CUDA miner

1293032343569

Comments

  • Bensam123Bensam123 Member Posts: 62
    Yeah getting "Read response failed: End of file" all the machines crash at the same time after about 18 hours or so while using stratum.
  • cflushcflush Member Posts: 24
    @coldmine try putting the -U argument after the -F argument
  • ramviramvi Member Posts: 10

    I am getting 18mhs with gtx 970 only using ethminer -U , any tips on what to tweak to get it to 20mhs?

    How are you getting 18 MH/s with GTX 970? I'm getting 6 MH/s with the same card.
    I'm running on Windows 10 with ethminer genoil 1.0.4b3, and CUDA 7.5 installed.
  • Marvell9Marvell9 Member Posts: 593 ✭✭✭
    ramvi said:

    I am getting 18mhs with gtx 970 only using ethminer -U , any tips on what to tweak to get it to 20mhs?

    How are you getting 18 MH/s with GTX 970? I'm getting 6 MH/s with the same card.
    I'm running on Windows 10 with ethminer genoil 1.0.4b3, and CUDA 7.5 installed.
    @ramvi you need to run windows 7 or Liniux to get the proper hash rate i have 3 970s one hashes at 21, one at 20.9 and one at 18
  • xenomorph113xenomorph113 Member Posts: 7
    ramvi said:

    I am getting 18mhs with gtx 970 only using ethminer -U , any tips on what to tweak to get it to 20mhs?

    How are you getting 18 MH/s with GTX 970? I'm getting 6 MH/s with the same card.
    I'm running on Windows 10 with ethminer genoil 1.0.4b3, and CUDA 7.5 installed.
    if youre using windows 10 you need to use the 347.52 drivers, ANYTHING newer will kill your hashrate

    windows 10 will always try to install 353.xx or newer, which will of course kill your hashrate
  • diluxdilux Member Posts: 22
    edited March 2016
    if i use the -U option i get:
    invalid argument -U

    or am i missing something here?

    ubuntu 14.04

    running etherminer without the -U flag works fine and also ethpools qtminer works fine and i get about 17MH with my gf970x (not oc'd)

    thing is that i switched to suprnova pool and running their command without the -U flag seems to result in that no blocks get submitted to my account there.

    i have another rig woth amd cards and that works without a hitch

    CUDA is installed and when running deviceQuery the result = pass.
  • Marvell9Marvell9 Member Posts: 593 ✭✭✭

    ramvi said:

    I am getting 18mhs with gtx 970 only using ethminer -U , any tips on what to tweak to get it to 20mhs?

    How are you getting 18 MH/s with GTX 970? I'm getting 6 MH/s with the same card.
    I'm running on Windows 10 with ethminer genoil 1.0.4b3, and CUDA 7.5 installed.
    if youre using windows 10 you need to use the 347.52 drivers, ANYTHING newer will kill your hashrate

    windows 10 will always try to install 353.xx or newer, which will of course kill your hashrate
    @xenomorph113 so on windows 10 how is the cuda CPU utilization? I have a cheap $60 pentian 3GHZ dual core on my windows box with 4 970s and the cpu utilization is so high by ehtminer the machine is constantly stuttering , it works but I have a feeling I'm losing hashes due to the performance.

    Do I need a stronger GPU ? I am using the most current version of the ethminer fork.

  • cflushcflush Member Posts: 24
    Ethminer performance for the 750Ti on Windows is poor compared to Linux. Everyone agrees.

    That said, can someone who is using 750Ti(s) on Windows tell me the best setup they have found?
    Windows 7 vs WIndows 10?
    Older cudaminer vs new release?
    optimal command line entries?

    I really appreciate your time and feedback. Thx.

