just a quick heads up, press s and verify your incorrect shares. I reverted to 5.3 because all above 6 gave me about 1% incorrect shares. after much testing I can confirm the incorrect share problem is real with the newer versions, all else being equal. why? no idea, maybe claymore could look into it. running 4 rigs, 380s, 390s, one 280x, 470s, win7 and one rig win10
Just checked it, currently v6.4 got 6 incorrect shares and 6140 correct ones. So it's about 0.1% of incorrect shares. It is because of fast math I use in kernels, if you don't like it you can use "-etha 2" but it's about 1% slower. DAG size grows and it causes more incorrect shares, as soon as they reach about 0.5% I will use "-etha 2" by default. However, v5.3 and v6.4 must show same rate of incorrect shares on same DAG (epoch). Okay your ignorance let me take the descission to ban claymore from my pool-sources, at time porting open-ethereum-pool to open-decred-pool ... but you are out
Oh, you can ban me from anything you want of course. Though I'd like to know what you want anyway, it seems I overlooked something interesting, this thread is full of newbies questions and grows all the time by several pages every day. If you are a dev, you can PM me and I will try to answer your questions. EDIT: found your question finally, my miner supports four different ETH stratum versions, check "-esm" option for that in OP: -esm Ethereum Stratum mode. 0 - eth-proxy mode (for example, dwarpool.com), 1 - qtminer mode (for example, ethpool.org), 2 - miner-proxy mode (for example, coinotron.com), 3 - nicehash mode. 0 is default. So if miner does not work with your pool, probably you use incorrect "-esm" value. PS. While searching within your latest posts to find a message in my thread, found your nice phrase "you could dual mine without closedsource claymore shit". Thanks you are so polite
|
|
|
Hello,
i Use
Ubuntu 1604 Claymore's Dual Ethereum AMD+NVIDIA GPU Miner v6.4 SAPPHIRE NITRO RADEON RX 470 8G
and i have test the option
-cvddc 1050 -mvddc 1050
but if i reduce the mV - nothing change on the Power Consumtion (I measure the power consumption with a calibrated electricity Watt meter)
what do i worng?
lg Martin
From OP or Readme: -cvddc set target GPU core voltage, multiplied by 1000. For example, "-cvddc 1050" means 1.05V. You can also specify values for every card, for example "-cvddc 900,950,1000,970". Supports latest AMD 4xx cards only in Windows. Note: for NVIDIA cards this option is not supported. -mvddc set target GPU memory voltage, multiplied by 1000. For example, "-mvddc 1050" means 1.05V. You can also specify values for every card, for example "-mvddc 900,950,1000,970". Supports latest AMD 4xx cards only in Windows. Note: for NVIDIA cards this option is not supported.
|
|
|
v9.4:
- Added Fury and RX480 cards native support.
|
|
|
Hi Claymore, Off topic this..... are you still updating your XMR miner? Is this a viable 3rd coin, after DCR and SIA for dual mining?
I don't see ways to make my XMR miner faster, but right now I test it with latest cards and in a few hours I will release new version that will detect latest cards and use best settings for them. xmr or lbry i think xmr is very cool algorytm xmr has bye orders more than eth XMR cannot be added to ETH dual miner as the second coin, though in theory it could be added instead of ETH, i.e. as first coin. I will check LBRY and some other coins very soon.
|
|
|
I tried v. 6.4 on Ubuntu 16.04 64 bit, driver is amdgpu-pro_16.30.03-315407, kernel is 4.4.0-34-generic the output is ./ethdcrminer64 -epool coinotron.com:3344 -ewal username.3 -esm 2 -epsw 3 -allpools 1 -mode 1 -etha 2 ͻ � Claymore's Dual ETH + DCR/SC GPU Miner v6.4 Beta � ͼ ETH: 1 pool is specified Main Ethereum pool is coinotron.com:3344 segmentation fault
Isn't this the target platform? Can I help debugging? (genoil 110 does mine. radeon 480) Felix Just tried same configuration - it works fine for me. PM me the full log.
|
|
|
On my GTX 1070 the 6.4 build is actually slightly slower with CUDA 8 which is strange because Genoil's miner with CUDA 8 is notably faster.
version 6.2 (CUDA 7.5) 26.6 ETH 710 SIA
version 6.4 (CUDA 8 ) 26.1 ETH 685 SIA
This is Win10 anniv with the latest nvidia drivers
You can use v6.4 CUDA 7.5 if you don't like speed of CUDA 8.0. Speed depends on windows&drivers&gpu so I include cuda 6.5, 7.5 and 8.0 builds, you can select the best for you.
|
|
|
Unfortunately, it's not so easy. Linux version of this miner had own sources (I know it's bad, the main reason was that other dev worked on Linux version), they were on dedicated Linux-based system, then HDD crashed and I lost everything on that system, last backup that I have is for v7.0. Of course I can take latest Windows version and update its sources to be able to compile them on Linux too, but it will take a few days of work, I just don't have so much time right now.
what about update for latest VC? thanks VC = video cards? Since monero becomes attractive again, in a day or so I will test miner with latest cards to see if I can improve something.
