CUDA miner

1464749515269

Comments

  • retherrether Member Posts: 258 ✭✭
    Mo35 said:

    rether said:

    Does anyone have a copy of the latest compiled with CUDA 6.5?

    ill try to find time this weekend and compile 1.0.7 , if im successful i`ll post it
    That'd be great...still fighting with Windows 10 and my 760. I have an AMD rig, but this is my main computer and it'd be nice to let it mine when not in use.

    Unfortunately, nothing I've done has given me decent speeds in Win10. I tried the previous build with CUDA 6.5 and the 347 drivers but still only getting like 3mh/s. If I take this card and put it in an old Vista system I have it hits 13 mh/s right away. Super annoying.
  • dlehenkydlehenky Member Posts: 2,249 ✭✭✭✭
    @Genoil All 22 v107b1 ethminer instances stopped mining when the last DAG changed. The process was still running, but none of them were doing anything, no cpu time being consumed, nothing. I restarted them all and they all began mining again fine. ????
  • retherrether Member Posts: 258 ✭✭
    I don't know if this is a bug or not, but when I use "--opencl-device" it always selects the first GPU only (device 0). For example, "--opencl-device 2" will still use device 0 only.
  • newkidONdablocknewkidONdablock interwebzMember Posts: 121
    same here, i use opencl platform for my r9s
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    dlehenky said:

    @Genoil All 22 v107b1 ethminer instances stopped mining when the last DAG changed. The process was still running, but none of them were doing anything, no cpu time being consumed, nothing. I restarted them all and they all began mining again fine. ????

    I should have an instance running with a debugger during the next dag change, can you ping me when it's due? What connection where you using? getwork or stratum? And which pool?

    @rether yes that's a bug i still need to look into. there is a flag --opencl-devices that allow you to specify more than 1 device. you can also use it with a single device. if that's also broken then it's an amd thing, because i tested that thoroughly with multiple nvidia opencl devices.
  • dlehenkydlehenky Member Posts: 2,249 ✭✭✭✭
    @Genoil Stratum to ethpool. Sure, I can ping you; I've been paying close attention the last couple of DAGs.
  • dr_pradr_pra Member Posts: 445 ✭✭✭
    If I find some time next week I will setup a DAG switch simulator so that you will be able to test your setup.
  • retherrether Member Posts: 258 ✭✭
    Genoil said:



    @rether yes that's a bug i still need to look into. there is a flag --opencl-devices that allow you to specify more than 1 device. you can also use it with a single device. if that's also broken then it's an amd thing, because i tested that thoroughly with multiple nvidia opencl devices.

    Ok, --opencl-devices seems to work but if I try to run more than one instance of ethminer the second one crashes right away. My intention with using --opencl-devices was to run separate command lines for different cards. I did use "--no-precompute" on all but the first instance.
  • LogicaluserLogicaluser Member Posts: 214 ✭✭
    edited April 2016
    @Genoil
    I suspect the DAG switch issues are related to lack of system RAM with multiple cards
    (something like previously used RAM not being freed up to load the new DAG?)

    Many (but not all) of my rigs have been getting stuck on the recent DAG transitions... however for yesterday's transition I noticed that all the rigs that mined straight through had 8GB or more of RAM.

    Another slightly related issue, the last few DAG increases no longer allow any of my 5 card AMD rigs to mine on systems with 4GB RAM (Win7 x64 and 15.7.1 drivers). Oddly enough, it didn't happen all on the same transition... some rigs refused to mine without more RAM a week or two ago, whereas one rig was just fine up until yesterday's transition, after which it too required more RAM.

    Oh and neither issue is unique to your builds, also happens with qtminer as well
  • dlehenkydlehenky Member Posts: 2,249 ✭✭✭✭
    @Logicaluser All my rigs have 8 GB, all got "stuck" on the DAG transition (v107b1).
  • cornlingercornlinger Member Posts: 107
    Hey guys,

    still struggling with my rig. Now I've installed the CUDA ethminer to fix all my problems.

    This all worked very well, but I just don't know how to start CUDA miner with stratum failover support. I wasn't able to find any tutorials or anything to get it running like I want it :(

    I use Ubuntu 14.04.4 LTS and a rig with 6x Radeon HD7970 on ethermine.org


    Can anybody help me to get CUDA miner running?


    Thank you!
    Florian

    PS: When I got all this fixed, I'm going to write a full tutorial in German and English to prevent other users from my failures and mistakes.
  • dlehenkydlehenky Member Posts: 2,249 ✭✭✭✭
    @cornlinger Just out of curiosity, have you ever run "ethminer --help"? All the options available are given there. It would show you, for instance, the '-FS' option, which is for specifying a stratum failover address.
  • cornlingercornlinger Member Posts: 107
    @dlehenky Yes, I've checked it and my miner is running, but I wasn't able to use it with stratum.