    -Cflush


  • Marvell9Marvell9 Member Posts: 593 ✭✭✭
  • promojopromojo Member Posts: 2
    edited March 2016
    Folks,who can help me setup my rig with 5x Asus 970 GTX win 10 x64(drivers 362.00-desktop-win10-64bit-international-whql)
    Ive tried to run all cards with following command in ethminer-0.9.41-genoil-1.0.4b3
    ethminer --cuda-devices 0 1 2 3 4 -U -F http://eth.suprnova.cc......
    So it creates dags,start digging and after 10 secs - black screen,internet connection error and ive got this -

  • wirelessnet2wirelessnet2 Member Posts: 385 ✭✭✭
    ok so @Genoil I tried mining ETH on Windows 10 with a GTX 980 on the newest Nvidia driver. I used the -U flag and no other flags. I was mining at a speed of about 8 MH/s. I then installed the 347.52 Driver, and then tried using the miner again. When i use CUDA (-U flag), the miner gives me "CUDA error in func 'ethash_cuda_miner::getNumDevices' at line 112 : CUDA driver version is insufficient for CUDA runtime version." I then switched to using OpenCL to mine, and was mining at a speed of 21 MH/s. I am using the 1.0.4b3 version of your miner. Three questions: 1. Is 21 MH/s good for a single GTX 980? and 2. How can I get CUDA working? and 3. Would there be a significant benefit to using CUDA vs OpenCL for ETH mining? i know other for other algos using a CUDA miner greatly increases speed, but so far it looks like for ETH mining OpenCL is barely behind CUDA
  • furionlovefurionlove Member Posts: 10
    Thank you @Genoil for all the awesome work on the miner.

    I do have 2 computers, one with Nvidia 760 and one with ATI 280x, however both are production machines running Virtual Machines and I use them for work. So I can't really mine at 100%.

    Is there any reliable way, to tell the miner to use only lets say 50% of cuda/ati opencl cores and ram?
  • Marvell9Marvell9 Member Posts: 593 ✭✭✭
    @Genoil

    I played around with the settings and found

    ethminer --farm-recheck 100 -G -F http://192.168.0.46:8546/miner/30/Marvell9.10 -U --cuda-devices 0 1 2 3 --cuda-extragpu-mem 0 --cuda-block-size 128 --cuda-grid-size 8192 --cuda-schedule auto




    if I remove "--cuda-schedule auto" cpu usage goes down to the normal cpu usage of 2-5% I see on my AMD boxes.

    Still the ethminer instance still seems laggy and on the pool I'm only getting around 70mhs I suspect due to the herky jerkiness of ethminer on that windows 7 box.

    When I bench mark each card I see 21mhs on 2 and 18.5 on the other two so I should be closer to 78mhs. My AMD boxes actually always show higher on the pool than what ethminer reports not this NVidia box.

    I guess on my next build I'll try and UBNNTU ethminer install , not sure what that all entails though never installed CUDA drivers manually on UBUNTU.

    I tried with and without the -G seems to make no difference btw with the newest version of your build.

  • kenshirothefistkenshirothefist Member Posts: 73
    @Genoil, is it possible for you to add per-device mining in OpenCL mode, similar as it is done for CUDA mode? In OpenCL mode, one is unable to select only a particular device, whereas this is possible in CUDA mode ... Thanks!
  • c3p8c3p8 Member Posts: 28
    where can i download the cuda miner?
  • Bensam123Bensam123 Member Posts: 62
    Anyone else having CPU utilization issues after todays dag file generation? Seems some of my miners are using 25% on ethminer at all times, not just when it's parsing.
  • mrpmrp Member Posts: 7
    edited March 2016
    coinotron stand alone stratum lose connection after internet disconnection or when ip changing, recconnect possible only manually

    but dwarf stratum work perfectly in same state

    may be implement dwarf stratum instead coinotron ?
  • teknohogteknohog Member Posts: 1
    skunk said:

    sot173 said:


    nvcc fatal : redefinition of argument 'std'

    got the very same issue (but under linux) after upgrading to cmake-3.3.1, downgrading to cmake-3.2.2 fixed it...
    can't tell if it's a cmake or a makefile bug though...
    I use cmake version 3.4.3, and I fixed this by removing the redundant 'std'. With make VERBOSE=1, it showed that the nvcc commandline had both "--std c++11" and "--std=c++11" so I removed the latter:

    sed -i 's/--std=c++11;//' libethash-cuda/CMakeLists.txt

    My 750 Ti now averages about 4.4 Mhash/s which doesn't feel too great, but I guess it's OK for this GPU, not quite the efficiency of Radeons, though.



