEM 5005)
GPU #4: SM 6.1 ASUS GTX 1060 3GB @ 1785 MHz (MEM 4004)
GPU #5: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #6: SM 6.1 ASUS GTX 1060 3GB @ 1785 MHz (MEM 4004)
GPU #7: SM 6.1 MSI GTX 1080 @ 1823 MHz (MEM 5005)
GPU #8: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #9: SM 6.1 MSI GTX 1080 @ 1823 MHz (MEM 5005)
GPU #10: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #11: SM 6.1 MSI GTX 1080 @ 1823 MHz (MEM 5005)
GPU #12: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #13: SM 6.1 Gigabyte GTX 1080 Ti @ 1633 MHz (MEM 5505)
GPU #14: SM 6.1 Gigabyte GTX 1080 Ti @ 1633 MHz (MEM 5505)
GPU #15: SM 6.1 Gigabyte GTX 1080 Ti @ 1633 MHz (MEM 5505)
GPU #16: SM 6.1 MSI GTX 1080 @ 1823 MHz (MEM 5005)
GPU #17: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #18: SM 6.1 Gigabyte GTX 1080 Ti @ 1633 MHz (MEM 5505)

Tried zm miner 0.5.8 it is awesome and works well. Tried ewbf cuda miner it is peace of shit.

[chrysophylax]

I was not able to compile with 19 gpu's, i've got a lot of errors. 16 gpu's version compiled without any troubles.
I htink tpruvot must check this situation.  

You need to modify 2 more files...... check here....  https://github.com/tpruvot/ccminer/pull/54
It build without any error for me..... But i don't have 19 GPU rig to test....
Please try and let us know.

Thanks, just compiled x32 x64 versions without any troubles.
But haven't tested yet< because i have removed cards from rig.

Ccminer is very unstable with 19 gpu's and die very fast on lyra2z and skein algo, so ccminer is not usable with 19gpu's. I think only tpruvot can find solution(((
Strange thing, but on both algos miner dies on gpu0 but in case when i am using 16 gpu config, there is no troubles with gpu0 on ccminer x64 ver and x32 ver have different gpu troubels. In my situration gpu0 is p104-100

GPU #0: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #1: SM 6.1 Gigabyte GTX 1080 Ti @ 1633 MHz (MEM 5505)
GPU #2: SM 6.1 ASUS GTX 1060 3GB @ 1785 MHz (MEM 4004)
GPU #3: SM 6.1 MSI GTX 1080 @ 1823 MHz (MEM 5005)
GPU #4: SM 6.1 ASUS GTX 1060 3GB @ 1785 MHz (MEM 4004)
GPU #5: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #6: SM 6.1 ASUS GTX 1060 3GB @ 1785 MHz (MEM 4004)
GPU #7: SM 6.1 MSI GTX 1080 @ 1823 MHz (MEM 5005)
GPU #8: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #9: SM 6.1 MSI GTX 1080 @ 1823 MHz (MEM 5005)
GPU #10: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #11: SM 6.1 MSI GTX 1080 @ 1823 MHz (MEM 5005)
GPU #12: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #13: SM 6.1 Gigabyte GTX 1080 Ti @ 1633 MHz (MEM 5505)
GPU #14: SM 6.1 Gigabyte GTX 1080 Ti @ 1633 MHz (MEM 5505)
GPU #15: SM 6.1 Gigabyte GTX 1080 Ti @ 1633 MHz (MEM 5505)
GPU #16: SM 6.1 MSI GTX 1080 @ 1823 MHz (MEM 5005)
GPU #17: SM 6.1 P104-100 @ 1734 MHz (MEM 5005)
GPU #18: SM 6.1 Gigabyte GTX 1080 Ti @ 1633 MHz (MEM 5505)

Tried zm miner 0.5.8 it is awesome and works well. Tried ewbf cuda miner it is peace of shit.

It sounds to me like your intensities may be completely off.

A huge mix of cards means that they need to be set individually, even though ccminer-tpruvot may try and auto adjust these intensities for each card.

Just a thought, but it is what it seems like.

#crysx

[cyberdeath]

The same config with 16 cards + without oc work well. so i don't think that there is trouble with intesivity.

[pallas]

Guys, is this legit? https://github.com/MSFTserver/ccminer/releases

It seems to be released by tpruvot but on MSFTserver repository... and description doesn't seem to be from him. What do you guys think?