    Current starting command:
    ./ethminer -F http://eu1.ethermine.org:5555/.worker -G --cl-local-work 128 --cl-global-work 8192 
    
    What to insert now? -FS won't work, the same with -S.
  • dlehenkydlehenky Member Posts: 2,249 ✭✭✭✭
    @cornlinger You do know the stratum port is 3333, not 5555, right?
  • SS2006SS2006 Member Posts: 2
    rether said:

    Mo35 said:

    rether said:

    Does anyone have a copy of the latest compiled with CUDA 6.5?

    ill try to find time this weekend and compile 1.0.7 , if im successful i`ll post it
    That'd be great...still fighting with Windows 10 and my 760. I have an AMD rig, but this is my main computer and it'd be nice to let it mine when not in use.

    Unfortunately, nothing I've done has given me decent speeds in Win10. I tried the previous build with CUDA 6.5 and the 347 drivers but still only getting like 3mh/s. If I take this card and put it in an old Vista system I have it hits 13 mh/s right away. Super annoying.
    Mo I'd greatly appreceiate that too. I'm using the 347.52 driver on windows 10. and would love to use cuda so I can limit my devices and hash power (so i can use my PC while mining). Right now i'm stuck with OpenCL because im using the latest ethminer compiled with the latest cuda (from cryptomining blog)
  • SS2006SS2006 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

    links are down :( do you mind kindly re-uploading? Unless you have the latest 1.0.7 beta compiled with cuda 6.5 :D
    Like you, im in windows 10 using 347.52 (if any newer driver works too for max hash rate, let me know!)
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    I've got 1.0.5 built with CUDA 6.5 here: https://github.com/Genoil/cpp-ethereum/tree/master/releases/cuda-6.5

    New 1.0.7 binaries are up as well. Nothing really changed since 1.0.7beta, except some small stuff.

    Some ideas for 1.0.8:

    - stratum work timeout (if no new work comes from the server within x seconds, reconnect or failover)
    - get OpenCL speed on CUDA exactly as fast as native CUDA
    - change console output to show more relevant information
    - fix DAG switch crash
    - binary kernels for faster startup time

    Not sure if everything will make it into 1.0.8.
  • dlehenkydlehenky Member Posts: 2,249 ✭✭✭✭
    @Genoil I'm currently reflashing all my GPUs, but once that's done, I'll work on the GPU errors/rejects. Hopefully, if it checks out for you, you could include it in v108, if you want.
  • newkidONdablocknewkidONdablock interwebzMember Posts: 121
    edited April 2016
    how do i connect to ethermine via stratum ? so I can measure my valid/invalid shares? I will also test v107 and post my results
    Post edited by newkidONdablock on
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    edited April 2016
    -S {URL} -FS {FailoverURL} -O 0xADDRESS.worker
  • newkidONdablocknewkidONdablock interwebzMember Posts: 121
    Genoil said:

    -S {URL} -FS {FailoverURL} -O 0xADDRESS.worker

    does this work with ethproxy? or with this i dont need ethproxy ?
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    No need. I don't know if it accepts stratum clients? That would be useful.
  • newkidONdablocknewkidONdablock interwebzMember Posts: 121
    edited April 2016
    testing now, speed/stability is as v106 only now accurately shows valids/invalids, good job bro
  • retherrether Member Posts: 258 ✭✭
    I don't know if anyone has an answer to this but using the 347 driver and a CUDA6.5 version seems to fix the issues with Win10 for 9xx series cards.

    However, for my single GTX 760 it does not fix it. If I take THAT card and put it in a Vista or Win7 system it will hash at about 13 or 14mh/s but on Win10 it still only gets about 3mh/s even with the CUDA6.5 version of ethminer and the 347 nvidia drivers.

    I honestly can't figure it out.
  • PustulPustul Member Posts: 2
    Hi rether,

    I've got the same problem. I've also tried old win7 driver with Cuda 6.5 on win10 but I only got 3-4 MH/s... Now I've built the miner in Ubuntu and I've got 13-14 MH/s out of the box. Too bad it doesn't work on Win10!
  • retherrether Member Posts: 258 ✭✭
    Pustul said:

    Hi rether,

    I've got the same problem. I've also tried old win7 driver with Cuda 6.5 on win10 but I only got 3-4 MH/s... Now I've built the miner in Ubuntu and I've got 13-14 MH/s out of the box. Too bad it doesn't work on Win10!

    Ya, there is something different with the 7xx series that keeps them from working on Win10 even with the older CUDA versions. I rolled back to a really old driver as well with no change. I wish I knew what it was!
  • PustulPustul Member Posts: 2
    Oh I forgot to mention that I have a GTX 670. But I think they are quite close.
  • retherrether Member Posts: 258 ✭✭
    Pustul said:

    Oh I forgot to mention that I have a GTX 670. But I think they are quite close.

    Ya, I think it's any of the non-9xx series cards that have this issue.
  • TankEngineTankEngine Member Posts: 2
    Hy
    I need help
    I always get this error in visual 2013, after i like to build ethminer.sln
    https://drive.google.com/file/d/0B-uFu7k6r1fgREJZNnRkbjNXekk/view?usp=sharing

    I tried 2 pc and win 10/7
    pls help!
  • GenoilGenoil 0xeb9310b185455f863f526dab3d245809f6854b4dMember Posts: 769 ✭✭✭
    @TankEngine did you install the CUDA toolkit?
Sign In or Register to comment.