<<  >> (p.13)
    Author Topic: [LOCKED] cpuminer-opt v3.12.3, open source optimized multi-algo CPU miner  (Read 444202 times)

    • pallas
    Legendary
    *
    Offline Offline

    Activity: 2716
    Merit: 1094


    Black Belt Developer


    View Profile
    January 29, 2016, 08:56:55 AM
     #241
    Quote from: joblo on January 29, 2016, 03:20:33 AM
    It didn't take long to hit a brick wall with windows.

    static inline void transform(cubehashParm *sp )

    Expected '(' to follow 'inline'

    WTF?

     I guess that settles it.

    visual studio or mingw?

  • joblo (OP)
    Legendary
    *
    Offline Offline

    Activity: 1470
    Merit: 1114


    View Profile
    January 29, 2016, 09:50:25 AM
     #242
    Quote from: pallas on January 29, 2016, 08:56:55 AM
    Quote from: joblo on January 29, 2016, 03:20:33 AM
    It didn't take long to hit a brick wall with windows.

    static inline void transform(cubehashParm *sp )

    Expected '(' to follow 'inline'

    WTF?

     I guess that settles it.

    visual studio or mingw?

    VS. It worked for me for ccminer so I was hoping if I got the project file right it would work.
    I was very careful with it matching the file list side by side with the makefile. There were
    windows versions of all the miners I harvested from but I didn't compile any of them
    myself.

    I'll try mingw again, this time in a VM
    AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
    https://bt.irlbtc.com/view/5226770.msg53865575#msg53865575
    BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,

  • djm34
    Legendary
    *
    Offline Offline

    Activity: 1400
    Merit: 1050


    View Profile WWW
    January 29, 2016, 10:00:15 AM
     #243
    Quote from: joblo on January 29, 2016, 09:50:25 AM
    Quote from: pallas on January 29, 2016, 08:56:55 AM
    Quote from: joblo on January 29, 2016, 03:20:33 AM
    It didn't take long to hit a brick wall with windows.

    static inline void transform(cubehashParm *sp )

    Expected '(' to follow 'inline'

    WTF?

     I guess that settles it.

    visual studio or mingw?

    VS. It worked for me for ccminer so I was hoping if I got the project file right it would work.
    I was very careful with it matching the file list side by side with the makefile. There were
    windows versions of all the miners I harvested from but I didn't compile any of them
    myself.

    I'll try mingw again, this time in a VM

    Epsylon3 version of cpuminer compiles well with visual studio
    djm34 facebook page
    BTC: 1NENYmxwZGHsKFmyjTc5WferTn5VTFb7Ze
    Pledge for neoscrypt ccminer to that address: 16UoC4DmTz2pvhFvcfTQrzkPTrXkWijzXw

  • joblo (OP)
    Legendary
    *
    Offline Offline

    Activity: 1470
    Merit: 1114


    View Profile
    January 29, 2016, 10:12:54 AM
     #244
    Quote from: djm34 on January 29, 2016, 10:00:15 AM
    Quote from: joblo on January 29, 2016, 09:50:25 AM
    Quote from: pallas on January 29, 2016, 08:56:55 AM
    Quote from: joblo on January 29, 2016, 03:20:33 AM
    It didn't take long to hit a brick wall with windows.

    static inline void transform(cubehashParm *sp )

    Expected '(' to follow 'inline'

    WTF?

     I guess that settles it.

    visual studio or mingw?

    VS. It worked for me for ccminer so I was hoping if I got the project file right it would work.
    I was very careful with it matching the file list side by side with the makefile. There were
    windows versions of all the miners I harvested from but I didn't compile any of them
    myself.

    I'll try mingw again, this time in a VM

    Epsylon3 version of cpuminer compiles well with visual studio

    I'll try compiling darkcoin minerd and cp3u. If they don't both compile then I know I have a bigger
    problem.
    AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
    https://bt.irlbtc.com/view/5226770.msg53865575#msg53865575
    BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,

  • zTheWolfz
    Full Member
    ***
    Offline Offline

    Activity: 231
    Merit: 150



    View Profile
    January 29, 2016, 01:22:55 PM
    Last edit: January 29, 2016, 01:39:26 PM by zTheWolfz
     #245
    Quote from: joblo on January 29, 2016, 10:12:54 AM
    Quote from: djm34 on January 29, 2016, 10:00:15 AM
    Quote from: joblo on January 29, 2016, 09:50:25 AM
    Quote from: pallas on January 29, 2016, 08:56:55 AM
    Quote from: joblo on January 29, 2016, 03:20:33 AM
    It didn't take long to hit a brick wall with windows.

    static inline void transform(cubehashParm *sp )

    Expected '(' to follow 'inline'

    WTF?

