GPU Mining Software - NVIDIA/AMD?

123578

Comments

  • lumierrelumierre Member Posts: 12
    edited May 2015
    Genoil said:

    @eth001 what do you need? I can create some binaries tomorrow.

    Could you share your eth build that your using to mine right now?

    I have managed to compile eth x64 without any errors but it just crashes when trying to mine. I have followed your mini guide on the 1st page even setting opencl-platform and opencl-device but it still doesn't mine with the current develop.

    My eth build only manages to go this far then crashes (same goes GPU mining with your April 22 build on 2nd page):
    Benchmarking on platform: { "platform": "AMD Accelerated Parallel Processing", "
    device": "Tahiti", "version": "OpenCL 1.2 AMD-APP (1445.5)" }
    Preparing DAG...
    i 12:49:48|main Loading from libethash...
    i 12:49:51|main Done loading.
    D 12:49:51|main pause took 0
    i 12:49:51|gpuminer0 workLoop 0 #00000000… #00000000…
    D 12:49:51|main kickOff took 0.001
    D 12:49:51|main pause took 0
    i 12:49:51|gpuminer1 workLoop 0 #00000000… #00000000…
    D 12:49:51|main kickOff took 0.002
    Warming up...


    Same with CPU mining:
    Benchmarking on platform: 8-thread CPU
    Preparing DAG...
    i 12:52:26|main Loading from libethash...
    i 12:52:26|main Done loading.
    D 12:52:26|main pause took 0
    D 12:52:26|main kickOff took 0.001
    D 12:52:26|main pause took 0
    D 12:52:26|main kickOff took 0.002
    D 12:52:26|main pause took 0
    D 12:52:26|main kickOff took 0.001
    D 12:52:26|main pause took 0
    D 12:52:26|main kickOff took 0.001
    D 12:52:26|main pause took 0
    D 12:52:26|main kickOff took 0
    D 12:52:26|main pause took 0
    D 12:52:26|main kickOff took 0
    D 12:52:26|main pause took 0
    D 12:52:26|main kickOff took 0
    D 12:52:26|main pause took 0
    D 12:52:26|main kickOff took 0.001
    Warming up...


    However, the eth exe you posted on April 20 (1st page) surprisingly works. Here are my results with a single r9 280x with i7 4770k:
    c:\genoil>eth -G -M --opencl-platform 1
    --- [ 12:29:35 | main ] 00004000000000000000000000000000000000000000000000000000
    00000000
    Benchmarking on platform: { "platform": "AMD Accelerated Parallel Processing", "
    device": "Tahiti", "version": "OpenCL 1.2 AMD-APP (1445.5)" }
    Preparing DAG...
    --- [ 12:29:52 | main ] pause took 0
    *** [ 12:29:52 | main ] Spawning gpuminer
    *** [ 12:29:52 | gpuminer ] Entering work loop...
    --- [ 12:29:52 | main ] kickOff took 0.007
    Using platform: AMD Accelerated Parallel Processing
    WUSING 0: Tahiti (OpenCL 1.2 AMD-APP (1445.5))
    a 1: Intel(R) Core(TM) i7-4770K CPU @ 3.50GHz (OpenCL 1.2 AMD-APP (1445.5)
    )
    rUsing device: Tahiti (OpenCL 1.2 AMD-APP (1445.5))
    ming up...
    Trial 1... 22092748
    Trial 2... 21998097
    Trial 3... 22085391
    Trial 4... 21998097
    Trial 5... 22085391
    *** [ 12:30:13 | gpuminer ] Finishing up worker thread...
    *** [ 12:30:13 | main ] Stopping gpuminer
    *** [ 12:30:13 | main ] Terminating gpuminer
    min/mean/max: 21998097/22051944/22092748 H/s
    inner mean: 22027195 H/s
    Phoning home to find world ranking...
    Error phoning home. ET is sad.
    c:\genoil>


    Your April 22 build (2nd page) works CPU mining but NOT GPU:
    Warming up...
    Trial 1... 137620
    Trial 2... 138538
    Trial 3... 135565
    Trial 4... 137566
    Trial 5... 137222
    i 13:05:58|main Terminating miner0
    i 13:05:58|miner0 Finishing up worker thread...
    i 13:05:58|main Terminating miner1
    i 13:05:58|miner1 Finishing up worker thread...
    i 13:05:58|main Terminating miner2
    i 13:05:58|miner2 Finishing up worker thread...
    i 13:05:58|main Terminating miner3
    i 13:05:58|miner3 Finishing up worker thread...
    i 13:05:58|main Terminating miner4
    i 13:05:58|miner4 Finishing up worker thread...
    i 13:05:58|main Terminating miner5
    i 13:05:58|miner5 Finishing up worker thread...
    i 13:05:58|main Terminating miner6
    i 13:05:58|miner6 Finishing up worker thread...
    i 13:05:58|main Terminating miner7
    i 13:05:58|miner7 Finishing up worker thread...
    min/mean/max: 135565/137302/138538 H/s
    inner mean: 137223 H/s