I CLONED THE REPOSITORY AND IT COMPILED--

Afterwards, I mined the new RavenCOIN (RVN) algorithm, x16r, on "yiimp.eu:3666".  The binary for the wallet-qt  synchronized rapidly on my Win7 x64 system, and received coins on payout.

CCminer 2.2.5 compiled on my Ubuntu 14.04 system under CUDA 8 (with both current patches).  Intensity is adjustable; I had to lower it to 16 for CCminer 2.2.5 to run on my 6x 750ti rig.  The default intensity is 19.

The notes read that CCminer 2.2.5 is a CUDA 9 creation.  There is no CUDA 9 toolkit for Ubuntu 14.04.  My 750ti rig kept crashing with a memory error in "line 41".  I may post a screen capture of the error later.

Does anybody know how to squeeze CUDA 9 onto Ubuntu 14.04?  Is there some code I could change in "line 41"?  Thanks...      --scryptr

First, you should use cuda 9.1.
Second, why don't you use the latest LTS? 14.04 is so old!

[cyberdeath]

The same config with 16 cards + without oc work well. so i don't think that there is trouble with intesivity.

Checked also ccminer on 18gpu config situation is same.
Checked also nicehash excavator with 18 cards works well but with 19 gpu config it dies )

[scryptr]

Guys, is this legit? https://github.com/MSFTserver/ccminer/releases

It seems to be released by tpruvot but on MSFTserver repository... and description doesn't seem to be from him. What do you guys think?

I CLONED THE REPOSITORY AND IT COMPILED--

Afterwards, I mined the new RavenCOIN (RVN) algorithm, x16r, on "yiimp.eu:3666".  The binary for the wallet-qt  synchronized rapidly on my Win7 x64 system, and received coins on payout.

CCminer 2.2.5 compiled on my Ubuntu 14.04 system under CUDA 8 (with both current patches).  Intensity is adjustable; I had to lower it to 16 for CCminer 2.2.5 to run on my 6x 750ti rig.  The default intensity is 19.

The notes read that CCminer 2.2.5 is a CUDA 9 creation.  There is no CUDA 9 toolkit for Ubuntu 14.04.  My 750ti rig kept crashing with a memory error in "line 41".  I may post a screen capture of the error later.

Does anybody know how to squeeze CUDA 9 onto Ubuntu 14.04?  Is there some code I could change in "line 41"?  Thanks...      --scryptr

First, you should use cuda 9.1.
Second, why don't you use the latest LTS? 14.04 is so old!

I AM RUNNING MODIFIED ETHOS--

I have also installed Excavator and XMR-STAK.  EthOS is a customized 14.04 release, and it does well with few errors.  It is very stable and reliable.       --scryptr

[joblo]

Guys, is this legit? https://github.com/MSFTserver/ccminer/releases

It seems to be released by tpruvot but on MSFTserver repository... and description doesn't seem to be from him. What do you guys think?

I CLONED THE REPOSITORY AND IT COMPILED--

Afterwards, I mined the new RavenCOIN (RVN) algorithm, x16r, on "yiimp.eu:3666".  The binary for the wallet-qt  synchronized rapidly on my Win7 x64 system, and received coins on payout.

CCminer 2.2.5 compiled on my Ubuntu 14.04 system under CUDA 8 (with both current patches).  Intensity is adjustable; I had to lower it to 16 for CCminer 2.2.5 to run on my 6x 750ti rig.  The default intensity is 19.

The notes read that CCminer 2.2.5 is a CUDA 9 creation.  There is no CUDA 9 toolkit for Ubuntu 14.04.  My 750ti rig kept crashing with a memory error in "line 41".  I may post a screen capture of the error later.

Does anybody know how to squeeze CUDA 9 onto Ubuntu 14.04?  Is there some code I could change in "line 41"?  Thanks...      --scryptr

First, you should use cuda 9.1.
Second, why don't you use the latest LTS? 14.04 is so old!

I AM RUNNING MODIFIED ETHOS--

I have also installed Excavator and XMR-STAK.  EthOS is a customized 14.04 release, and it does well with few errors.  It is very stable and reliable.       --scryptr

I think you'd have to compile cuda 9 and a compatible driver for 14.04, but source code isn't open.

[firemylasers]

Welcome to the discussion thread for my ccminer fork.

https://imgur.com/BaLuHQy

What´s the boo and yes about?

"Boo" means that the share it's referring to was rejected (in this case the reason for the reject can be found on the next line). "Yes" means that the share it's referring to was accepted.