     I guess that settles it.

    visual studio or mingw?

    VS. It worked for me for ccminer so I was hoping if I got the project file right it would work.
    I was very careful with it matching the file list side by side with the makefile. There were
    windows versions of all the miners I harvested from but I didn't compile any of them
    myself.

    I'll try mingw again, this time in a VM

    Epsylon3 version of cpuminer compiles well with visual studio

    I'll try compiling darkcoin minerd and cp3u. If they don't both compile then I know I have a bigger
    problem.

    Not sure if there is a un-compiled version or Linux version of cpuminer_x64_SSE2, but the .exe of this
    will mine x11 & a few other coins in Windows7-x64. No AES support, none of the cpuminer_x64_avx flavors
    worked on any of my rigs, unsupported hardware? Using the cpuminer_x64_sse2 on the AMD rig ATM mining x11.
    AMD rig has Server R2012 R2 as a OS. 42 day trail left on that. lol

    Edit: On another note I have a old AMD FX duel CPU motherboard & hardware setting collecting dust if I can fine the time
    from RL I'll setup a test bed with that "IF" I can find a spare PSU to use with it & a spot somewhere to place it. lol

  • joblo (OP)
    Legendary
    *
    Offline Offline

    Activity: 1470
    Merit: 1114


    View Profile
    January 29, 2016, 04:30:56 PM
     #246
    Quote from: zTheWolfz on January 29, 2016, 01:22:55 PM

    Not sure if there is a un-compiled version or Linux version of cpuminer_x64_SSE2, but the .exe of this
    will mine x11 & a few other coins in Windows7-x64. No AES support, none of the cpuminer_x64_avx flavors
    worked on any of my rigs, unsupported hardware? Using the cpuminer_x64_sse2 on the AMD rig ATM mining x11.
    AMD rig has Server R2012 R2 as a OS. 42 day trail left on that. lol

    Edit: On another note I have a old AMD FX duel CPU motherboard & hardware setting collecting dust if I can fine the time
    from RL I'll setup a test bed with that "IF" I can find a spare PSU to use with it & a spot somewhere to place it. lol

    I don't follow your first paragraph. Are you talking about another fork? What is cpuminer_x64_sse2?
    AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
    https://bt.irlbtc.com/view/5226770.msg53865575#msg53865575
    BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,

  • zTheWolfz
    Full Member
    ***
    Offline Offline

    Activity: 231
    Merit: 150



    View Profile
    January 29, 2016, 04:51:15 PM
    Last edit: January 29, 2016, 05:38:38 PM by zTheWolfz
     #247
    Quote from: joblo on January 29, 2016, 04:30:56 PM
    Quote from: zTheWolfz on January 29, 2016, 01:22:55 PM

    Not sure if there is a un-compiled version or Linux version of cpuminer_x64_SSE2, but the .exe of this
    will mine x11 & a few other coins in Windows7-x64. No AES support, none of the cpuminer_x64_avx flavors
    worked on any of my rigs, unsupported hardware? Using the cpuminer_x64_sse2 on the AMD rig ATM mining x11.
    AMD rig has Server R2012 R2 as a OS. 42 day trail left on that. lol

    Edit: On another note I have a old AMD FX duel CPU motherboard & hardware setting collecting dust if I can fine the time
    from RL I'll setup a test bed with that "IF" I can find a spare PSU to use with it & a spot somewhere to place it. lol

    I don't follow your first paragraph. Are you talking about another fork? What is cpuminer_x64_sse2?
    Its found in the \NiceHashMiner_v1.2.2.2\bin folder after unpacked

    runline: cpuminer_x64_SSE2 -a x11 -o stratum+tcp://x11.usa.nicehash.com:3336 -u 18tvS3deKZK5q4eTtPRWYeEMWmGmuErHgz.H8QG6 -p d=0.01 -t 32



    Edit: Apparently is seems to be cpuminer-muti 1.1 renamed.? My guess is Nichhash software conf line makes it use sse2 when ran with it,
    but with no options for any of the GPU algos to be mined on the CPU. Since I don't see sse2 being used in my run or maybe sse2 is used but just not printed as in use.