    After 2 days uninstalling and reinstalling stuff, building eth multiple times, still no luck trying to mine. Benchmarking alone with the latest develop doesn't work for me (CPU and GPU). I don't know if my builds are wrong or if the multi-GPU support made the new eth incompatible with my computer or both. I'm only using 1 gpu by the way. A working eth build would be greatly appreciated right now. I'd be happy to report my findings.
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    edited May 2015
    @lumierre I haven't updated my source tree for a while. They have a completely revamped libethash in it now, so I'm curious what causes the issues. Unfortunately I haven't got working eth/ethminer where I am currently and I'm not back at the working version untile monday. If I have time (don't think so) I may look into building it myself here and trying again. Maybe ethminer gives you better results?

    --edit--

    quickly updated to latest develop and built ethminer. CPU mining gives me exact same result as you. Haven't got a GPU in this machine ATM, maybe later.
    Post edited by Genoil on
  • mikmik Member Posts: 57 ✭✭
    Ok guys you all did a lot of tests, like me. But did any of you ever really mine with eth.exe to an address? I that case please tell me:
    1. How do I import/generate an address in eth.exe?
    2. What is the command line to start mining to this address? I mean, not only a testrun with '-G -M'!
    Tx a lot guys!!!
  • literarydeviceliterarydevice TorontoMember Posts: 13
    Hey Mik,
    mik said:

    1. How do I import/generate an address in eth.exe?
    2. What is the command line to start mining to this address? I mean, not only a testrun with '-G -M'!
    1. Generate an address using geth account new
    2. The following command should work to mine to a specific address: eth.exe -G -m on -a "ether address" (without quotes, and note the lower-case m)
  • Wil611Wil611 Member Posts: 77 ✭✭
    If spent the last week trying to get GPU mining to work on Ubuntu 14.04 with an AMD Radeon R9 200 series compiling the cpp-ethereum source via the wiki with little success. The closest I've come is generation of the DAG after which it crashes. If anyone can give me some pointers it would be greatly appreciated.

    DAG 11:05:22|eth Generating DAG file. Progress: 88 %
    DAG 11:05:26|eth Generating DAG file. Progress: 89 %
    DAG 11:05:31|eth Generating DAG file. Progress: 90 %
    DAG 11:05:35|eth Generating DAG file. Progress: 91 %
    DAG 11:05:40|eth Generating DAG file. Progress: 92 %
    DAG 11:05:44|eth Generating DAG file. Progress: 93 %
    DAG 11:05:48|eth Generating DAG file. Progress: 94 %
    DAG 11:05:53|eth Generating DAG file. Progress: 95 %
    DAG 11:05:57|eth Generating DAG file. Progress: 96 %
    DAG 11:06:02|eth Generating DAG file. Progress: 97 %
    DAG 11:06:06|eth Generating DAG file. Progress: 98 %
    DAG 11:06:11|eth Generating DAG file. Progress: 99 %
    DAG 11:06:15|eth Generating DAG file. Progress: 100 %
    ℹ 11:06:15|eth Done loading.
    ℹ 11:06:15|gpuminer0 workLoop 0 #00000000… #00000000…
    Using platform: AMD Accelerated Parallel Processing
    Using device: Tahiti(OpenCL 1.2 AMD-APP (1642.5))
    ℹ 11:06:15|gpuminer1 workLoop 0 #00000000… #00000000…
    Using platform: AMD Accelerated Parallel Processing
    Using device: Intel(R) Core(TM) i3-4130 CPU @ 3.40GHz(OpenCL 1.2 AMD-APP (1642.5))
    Warming up...
    Segmentation fault (core dumped)
  • mikmik Member Posts: 57 ✭✭
    @literarydevice
    Hi - thanks a lot!!! That is what I was looking for :-)
    Where did you get that info from? Is there a link?
    I couldn't find any doc which switches are available. --help or -h is not implemented.
  • mikmik Member Posts: 57 ✭✭

    Hey Mik,

    mik said:

    1. How do I import/generate an address in eth.exe?
    2. What is the command line to start mining to this address? I mean, not only a testrun with '-G -M'!
    1. Generate an address using geth account new
    2. The following command should work to mine to a specific address: eth.exe -G -m on -a "ether address" (without quotes, and note the lower-case m)
    Just tried it but unfortunately eth.exe exits with Debug Error:
    R6010
    -abort() has benn called
    Will try to look into it later :(