  • wirelessnet2wirelessnet2 Member Posts: 385 ✭✭✭
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    Unfortunately I'm back. With ETH close to $10 currently, it seems logical to invest some time into proper Compute 3.0 support (GTX 6xx, 760, 770, K520) above all else. In other words: AWS g2 spot instances can be used to mine with a (tiny) profit in certain regions.

    Stratum support needs to be fixed as well, doesn't seem to be too reliable currently.

    @regulars thanks for answering all the n00b questions during my absence.
  • HailSaganHailSagan Member Posts: 2
    Mo35 said:

    @Genoil Yeah , figured myself :) , Cuda 6.5.14 wich is available dosnt support compute 5.2 , there is patch wich upgrades toolkit to 6.5.19 with compute 5.2 support.
    https://developer.nvidia.com/cuda-downloads-geforce-gtx9xx

    succesfully compiled 1.0.3 master branch and 1.0.4b3 for cuda 6.5 with compute 5.2 card support
    feel free to use if you need , but preferably let Genoil check them and aprove release beforehand
    http://wikisend.com/download/361278/ethminer1.0.3_master_6.5cuda.zip
    http://wikisend.com/download/960168/ethminer1.0.4b3_stratum_6.5cuda.zip

    "how to" will eventualy follow when i write up my adventures as newbie in compiling stuff

    Hey, did you still have those binaries? The links got expired :/ maybe you could upload those on a "never expire" server like OneDrive or Google Drive? Thanks!
  • Marvell9Marvell9 Member Posts: 593 ✭✭✭
    I might have found an easier solution for windows 10 rather than hacking the registery, currently i have 6 cards on a board hashing at 115mhs windows 10, two of the cards are used lemons getitng 16.5 mhs even on win 7 but the rest of the cards get a solid 20.9+ mhs with a mild OC.

    The asus Strx card i have seen as high as 23mhs with a 30% OC but i dont know if i want to stress the card that much.

    Ill post the steps i took after 24 hours testing on the pool.
  • wompwomp Member Posts: 5
    @Genoil - would love to see GTX 770 support!

    @Marvell9 - What is the registry hacking method? I've been wrestling with Windows 10 and haven't seen anything about registry tweaks to get mining working properly...
  • Marvell9Marvell9 Member Posts: 593 ✭✭✭
    @womp its a few pages back on this thread somewhere is super invloved though I would rather do an Ubuntu inststall rather than go that route
  • wompwomp Member Posts: 5
    @Marvell9 ugh yeah that doesn't look good. Looking forward to seeing the steps that you took.
  • vexdvexd Member Posts: 2
    promojo said:

    Folks,who can help me setup my rig with 5x Asus 970 GTX win 10 x64(drivers 362.00-desktop-win10-64bit-international-whql)
    Ive tried to run all cards with following command in ethminer-0.9.41-genoil-1.0.4b3
    ethminer --cuda-devices 0 1 2 3 4 -U -F http://eth.suprnova.cc......
    So it creates dags,start digging and after 10 secs - black screen,internet connection error and ive got this -


    Same error here, on windows 8.1, with genoil ethminer 1.0.3.
  • Marvell9Marvell9 Member Posts: 593 ✭✭✭
    I'll post a full wint10 guide when i get an extra mother board either tomorrow or Thursday to install these two 970s i have sitting around , for one you have to install the 350.12 drivers it seems and to do that you have to disable some win10 auto updating stuff ill post the guide with screen shots.
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    @vexd @promojo the black screen and timeouts are caused by the kernel running longer than the timeout defined by TDR: https://msdn.microsoft.com/en-us/library/windows/hardware/ff569918(v=vs.85).aspx

    This is not supposed to happen, but in case of Windows 10, I can imagine it happens. 20Mh/s for 5 x 970 is way too low of course. If you want to continue with this setup anyway, I suggest you lower --cuda-grid-size (default 8192) by factor 2 until it runs stable.
Sign In or Register to comment.