     

  • joblo (OP)
    Legendary
    *
    Offline Offline

    Activity: 1470
    Merit: 1114


    View Profile
    January 29, 2016, 07:00:06 PM
     #248
    Quote from: zTheWolfz on January 29, 2016, 04:51:15 PM
    Quote from: joblo on January 29, 2016, 04:30:56 PM
    Quote from: zTheWolfz on January 29, 2016, 01:22:55 PM

    Not sure if there is a un-compiled version or Linux version of cpuminer_x64_SSE2, but the .exe of this
    will mine x11 & a few other coins in Windows7-x64. No AES support, none of the cpuminer_x64_avx flavors
    worked on any of my rigs, unsupported hardware? Using the cpuminer_x64_sse2 on the AMD rig ATM mining x11.
    AMD rig has Server R2012 R2 as a OS. 42 day trail left on that. lol

    Edit: On another note I have a old AMD FX duel CPU motherboard & hardware setting collecting dust if I can fine the time
    from RL I'll setup a test bed with that "IF" I can find a spare PSU to use with it & a spot somewhere to place it. lol

    I don't follow your first paragraph. Are you talking about another fork? What is cpuminer_x64_sse2?
    Its found in the \NiceHashMiner_v1.2.2.2\bin folder after unpacked

    runline: cpuminer_x64_SSE2 -a x11 -o stratum+tcp://x11.usa.nicehash.com:3336 -u 18tvS3deKZK5q4eTtPRWYeEMWmGmuErHgz.H8QG6 -p d=0.01 -t 32



    Edit: Apparently is seems to be cpuminer-muti 1.1 renamed.? My guess is Nichhash software conf line makes it use sse2 when ran with it,
    but with no options for any of the GPU algos to be mined on the CPU. Since I don't see sse2 being used in my run or maybe sse2 is used but just not printed as in use.

     

    OK, that's not mine, you'll have to talk to nicehash about that.
    AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
    https://bt.irlbtc.com/view/5226770.msg53865575#msg53865575
    BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,

  • zTheWolfz
    Full Member
    ***
    Offline Offline

    Activity: 231
    Merit: 150



    View Profile
    January 29, 2016, 08:13:44 PM
    Last edit: January 29, 2016, 08:33:32 PM by zTheWolfz
     #249
    Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
    It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

    I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
    Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
    But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.

  • joblo (OP)
    Legendary
    *
    Offline Offline

    Activity: 1470
    Merit: 1114


    View Profile
    January 29, 2016, 09:22:41 PM
     #250
    Quote from: zTheWolfz on January 29, 2016, 08:13:44 PM
    Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
    It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

    I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
    Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
    But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.

    I'm confused. Your first paragraph says you aren't asking for help and the AMD rig is working.
    But the second one is saying you are having a problems with 3.0.7 and the AMD rig isn't working.

    Is the problem with cpuminer-opt or the rig?

    Thanks for the info abnout cpumier-multi. It seems to claim SSE2 optimizations so how does it
    compare with cpuminer-opt? Maybe they did a better job.

    You're probably right about 3.0.2 performance. I'm reviewing the performance of all releases because
    I think some of my tweaks actually hurt performance.

    Thanks for the extensive testing.
    AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
    https://bt.irlbtc.com/view/5226770.msg53865575#msg53865575
    BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,

  • zTheWolfz
    Full Member
    ***
    Offline Offline

    Activity: 231
    Merit: 150



    View Profile
    January 29, 2016, 10:24:15 PM
     #251
    Quote from: joblo on January 29, 2016, 09:22:41 PM
    Quote from: zTheWolfz on January 29, 2016, 08:13:44 PM
    Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
    It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

    I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
    Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
    But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.