  • timothyjcoultertimothyjcoulter Member Posts: 14
    I seem to be getting crashes with the latest code on the `develop` branch, even just during profile. Seems to be DAG related. If anyone else is running into a similar issue, would love to hear about it.
  • zeelzebubzeelzebub Member Posts: 13
    @Wil611 - Do you have the AMD driver installed? I've been experimenting with different versions and that seems to be the sticking point for me. My rig running the 14.20 proprietary AMD driver seems to run the best - with other drivers I get crashes similar to what you're describing.
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    @timothyjcoulter yes I have the same problem. It's definetely DAG related. It's not being generated at all while it thinks it has one, so it crashes. Both CPU and GPU. I'm trying to see if I can fix it, but admittedly what goes on in that code is a bit beyond my comprehension :)
  • Wil611Wil611 Member Posts: 77 ✭✭
    edited May 2015
    @zeelzebub - I finally got it working. what was happening is that the client was trying to run the opencl code againts the CPU i had to use this to launch the client

    eth -i -G --opencl-device 0

    then started mining with the minestart command.
  • timothyjcoultertimothyjcoulter Member Posts: 14
    @zeelzebub Interesting. I'm running AMD drivers 14.12. Will try upgrading now.
  • timothyjcoultertimothyjcoulter Member Posts: 14
    Update: Tried the 15.4 beta drivers and still no dice. They include better performance for GTA, so there's that. ;)
  • n12on12o Member Posts: 6
    So, the benchmark works for me. But then I get this:

    ⚡ 19:19:41|gpuminer0 Worker stopping 0.416186 s
    ℹ 19:19:41|gpuminer0 workLoop 1 #290decd9… #290decd9…
    ✘ 19:19:41|gpuminer0 Error GPU mining. GPU memory fragmentation?

    I've seen this already in this thread. Has anybody fixed it?
  • arkanaprotegoarkanaprotego Member Posts: 3
    @n12o For me, it just went away after restarting eth a few times, but I haven't mined a thing, so I doubt it is actually working...
  • knobsonkookknobsonkook Member Posts: 14
    Hello, gues!
    Can anyone explain me how to start mining?
    I have Eth V. 0.9.21 on Ubuntu 14.04 x64 installed by this instruction https://github.com/ethereum/cpp-ethereum/wiki/Installing-clients
    But when run eth -M -G I receive
    terminate called after throwing an instance of 'cl::Error'
    what(): clGetPlatformIDs
    Aborted (core dumped)


    The same result when run eth -M -G --opencl-device 0

    Who can explane what is wrong?

    ??
  • mikmik Member Posts: 57 ✭✭
    mik said:

    @literarydevice
    Hi - thanks a lot!!! That is what I was looking for :-)
    Where did you get that info from? Is there a link?
    I couldn't find any doc which switches are available. --help or -h is not implemented.

    Just want to add, that I finally found a doc on the 'eth' options here :)
    It also looks like I used an outdated version, since -h seems to be impleted now (though untested by me ;))

  • luc4luc4 Member Posts: 7

    Hello, gues!
    Can anyone explain me how to start mining?
    I have Eth V. 0.9.21 on Ubuntu 14.04 x64 installed by this instruction https://github.com/ethereum/cpp-ethereum/wiki/Installing-clients
    But when run eth -M -G I receive
    terminate called after throwing an instance of 'cl::Error'
    what(): clGetPlatformIDs
    Aborted (core dumped)


    The same result when run eth -M -G --opencl-device 0

    Who can explane what is wrong?

    ??

    same here. also tried with different drivers. I'm on Xubuntu 14.04.
  • julian1julian1 Member Posts: 1
    edited May 2015
    It's definetely DAG related. It's not being generated at all while it thinks it has one, so it crashes. Both CPU and GPU.
    I can't get the ethereum-cpp client to make outbound connections to download the chain. branch develop compiled yesterday and today.

    Edit. Hmmm pulled, and compiled and now working.
    Post edited by julian1 on
  • luc4luc4 Member Posts: 7
    zeelzebub said:

    @Wil611 - Do you have the AMD driver installed? I've been experimenting with different versions and that seems to be the sticking point for me. My rig running the 14.20 proprietary AMD driver seems to run the best - with other drivers I get crashes similar to what you're describing.

