Topic: Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v15.0 (Windows/Linux) - page 1109. (Read 6590718 times)

@claymore :
Sure, there is no reason to start talking about hashrate, as that depends completely from your configuration.

Tried to start your miner, to test on eth, I got this result ;

20:16:01:274   abf62840   cudaGetDeviceCount failed (35, CUDA driver version is insufficient for CUDA runtime version), probably no CUDA devices

My version is cuda 7.5 , the highest I get offered for Linux. Prolly that's the error ?
On which of your versions I can fall back ?

thanks !

1. It seems you see and test only Sammy007 stratum implementation. Unfortunately, there are several others. Here is how I added pools and various stratums:
a. Take stratum software (eth-proxy, qtminer, miner-proxy).
b. Learn sources, sniff traffic, see data.
c. Implement it.
Now you take another (fourth? fifth?) stratum implementation which was created after I added all -esm values, and tell me that my miner works incorrectly for all "-esm 0" pools. It is not true because "-esm 0" mode was created for Dwarfpool originally and it works fine there. But it can be true Sammy007 pools, though I remember when I saw first Sammy007-based pool I tested it and it worked well. If it does not, send me some pool address and explain the issue(s) in details, if necessary I will add "-esm 4" for it if other modes don't work. Or, if sending rig name as "worker" in login string will help and solve all issues - just let me know, I will add it in next update and we all will be happy.
Again: "-esm 0" was created to work exactly as eth-proxy software, it was not designed for Sammy007 pool, this pool was designed later.

2. I measure time spent in GPU and calculate hashrate as NumberOfHashes/GpuTime, that is all, I don't know how other miners calculate hashrate. If you think that I show more hashrate that miner actually gives, I cannot prove anything because my miner is closed source. The only way to measure real hashrate - test my miner and others and see how much they earn. It's not so difficult if you have several GPUs and not in a hurry, and I'm sure many people already did it. Everyone who sees that Genoil's miner (or ethminer) earns more can use his miner, and vice versa. Talking more about differences in hashrate in miner and pool is useless, all my thoughts I've written in FAQ.

1. Definition of eth_submitLogon :
    {"jsonrpc": "2.0", "id": 2, "method": "eth_submitLogin", "worker": "", "params": ["", "x"]}
    Code : https://github.com/sammy007/open-ethereum-pool/blob/master/proxy/stratum.go#L113

2. You are mentioned that it would be resulting due connection quallity and so on,
    but why show other miners accurat hashrates like orginal cpp-ethereum miner or genoil fork ?

1. Ok, take Dwarfpool as example of "-esm 0" pool. It uses own eth-proxy software to connect ethminer via stratum to the pool. Sources of eth-proxy are available. Show me the code line that sends worker name in "eth_submitLogin" command and I will agree that I'm wrong. Other "-esm 0" pools must work with eth-proxy properly (because this mode is called "0 - eth-proxy mode" in readme), otherwise you should used "-esm 1" or "-esm 2" mode for them.
2. Please read OP or Readme, "FAQ" section, question "Why pool shows less hashrate than miner?"

that is not true... all pools at your named esm 0 mode stratum implementation are using it also in eth_submitLogin,
but they have implement a workaround to set a fixed workerid when not given like your miner and nicehash do...
so that it isn't possible to to identify different rigs ...

by the way your hashrate displayed at miner isn't corrct shows exactly 5 % more then the real hashrate messured at stratum-pool directly connected with a lat. of 0 - 1 ms
so that this could not be a connection to pool related issue

OpenCL: clGetDeviceIDs
ADL: ADL_Adapter_NumberOfAdapters_Get


"eth1.0" is not related to -eworker parameter, it just marks miner, miner sends "eth1.0" only in "-esm 0" mode and only for "eth_submitLogin" command, in this command you can ignore "worker" at all, it is not used by pools. Though I can send -eworker value there if you want. -eworker value is used in "eth_submitWork" and "eth_submitHashrate" commands only, but most pools that want "-esm 0" mode ignore it even there and take worker name from wallet address (WALLET.WORKER).
PS. If "worker: eth1.0" string in "eth_submitLogin" looks strange for you, I can explain more First version of my miner did not have "-eworker" parameter (and "-esm" too), it used "worker" in login string just as miner signature, not as rig name. Then I found that there are several ETH-stratum versions and some of them send rig name in "worker" parameter in some commands, I added it to support these pools. But I did not touch original code for "eth_submitLogin" (because it was not necessary) so it still sends the miner marker.