    I'm confused. Your first paragraph says you aren't asking for help and the AMD rig is working.
    But the second one is saying you are having a problems with 3.0.7 and the AMD rig isn't working.

    Is the problem with cpuminer-opt or the rig?

    Thanks for the info abnout cpumier-multi or . It seems to claim SSE2 optimizations so how does it
    compare with cpuminer-opt? Maybe they did a better job.

    You're probably right about 3.0.2 performance. I'm reviewing the performance of all releases because
    I think some of my tweaks actually hurt performance.

    Thanks for the extensive testing.

    AMD is working with cpuminer-muti or cpuminer_x64_SSE2 in windows, in VM your versions doesn't work for the AMD rig.
    Yours loads now and appears to be working but at a low load rate "20% or less" and I never see any shares completed with the amount of time I've let it run.
    Over 2 hours nothing.

  • joblo (OP)
    Legendary
    *
    Offline Offline

    Activity: 1470
    Merit: 1114


    View Profile
    January 30, 2016, 01:19:36 AM
    Last edit: January 30, 2016, 01:39:13 AM by joblo
     #252
    Quote from: zTheWolfz on January 29, 2016, 10:24:15 PM
    Quote from: joblo on January 29, 2016, 09:22:41 PM
    Quote from: zTheWolfz on January 29, 2016, 08:13:44 PM
    Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
    It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

    I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
    Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
    But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.

    I'm confused. Your first paragraph says you aren't asking for help and the AMD rig is working.
    But the second one is saying you are having a problems with 3.0.7 and the AMD rig isn't working.

    Is the problem with cpuminer-opt or the rig?

    Thanks for the info abnout cpumier-multi or . It seems to claim SSE2 optimizations so how does it
    compare with cpuminer-opt? Maybe they did a better job.

    You're probably right about 3.0.2 performance. I'm reviewing the performance of all releases because
    I think some of my tweaks actually hurt performance.

    Thanks for the extensive testing.

    AMD is working with cpuminer-muti or cpuminer_x64_SSE2 in windows, in VM your versions doesn't work for the AMD rig.
    Yours loads now and appears to be working but at a low load rate "20% or less" and I never see any shares completed with the amount of time I've let it run.
    Over 2 hours nothing.

    OK so there is a problem with cpuminer-opt in a Linux VM on your AMD rig but the other miners work ok
    natively in Windows on the same machine? Is this correct?

    How does the performance compare between cpuminer-multi and cpuminer_x64_SSE2. It has SSE2 in
    the name but does it have the performance? It should be significantly faster than than multi.

    Edit and another thing you could try the other miners in the VM?This eliminates another variable. Thanks.
    When did it break for cpuminer-opt?
    AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
    https://bt.irlbtc.com/view/5226770.msg53865575#msg53865575
    BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,

  • zTheWolfz
    Full Member
    ***
    Offline Offline

    Activity: 231
    Merit: 150



    View Profile
    January 30, 2016, 01:53:04 AM
     #253
    Quote from: joblo on January 30, 2016, 01:19:36 AM
    Quote from: zTheWolfz on January 29, 2016, 10:24:15 PM
    Quote from: joblo on January 29, 2016, 09:22:41 PM
    Quote from: zTheWolfz on January 29, 2016, 08:13:44 PM
    Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
    It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

    I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
    Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
    But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.

    I'm confused. Your first paragraph says you aren't asking for help and the AMD rig is working.
    But the second one is saying you are having a problems with 3.0.7 and the AMD rig isn't working.

    Is the problem with cpuminer-opt or the rig?

    Thanks for the info abnout cpumier-multi or . It seems to claim SSE2 optimizations so how does it
    compare with cpuminer-opt? Maybe they did a better job.

    You're probably right about 3.0.2 performance. I'm reviewing the performance of all releases because
    I think some of my tweaks actually hurt performance.

    Thanks for the extensive testing.

    AMD is working with cpuminer-muti or cpuminer_x64_SSE2 in windows, in VM your versions doesn't work for the AMD rig.
    Yours loads now and appears to be working but at a low load rate "20% or less" and I never see any shares completed with the amount of time I've let it run.
    Over 2 hours nothing.