|
|
|
Claymore, could you please update the Linux version? I'm using version 9.3 in windows perfectly, but 9.1 on linux does not connect to the same pool that i'm using on windows (tested with moneropool and monero.crypto-pool.fr).. sorry for my crappy english Unfortunately, it's not so easy. Linux version of this miner had own sources (I know it's bad, the main reason was that other dev worked on Linux version), they were on dedicated Linux-based system, then HDD crashed and I lost everything on that system, last backup that I have is for v7.0. Of course I can take latest Windows version and update its sources to be able to compile them on Linux too, but it will take a few days of work, I just don't have so much time right now.
|
|
|
v6.4:
- added "-etha 2" option value for gpu-pro Linux drivers. - fixed issue with a bit lower mining speed of previous version (AMD cards). - Improved speed by about 3% for NVIDIA 10xx cards in Windows 10 (CUDA 8.0 build). - NVIDIA: added build with CUDA 8.0, it is default now, there are also CUDA 6.5 and 7.5 builds in separate folders. - NVIDIA: added temperatures/fans monitoring. Note that fans management is not available, only monitoring is possible. - EthMan now shows "off" or "stopped" status for disabled/stopped cards in detailed hashrate stats.
|
|
|
@claymore
this is something i havent quite figured out.
have an ubuntu 15.10 x64 rig, 15.12 driver (15.302 it says) 8 gig ram cdm 6.3. has 390 8 gig and 380 2 gig. 390 was primary display, 380 2nd with no monitor. all was well, no environmental settings needed. ran for months stable as a rock.
so my seasonic x750 burned a pcie cable and i will rma it, have a evga gs850 on its way. till it gets here i have pulled the 390, so now the 380 is the only card and primary display. and it will not run, cant create the dag. cdm logs show 1.8 gigs vram.
here is what ive tried: set all those GPU_etc environmental variable. nope. stopped xorg (now cdm logs show 2 gig vram). nope same error, cant create dag. throw in an old HD6770 as primary card running the display so the 380 is now 2nd card with no monitor hooked to it. no go, same error. also tried cdm 5.3 (?, the last 5.x one). negative.
guess the question is, does the dag get created on 1st supported card and then transferred to the others? does the card creating the dag need more vram than other cards? this 380 ran fine alongside the 390, but wont run by itself, even without xorg sucking down 256 mb or whatever the gui uses.
Miner creates DAG on every card. No transfers. If you get error it means that OpenCL cannot allocate enough buffer. Why? I don't know. Perhaps OS allocated some memory for itself. Or OpenCL drivers don't like your configuration for some reason and therefore don't allow to allocate max memory. Unfortunately I don't know drivers logic about max memory limits calculation, I only know that is not simple and that's the reason of many problems.
|
|
|
@Claymoer trying to down loads the drivers for 10-serires from the op tells me the folder doesn't exist can you pls fix the link?
OP updated with new info: NVIDIA DRIVERS: 9xx cards in Windows 7 x64: just use latest/recent drivers from Nvidia website (for example, 368.81). Note that latest 372.54 is slower than 368.81. 9xx cards in Windows 10 x64: you have to use old drivers (for example, 352.xx) and miner built for cuda6.5. 10xx cards in Windows 7 x64: just use latest 372.54 drivers from Nvidia website. 10xx cards in Windows 10 x64: just use latest 372.54 drivers from Nvidia website, note that you must have Win10 Anniversary update.
|
|
|
v6.3 gives bit less hashrate on RX470 then 6.2beta.
You are correct. Within 24 hours I will release an update that has same speed as 6.2b (actually it's bit faster because I made some minor improvements), also it has some speed improvements for nvidia 10xx cards in Windows 10 and temps/fans monitoring for nvidia cards.
|
|
|
Claymore, could you do some ethman modification: let ethman show "off" instead gpu temperature of those cards, wich was switched off by max temp reached or another reason
Ok, will be added in next update.
|
|
|
Claymore, It seems to sometimes miner doesn't restart itself, but only when I press a key (I press 's' from time to time to see if everything is ok) Here's the log: BEware over 8mb lenght.. https://www.pastefs.com/pid/4083 showed only when I press the key... If you use Win10, it can be if you clicked inside of terminal window so it has text focus.
|
|
|
Someone can answer why on my PC(1070 372.54,win 10 AU) does't work option -tt 34 -tstop 45\-45(GPU just does't turn off and miner does't close)!
OP or Readme, "KNOWN ISSUES" section: ... - Temperature/fan management for nVidia cards is not supported currently. ...
|
|
|
I had to change OpenCL kernels because of problems with gpu-pro Linux driver. Since it seems it works a bit slower, in next version I will return old kernels for Windows and will use new kernels for Linux only. Right now for Windows you can use v6.2 Beta.
The new version 6.3 runs on Linux without errors, but 7% less hashrate. What Linux? What drivers? What cards? If you want to report some issue, please share a bit more info.
|
|
|
I had to change OpenCL kernels because of problems with gpu-pro Linux driver. Since it seems it works a bit slower, in next version I will return old kernels for Windows and will use new kernels for Linux only. Right now for Windows you can use v6.2 Beta.
|
|
|
I've tried the new 6.3 version with linux ubuntu 14.04 ./ethdcrminer64: /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version `GLIBCXX_3.4.21' not found (required by ./ethdcrminer64) going back to 6.2 for now.
Ok, rebuilt in 14.04, please redownload it.
|
|
|
|