    Now i tried with the 14.20 beta driver. now DAG doesn't even start anymore... it does nothing
  • literarydeviceliterarydevice TorontoMember Posts: 13
    @mik I got that info from eth --help, after encountering the same problem you faced. One of the original posts in this thread said to use eth -G -M, but that was erroneous.
  • creoncreon Member Posts: 2
    edited May 2015
    @luc4 @knobsonkook check your opengl / opencl installation. If a basic method like clGetPlatformIDs fails then there seems to be something wrong with those libraries.

    @n12o Wait until the DAG is fully created. Just ignore the GPU errors during that time. When the DAG was fully created, restart the miner. This at least helps for me when I see this error, which is the case whenever I deleted the blockchain.

    First I'd like to thank the developers for this very nice piece of software, both on execution and on code level. I am running two NVIDIA GTX 970 and two NVIDIA Ti 750 and consistently get around 45 Mh/sec. For a non optimized open CL kernel this is a great performance on NVIDIA GPUs.

    Now I just punched around in the open cl kernel and recognized that I can get around 5 Mh/sec more on each of the GTX 970 if I reduce the number of threads per hash to 4. However, I somehow don't find any blocks anymore if I do this, so I seem to break something in the inner logic of the kernel. Any hint?

    Playing around with the workgroup and search batch size so far did not lead to any improvements. Is here anyone else who tries to adapt the source a bit to NVIDIA?
    Post edited by creon on
  • n12on12o Member Posts: 6
    @creon What ended up working for me was to use eth and not ethminer.
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    @creon, i've been poking around in the kernel too, I think the number of threads per hash is crucial to its working. You might instead want to look at the AMD optimizations already in there and see how much sense they make to the NVidia compiler.

    I did a half-hearted attempt at a CUDA port a few weeks ago but gave up on it. Don't have the time to acquire the required skills to actually make it work.
  • hope2907hope2907 Member Posts: 5
    edited May 2015
    I 'm using 7970 and only run at 8Mh/s compare to other which is running at 22mhs
    I'm using 13.1 but it make nosense a gap is too big
  • urubuurubu Member Posts: 19
    Thanks to everyone I just got my 3x 290x rig mining. I have been trying to figure it out for a few days.
    I am running geth console.
    Started the rpc,
    start ethminer -D 33000 to load the DAG.
    then started an ethminer -G -opencl-device 1 instance for each gpu, changing the number for the opencl device. Make sure it doesn't start one for the cpu.
    For some reason, you have to keep restarting each ethminer until it starts hashing and doesn't give the memory error.
  • lumierrelumierre Member Posts: 12
    edited May 2015
    So I gave up trying to mine in windows. I switched to ubuntu 14.04 and I'm finally mining. Most problems I encountered in ubuntu was solved by correctly installing up to date drivers and opencl. Ensure they are installed correctly by typing fxglrinfo and clinfo.
  • luc4luc4 Member Posts: 7
    urubu said:

    Thanks to everyone I just got my 3x 290x rig mining. I have been trying to figure it out for a few days.
    I am running geth console.
    Started the rpc,
    start ethminer -D 33000 to load the DAG.
    then started an ethminer -G -opencl-device 1 instance for each gpu, changing the number for the opencl device. Make sure it doesn't start one for the cpu.
    For some reason, you have to keep restarting each ethminer until it starts hashing and doesn't give the memory error.

    I'm trying to get my rig working for about 3 days now (~10hours sleep total LOL) without success. i also followed your guide...
    ethminer -C -M works...
    ethminer -G -M doesn't .:

    - with the preinstalled x.org x server (opensource) it gives me the 'cl::Error'
    - all the other fglrx drivers from 13.XX to 14.20 all do nothing eth or ethminer justs sits there using 100% CPU(one core)

    it's really frustrating.. any help is appreciated
  • urubuurubu Member Posts: 19
    (going by memory)
    with amd gpu's
    with ubuntu 64
    Install the fglrx driver
    install go-ethereum
    install cpp-ethereum

    (in terminal)
    geth console
    admin.newAccount()
    (remember your password)
    eth.coinbase
    ( is your account number,remember it)

    admin.startRPC("127.0.0.1", 8545, "*")

    (start other terminal)
    cd cpp-ethereum/build/ethminer
    ( or your ethminer folder)
    ethminer -D 33000 to load the DAG
    ethminer -G --opencl-device 0
    (do this for each gpu in a seperate terminal
    if it says gpu memory fault, ctrl-c and restart it
    if it says core segmentation, make sure you are not starting it on the cpu)
  • Wil611Wil611 Member Posts: 77 ✭✭
    @urubu when I run geth and then attempted to run eth it tells me that the port is already in use how are you setting the ports for your ethminers?
Sign In or Register to comment.