    OK so there is a problem with cpuminer-opt in a Linux VM on your AMD rig but the other miners work ok
    natively in Windows on the same machine? Is this correct?

    How does the performance compare between cpuminer-multi and cpuminer_x64_SSE2. It has SSE2 in
    the name but does it have the performance? It should be significantly faster than than multi.

    Edit and another thing you could try the other miners in the VM?This eliminates another variable. Thanks.
    When did it break for cpuminer-opt?
    #1 yes

    #2 Haven't had much of a change to take a close look at those numbers yet.
    RL is being a pain atm.

    #3 When I get the time I'll be more then glad to, I did try or tried to compile cpuminer-multi  with no luck in the VM so far.
    No source for cpuminer_x64_SSE2 to compile that I'm aware of.

    #4 hasn't never worked for the AMD rig, but by this time next week with any luck I'll have my new CPU's up and running so it will no longer matter.
    I will try to get you as much info as I can if you'd like that before the change over, but looks like I maybe the only one running this type of setup.
    So that may not help anyone, but I willing to see what I can dig up.

  • joblo (OP)
    Legendary
    *
    Offline Offline

    Activity: 1470
    Merit: 1114


    View Profile
    January 30, 2016, 06:45:49 AM
    Last edit: January 30, 2016, 10:52:27 PM by joblo
     #254
    The following may be of interest to devs of cpuminer, ccminer and maybe others.

    Update:

    More architectural changes made. Most declarations moved from cpu-miner.c
    to miner.h so no files will need to include it. The system I am famiiar with could prevent
    being embedded adn it was the default. The owner had to explicitly allow  inclusion
    and to whom. Nice feature.

    I still have to migrate full_test and get_opt to use the gate. I will have to split up the options
    into common options and algo specific options. cpuminer-c will handler the common options
    and the algos will handle the algo-specific options that apply to them.

    I will also look into enhancing the registration process to have the algo and its custom
    options added to the argument list and help file. These are the only remaining holes in the gate,
    to my knowledge.

    The gate is almost complete. It's called a gate because it is the only path between cpu-miner.c
    and the algos. A very popular technique where I come from.

    A developpers preview should be avilable in a couple of days.

    I am hopeful the rearchitecting will help solve some of the other nagging issues I have been having,
    like flipping between undefined and multi defined.

    Previous post:
    --------
    I've done a lot of rebuilding under the hood because  miner_thread had become
    polluted with too much algo specific code. Well, that is all gone.
    Folllowing is a summary of the changes. I will make a preview available before release.

    - restructured the file system to be more modular with respect to algos

    - removed bloat added to support multiple CPU targets.

    - removed all algo specific code from miner.h and cpu-miner.c, and eventually util.c
       and anywhere else I find it, it all lives now in the algo's source file.

    - adding algos doesn't require any changes to base code, a registry interface file requires adding one entry.

    - redesigned the scanhash engine to use function pointers instead of switch/case.

    - removed many long switch/case statements.

    - created a registry where algos register their functions with cpu-miner.

    - registry functions include scanhash, hash, init_ctx, get_max64, and get_custom_opts with plans
      to add full_test and any others I find.

    - enabled initialization of algo context outside of miner_thread loop.

    - changes do not increase performance directly but enables more efficient management of contexts
      which helps certain algos, specificaly quark.
    AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
    https://bt.irlbtc.com/view/5226770.msg53865575#msg53865575
    BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,

  • pallas
    Legendary
    *
    Offline Offline

    Activity: 2716
    Merit: 1094


    Black Belt Developer


    View Profile
    January 30, 2016, 09:30:16 AM
     #255
    Nice work! I like clean and modular code!

  • joblo (OP)
    Legendary
    *
    Offline Offline

    Activity: 1470
    Merit: 1114


    View Profile
    January 30, 2016, 02:28:40 PM
     #256
    Quote from: pallas on January 30, 2016, 09:30:16 AM
    Nice work! I like clean and modular code!