[joblo]

The same config with 16 cards + without oc work well. so i don't think that there is trouble with intesivity.

What does "die very fast" mean? How far it gets and how it dies is very important.
Is there an error message? Does it start hashing?

And how many is "16 cards +"? What is the maximum that work?

[p1r473]

Getting "does not validate on CPU" and "memory access" problems with Tribus on Tpruvot x86. Doesn't happen immediately, but if left on for a while, it starts to do it, maybe 5-10 min.
No overclock and latest Nvidia drivers, default intensity. 2 1080 TI.

[chrysophylax]

The same config with 16 cards + without oc work well. so i don't think that there is trouble with intesivity.

Remember also.

nVidia themselves have put on a hardcap of max GPU in their drivers for 'gaming' GPU's in one machine.

Seeing you have under the 16 'gaming' GPU and mining GPU also, this shouldn't be an issue.

We are about to build a few of these this coming week, with ALL 'gaming' GPU's, so I think we will cap our systems at 16 GPU for every miner.

We will be mining under the CWI name, so look out for that sort of performance when it comes about, as we will be testing both ccminer-tpruvot and CWIgm-0.9.9

#crysx

[cyberdeath]

The same config with 16 cards + without oc work well. so i don't think that there is trouble with intesivity.

What does "die very fast" mean? How far it gets and how it dies is very important.
Is there an error message? Does it start hashing?

And how many is "16 cards +"? What is the maximum that work?

i have tested 16 18 and 19 cards. 17 haven't tested. 16 works just fine and i have stoped experiments.
when it is 19 gpus it started hashing always after intensity sets and sometimes it even 45 seconds shows hashes of 4 or 5 cards.
when it was 18 gpus it die faster than 19gpu config in stage of intensity sets.
I can't now show error's because i have removed 3 cards. just ccminer crashed system was stable.

[cyberdeath]

The same config with 16 cards + without oc work well. so i don't think that there is trouble with intesivity.

Remember also.

nVidia themselves have put on a hardcap of max GPU in their drivers for 'gaming' GPU's in one machine.

Seeing you have under the 16 'gaming' GPU and mining GPU also, this shouldn't be an issue.

We are about to build a few of these this coming week, with ALL 'gaming' GPU's, so I think we will cap our systems at 16 GPU for every miner.

We will be mining under the CWI name, so look out for that sort of performance when it comes about, as we will be testing both ccminer-tpruvot and CWIgm-0.9.9

#crysx

i have used the latest drivers 23.21.13.9077
i have used 13 gaming and 6 mining because b250 mining expert mobo only boots in such configuration if you will try 14 gaming your mobo will not boot even.
you mean that i have to use one driver for gaming gpus and another for mining gpus? i just don't know how to do such idea.

[joblo]

The same config with 16 cards + without oc work well. so i don't think that there is trouble with intesivity.

What does "die very fast" mean? How far it gets and how it dies is very important.
Is there an error message? Does it start hashing?

And how many is "16 cards +"? What is the maximum that work?

i have tested 16 18 and 19 cards. 17 haven't tested. 16 works just fine and i have stoped experiments.
when it is 19 gpus it started hashing always after intensity sets and sometimes it even 45 seconds shows hashes of 4 or 5 cards.
when it was 18 gpus it die faster than 19gpu config in stage of intensity sets.
I can't now show error's because i have removed 3 cards. just ccminer crashed system was stable.

If it gets worse with more cards it's not likely a software issue, probably a resource issue like power.

[cyberdeath]

The same config with 16 cards + without oc work well. so i don't think that there is trouble with intesivity.

What does "die very fast" mean? How far it gets and how it dies is very important.
Is there an error message? Does it start hashing?

And how many is "16 cards +"? What is the maximum that work?

i have tested 16 18 and 19 cards. 17 haven't tested. 16 works just fine and i have stoped experiments.
when it is 19 gpus it started hashing always after intensity sets and sometimes it even 45 seconds shows hashes of 4 or 5 cards.
when it was 18 gpus it die faster than 19gpu config in stage of intensity sets.
I can't now show error's because i have removed 3 cards. just ccminer crashed system was stable.

If it gets worse with more cards it's not likely a software issue, probably a resource issue like power.

man i have enough power i am using server psu's, don't waste your time

[thin]

Ccminer is very unstable with 19 gpu's ..

I just wander why would anyone create system like that and get a  constant problems. Comparing to total system cost, an extra motherboard+cpu+memory would not be that big investment, but release complexity by times.