I have no ideas, sorry, it can be something related to videodrivers.

@Claymore
My Rig temperature is not being reported by 7.1 on windows 10. I have 8 AMD GPU's and it use to work fine. I closed the miner and changed the DRC % from 30 to 20 due to summer hitting us and when i started it only shows GPU 0 temperature.

why did your "fine" miner submit eth1.0 as workername everytime also when -eworker parameter is set ?

@Claymore

can you explain in what order your read the cards or number the cards ?
With old crimson driver it was easy to determine a faulty card because of aticonfig --lsa  printed the gpu number and the PCI ID.. With the PCI ID it was easy to find the PCI-E slot.

Now with amdgpu-pro driver there is no aticonfig tool which prints gpu number and PCI IDs .

So now, if your miner reports an error because of OpenCL hang i only get the GPU number from your miner, but i've no idea what PCI-ID is behind that GPU number to identify faulty card physically.

Thanks

I'm not even able to take the limits maximum at WattMan, the same error shows and restarts the system whenever I apply the clock to the maximum.

Isn't anyone else having the similar problem with overclocking?

It's not strange, you are mining not ETH but ETC, so you must use "-allcoins 1" option.
It seems now ETC has a bit less number of epochs than ETH, miner knows current ETH epoch and detects that it mines not ETH and shows this message.
I will fix it in next update so you will be able to mine ETC without specifying "-allcoins 1" option.

GPU_FORCE_64BIT_PTR 0
GPU_MAX_HEAP_SIZE 100
GPU_USE_SYNC_OBJECTS 1
GPU_MAX_ALLOC_PERCENT 100
GPU_SINGLE_ALLOC_PERCENT 100

NANOPOOL IS IN GREAT PAIN--

I mine there, and the wallet they have is vulnerable, and I am eating hacker dirt as they wait for a protected wallet.       --scryptr

No worries. I think the error has to do with amdgpu driver in ubuntu 16.04
I do in fact get the same error when trying to start ethminer (classic) genoil ethminer and nerd ralphs version of ethminer in ssh session. My apologies for saying earlier that ethminer ran in ssh. It did not.

Here is what I was able to do to fix it:

I had a standard startup script that I was using to start claymore "startclaymore"
I added a reference to that script in /etc/rc.local
Now it runs on startup and doesn't give me the segmentation fault

Therefore I have killed 2 birds with one stone: I have a way to start claymore with system boot and it actually runs instead of throwing a segfault.
Now to dig deeper - rc.local runs the script as root - so I tried starting claymore with sudo and it runs. Segfault occurs only when running as standard user instead of root...

Ideas ?

THE MISSING LIBJANSSON FILES--

EthOS has a peculiar syntax, you must use "sudo apt-get-ubuntu install libjansson*".  Makes sure all is spelled correctly, the double "ss" in libjansson is real.  The EthOS developers do not wish their specialized environment to become broken, so use "apt-get-ubuntu" instead of "apt-get".

Th command worked immediately and quickly for me.       --scryptr

I MISSED YOUR POST, BUT--

The multiple exclamation points have a worthy and needed emphasis, however, in Linux, they trigger an interpetation that a BASH script is present.  Windows users hoping to start Claymore and earn money without reading the README!!!.txt are not affected, and will still ask questions already answered.  However, in a Linux console window, opening the file results in an error, and the file must be renamed.  In Linux, files are renamed with the MOVE "MV" command.  In a console wnodw (simiilar to a DOS box or command window in Windows), the exclamation points lead to an error.  The file must be moved to a different name, as in "mv README*.txt readme.txt".  I don't know exactly why, but the exclamation point is syntax in BASH (BOURNE shell, the movie has a flavor), and makes the file difficult to handle in text-only mode.  For example, trying to move README!!!.txt to readme.txt with "mv" as in "mv README!!!.txt readme.txt" results in an error, the exclamation points must be addressed anonymously with the asterisk "*".        --scryptr

Strange error in version 7.1 =)

Could be that this missing libjansson packages are the problem? I couldn´t install them at first try from ubuntu trusty universal repositories nor from .deb package. Will try from sources tomorrow.

It does work in windows 10, however, I'm using older 7xxx cards.  Alternative is MSI Afterburner and just set the clocks manually and tune while your mining and save a profile for later use.

I have posted error stating there is something wrong with readme.txt and history.txt date (was in the future) maybe that's problem