    Thanks, I haven't yet found a way to run multiple entry functions upon program load..
    The system I'm familiar with could declare any file with a main function to have it be run
    automaticaaly when the program is loaded. Essentially it functions similarly to a OS init
    system but for an application. If I can do that in c/c++ I could plug the only remaining
    crack in the wall between algo code and base code. If I can't do it  from within the
    algo file I might have to write an init system.
    AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
    https://bt.irlbtc.com/view/5226770.msg53865575#msg53865575
    BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,

  • joblo (OP)
    Legendary
    *
    Offline Offline

    Activity: 1470
    Merit: 1114


    View Profile
    January 30, 2016, 03:06:32 PM
     #257
    Quote from: zTheWolfz on January 30, 2016, 01:53:04 AM
    Quote from: joblo on January 30, 2016, 01:19:36 AM
    Quote from: zTheWolfz on January 29, 2016, 10:24:15 PM
    Quote from: joblo on January 29, 2016, 09:22:41 PM
    Quote from: zTheWolfz on January 29, 2016, 08:13:44 PM
    Wasn't asking for help with it was just bringing it to you attention in case you didn't know about it.
    It and cpumimer-muti both run GPU algos with my limited testing with each on Intel & AMD CPU's in windows no VM or 2nd OS needed.

    I did compile 3.07 on 2 Intel rigs & the AMD rig, still no luck with the AMD, only seeing a 20% load max no matter the core amount used.
    Intel on the other hand runs about the same as 3.02 that was in use, 3.0.7 from a quick look was just a tad lower on hashrate than 3.0.2.
    But that was just the start up numbers, long term that could change, will have to check it later once its been working a while.

    I'm confused. Your first paragraph says you aren't asking for help and the AMD rig is working.
    But the second one is saying you are having a problems with 3.0.7 and the AMD rig isn't working.

    Is the problem with cpuminer-opt or the rig?

    Thanks for the info abnout cpumier-multi or . It seems to claim SSE2 optimizations so how does it
    compare with cpuminer-opt? Maybe they did a better job.

    You're probably right about 3.0.2 performance. I'm reviewing the performance of all releases because

    I think some of my tweaks actually hurt performance.

    Thanks for the extensive testing.

    AMD is working with cpuminer-muti or cpuminer_x64_SSE2 in windows, in VM your versions doesn't work for the AMD rig.
    Yours loads now and appears to be working but at a low load rate "20% or less" and I never see any shares completed with the amount of time I've let it run.
    Over 2 hours nothing.

    OK so there is a problem with cpuminer-opt in a Linux VM on your AMD rig but the other miners work ok
    natively in Windows on the same machine? Is this correct?

    How does the performance compare between cpuminer-multi and cpuminer_x64_SSE2. It has SSE2 in
    the name but does it have the performance? It should be significantly faster than than multi.

    Edit and another thing you could try the other miners in the VM?This eliminates another variable. Thanks.
    When did it break for cpuminer-opt?
    #1 yes

    #2 Haven't had much of a change to take a close look at those numbers yet.
    RL is being a pain atm.

    #3 When I get the time I'll be more then glad to, I did try or tried to compile cpuminer-multi  with no luck in the VM so far.
    No source for cpuminer_x64_SSE2 to compile that I'm aware of.

    #4 hasn't never worked for the AMD rig, but by this time next week with any luck I'll have my new CPU's up and running so it will no longer matter.
    I will try to get you as much info as I can if you'd like that before the change over, but looks like I maybe the only one running this type of setup.
    So that may not help anyone, but I willing to see what I can dig up.

    No pressure, you can test whver you pleae or not at all. I'm very appreciative of everything you've done so far.

    It appears the VM is the issue. I don't know why it woulg connect to stratum and get new work but never
    start hashing. At least I know where to start looking.

    I can test cpuminer_x64_SSE2 myself, it was just a curiosity.

    Many thanks.
    AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
    https://bt.irlbtc.com/view/5226770.msg53865575#msg53865575
    BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,

  • joblo (OP)
    Legendary
    *
    Offline Offline

    Activity: 1470
    Merit: 1114


    View Profile
    January 30, 2016, 04:17:20 PM
     #258
    How to use algo gate, for devs.

    algo gate is the new miner_thread engine for running an algo in the correct environment.

    It makes adding new algos easy. Here's how:

    1. define the usual functions in the algo's source file. scanhash and hash functions are mandatory

    2. Define a registration function

    Code:
    bool register_x11_algo( algo_gate_t* gate )
    {
      gate->init_ctx               = &init_x11_aes_ctx;
      gate->scanhash            = &scanhash_x11_aes;
      gate->hash                    = &x11_hash_aes;
    // gate->get_custom_opt = &x11_get_custom_opt;
    // gate->get_max64         = &x11_get_max64;
      return true;
    };

    3. add "#include "algo-gate-api.h" to the algo's source file.

    4. add a new entry to initialize_algo_gate in algo-gate-api.c to register your algo

    Code:
         case ALGO_X11:
            register_x11_algo( &algo_gate );
            break;

    5. Compile and run. When cpuminer starts it will use the algo registry
        without needing to know anything about the algo except it's name.

    That's all there is to it. Any miner using the Garzick/Pooler front end can implement this.
    The only pain is retrofitting all the existing algos.


    More to come
    AKA JayDDee, cpuminer-opt developer. https://github.com/JayDDee/cpuminer-opt
    https://bt.irlbtc.com/view/5226770.msg53865575#msg53865575
    BTC: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT,

  • zTheWolfz
    Full Member
    ***
    Offline Offline

    Activity: 231
    Merit: 150



    View Profile
    January 30, 2016, 05:55:55 PM
     #259
    Great work.

    Well looks like I will be doing that upgrade sooner than I thought. Seller shipped my stuff
    usps 2 day priority showing it should be here today in a matter of hour actually, Postman runs early on Sat.   Cool

  • zTheWolfz
    Full Member
    ***
    Offline Offline

    Activity: 231
    Merit: 150



    View Profile
    January 30, 2016, 07:39:10 PM
    Last edit: January 30, 2016, 08:18:15 PM by zTheWolfz
     #260
    Okay here is were I am: Installed Ubuntu on a small 64GB SSD I had laying around.
    Everything looks to have gone fine with the compile of v3.0.7. But

    I'm getting this below once mining is started.

    Illegal instruction (core dumped)
     then the miner dies...
    Code:
    thewolf@thewolf-H8QG6:~/Downloads/cpuminer-opt-3.0.7$ ./cpuminer -a x11 -o stratum+tcp://x11.usa.nicehash.com:3336 -u 18tvS3deKZK5q4eTtPRWYeEMWmGmuErHgz.H8QG6 -p x

             **********  cpuminer-opt  v3.0.7  ***********
         BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT
         Forked from TPruvot's cpuminer-multi-1.2pre with credits
         to Lucas Jones, elmad, palmd, djm34, pooler, ig0tik3d,
         Wolf0 and Jeff Garzik.

    Checking CPU capatibility...
            AMD Opteron(tm) Processor 6124 HE
    CPU & SW Build Supports AES_NI: NO
    [2016-01-30 13:34:37] Starting Stratum on stratum+tcp://x11.usa.nicehash.com:3336
    [2016-01-30 13:34:37] 32 miner threads started, using 'x11' algorithm.
    [2016-01-30 13:34:37] Stratum difficulty set to 0.04
    [2016-01-30 13:34:37] x11.usa.nicehash.com:3336 x11 block 662269
    Illegal instruction (core dumped)
    thewolf@thewolf-H8QG6:~/Downloads/cpuminer-opt-3.0.7$

    On zpool.ca  & xpool.ca I get  the below.

    Code:

    thewolf@thewolf-H8QG6:~/Downloads/cpuminer-opt-3.0.7$ ./cpuminer -o stratum+tcp://mine.zpool.ca:3533 -u 18tvS3deKZK5q4eTtPRWYeEMWmGmuErHgz -p x -t 16

             **********  cpuminer-opt  v3.0.7  ***********
         BTC donation address: 12tdvfF7KmAsihBXQXynT6E6th2c2pByTT
         Forked from TPruvot's cpuminer-multi-1.2pre with credits
         to Lucas Jones, elmad, palmd, djm34, pooler, ig0tik3d,
         Wolf0 and Jeff Garzik.

    Checking CPU capatibility...
            AMD Opteron(tm) Processor 6124 HE
    CPU & SW Build Supports AES_NI: NO
    [2016-01-30 13:54:17] Starting Stratum on stratum+tcp://mine.zpool.ca:3533
    [2016-01-30 13:54:17] 16 miner threads started, using 'scrypt' algorithm.
    [2016-01-30 13:54:18] Stratum difficulty set to 0.016
    [2016-01-30 13:54:18] mine.zpool.ca:3533 scrypt block 577802
    [2016-01-30 13:54:18] CPU #8: 4.50 kH/s
    [2016-01-30 13:54:18] CPU #0: 4.49 kH/s
    [2016-01-30 13:54:18] CPU #14: 4.46 kH/s
    [2016-01-30 13:54:18] accepted: 0/1 (0.00%), 13.44 kH/s nooooo
    [2016-01-30 13:54:18] reject reason: Invalid share
    [2016-01-30 13:54:18] CPU #0: 4.42 kH/s
    [2016-01-30 13:54:18] CPU #11: 4.50 kH/s
    [2016-01-30 13:54:18] CPU #1: 4.49 kH/s
    [2016-01-30 13:54:19] CPU #3: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #4: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #2: 4.24 kH/s
    [2016-01-30 13:54:19] CPU #6: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #5: 4.33 kH/s
    [2016-01-30 13:54:19] CPU #7: 4.47 kH/s
    [2016-01-30 13:54:19] CPU #9: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #5: 4.42 kH/s
    [2016-01-30 13:54:18] reject reason: Invalid share
    [2016-01-30 13:54:18] CPU #0: 4.42 kH/s
    [2016-01-30 13:54:18] CPU #11: 4.50 kH/s
    [2016-01-30 13:54:18] CPU #1: 4.49 kH/s
    [2016-01-30 13:54:19] CPU #3: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #4: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #2: 4.24 kH/s
    [2016-01-30 13:54:19] CPU #6: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #5: 4.33 kH/s
    [2016-01-30 13:54:19] CPU #7: 4.47 kH/s
    [2016-01-30 13:54:19] CPU #9: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #5: 4.42 kH/s
    [2016-01-30 13:54:18] CPU #14: 4.46 kH/s
    [2016-01-30 13:54:18] accepted: 0/1 (0.00%), 13.44 kH/s nooooo
    [2016-01-30 13:54:18] reject reason: Invalid share
    [2016-01-30 13:54:18] CPU #0: 4.42 kH/s
    [2016-01-30 13:54:18] CPU #11: 4.50 kH/s
    [2016-01-30 13:54:18] CPU #1: 4.49 kH/s
    [2016-01-30 13:54:19] CPU #3: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #4: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #2: 4.24 kH/s
    [2016-01-30 13:54:19] CPU #6: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #5: 4.33 kH/s
    [2016-01-30 13:54:19] CPU #7: 4.47 kH/s
    [2016-01-30 13:54:19] CPU #9: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #5: 4.42 kH/s
    [2016-01-30 13:54:18] CPU #11: 4.50 kH/s
    [2016-01-30 13:54:18] CPU #1: 4.49 kH/s
    [2016-01-30 13:54:19] CPU #3: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #4: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #2: 4.24 kH/s
    [2016-01-30 13:54:19] CPU #6: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #5: 4.33 kH/s
    [2016-01-30 13:54:19] CPU #7: 4.47 kH/s
    [2016-01-30 13:54:19] CPU #9: 4.50 kH/s
    [2016-01-30 13:54:19] CPU #5: 4.42 kH/s
    [2016-01-30 13:54:19] CPU #15: 4.50 kH/s
    [2016-01-30 13:54:19] accepted: 0/2 (0.00%), 71.50 kH/s nooooo
    [2016-01-30 13:54:19] reject reason: Invalid share
    [2016-01-30 13:54:19] accepted: 0/3 (0.00%), 71.50 kH/s nooooo
    [2016-01-30 13:54:19] reject reason: Invalid share
    [2016-01-30 13:54:19] CPU #9: 4.50 kH/s
    [2016-01-30 13:54:19] accepted: 0/4 (0.00%), 71.51 kH/s nooooo
    Pages: « 1 2 3 4 5 6 7 8 9 10 11 12 [13] 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 ... 197 »
      Print  
Page 12
Viewing Page: 13

IRLBTC™ © 2025 IRLBTC.com