[ANN] sgminer v5 - optimized X11/X13/NeoScrypt/Lyra2RE/etc. kernel-switch miner

[Xmm]

Please anybody upload sgminer+conf with .bin files for 280x. Try lots of versions and configs result HW errors or 268Khs only

I'm not the only one with such a request
Thank You.


R9 290:
Try using https://github.com/wolf9466/neo-sgminer5-hawaii with
-I 15 -g 1 -w 64 --thread-concurrency 32768  (use Wolf0 .bin) on R290 and receive only 275 Khs

[kobol21]

SGMINER-

After cloning the latest git (81d9c9e574866cd5e10794875872c9c08b11bd2b), 3 days old today, I follwed the *.nix build instructions:

git submodule init
git submodule update
autoreconf -i
CFLAGS=CFLAGS="-O2 -Wall -march=native -I /opt/AMDAPPSDK-2.9-1/include/" LDFLAGS="-L/opt/AMDAPPSDK-2.9-1/lib/x86_64/" ./configure --enable-neoscrypt
make

I get a compile error:

make[2]: Leaving directory `/home/borqd/master/sph'
make[2]: Entering directory `/home/borqd/master'
  CC       sgminer-sgminer.o
sgminer.c: In function ‘stratum_sthread’:
sgminer.c:5500:15: warning: unused variable ‘nonce2_64’ [-Wunused-variable]
     uint64_t *nonce2_64;
               ^
sgminer.c: In function ‘main’:
sgminer.c:8476:3: error: ‘s’ undeclared (first use in this function)
   s = strdup(argv[0]);
   ^
sgminer.c:8476:3: note: each undeclared identifier is reported only once for each function it appears in
make[2]: *** [sgminer-sgminer.o] Error 1
make[2]: Leaving directory `/home/borqd/master'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/borqd/master'
make: *** [all] Error 2

My computer is at the command prompt with no runtime sgminer file.  What is wrong?       -scryptr

edit:  I checked the source code and the error reports appear to be correct.  I am guessing that the errors are typos, but I haven't been able to correct them.       --scryptr

I'm have this error too. Please help

Same problem for me. Anyone please help ?

[whitecloud]

Hi,
Win 8.1 x64, 4.9, R200 280x.
I put 14.6 RC2 AMD OCL (4 files with *.dll estension) on folder with sgminer, but:

Started sgminer 5.1.0-dev
* using Jansson 2.6
clDeviceNum returned error, no GPUs usable
All devices disabled, cannot mine!

What is wrong ?

Try adding gpu-platform 1  to your config.

Works!
Thank you!

[Pansyfaust]

Hi,
Win 8.1 x64, 4.9, R200 280x.
I put 14.6 RC2 AMD OCL (4 files with *.dll estension) on folder with sgminer, but:

Started sgminer 5.1.0-dev
* using Jansson 2.6
clDeviceNum returned error, no GPUs usable
All devices disabled, cannot mine!

What is wrong ?

Try adding gpu-platform 1  to your config.

Works!
Thank you!

Happy hashing

[scryptr]

SGMINER-

After cloning the latest git (81d9c9e574866cd5e10794875872c9c08b11bd2b), 3 days old today, I follwed the *.nix build instructions:

git submodule init
git submodule update
autoreconf -i
CFLAGS="-O2 -Wall -march=native -I /opt/AMDAPPSDK-2.9-1/include/" LDFLAGS="-L/opt/AMDAPPSDK-2.9-1/lib/x86_64/" ./configure --enable-neoscrypt
make

I get a compile error:

make[2]: Leaving directory `/home/borqd/master/sph'
make[2]: Entering directory `/home/borqd/master'
  CC       sgminer-sgminer.o
sgminer.c: In function ‘stratum_sthread’:
sgminer.c:5500:15: warning: unused variable ‘nonce2_64’ [-Wunused-variable]
     uint64_t *nonce2_64;
               ^
sgminer.c: In function ‘main’:
sgminer.c:8476:3: error: ‘s’ undeclared (first use in this function)
   s = strdup(argv[0]);
   ^
sgminer.c:8476:3: note: each undeclared identifier is reported only once for each function it appears in
make[2]: *** [sgminer-sgminer.o] Error 1
make[2]: Leaving directory `/home/borqd/master'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/borqd/master'
make: *** [all] Error 2

My computer is at the command prompt with no runtime sgminer file.  What is wrong?       -scryptr

edit:  I checked the source code and the error reports appear to be correct.  I am guessing that the errors are typos, but I haven't been able to correct them.       --scryptr

I'm have this error too. Please help

Same problem for me. Anyone please help ?

Mr. troky authored a new build with corrections a couple hours ago (7f88bab72576c31d1013dd8b161b4231587173b4).   It compiles on Ubuntu 14.04.1.  It runs.   I need to learn to use the new features.        -scryptr

[damm315er]

Is there a windows version compiled with the corrected libraries yet?

[platinum4]

Is there a windows version compiled with the corrected libraries yet?

We're waiting on Wolf0 and/or Slix for that I think.

[Ignition75]

4 x 290X (Hynix) overclocked @ 1200/1500

http://store.picbg.net/pubpic/0E/A7/80c68e6dbdb20ea7.png

For configuration check: https://github.com/wolf9466/neo-sgminer5-hawaii/blob/master/neoscrypt.conf

wait a minute it no good getting at Wolf0 about this couse this guy got loads of hash BUT look at those temps and fan speeds,,,
that has got to be water cooled or not 14.6 drivers or inside an igloo

Not really, dude - open air rig in cold weather with good coolers will produce that.

Nope. Watercooling FTW and driver Catalyst 14.6

Nice 

[spoonuk]

Any higher speed than 325kH/s on 280X ?

What's your config on 280x? I'm only getting about 285kh/s per card.

[Oldriga]

Any higher speed than 325kH/s on 280X ?

What's your config on 280x? I'm only getting about 285kh/s per card.

Same question to go6ooo1212. My speed only 272Kh/s

[qwep1]

Elpida vs. Hynix, people.

286 Elpida 1100\1500
315 Hynix  1100\1500   280X

[gotqn]

Could anyone share HashRate/settings for Darkcoin using R280x?

[go6ooo1212]

Definitely Elpida sucks, I got rid of all cards with elpida memory...

[semajjames]

my r9 290 elipa's get the same as my r9 290 hynix's on neoscrypt

only difference is the hynix can go much higher engin clocks because they are flashed with 290x bios

[damm315er]

my r9 290 elipa's get the same as my r9 290 hynix's on neoscrypt

only difference is the hynix can go much higher engin clocks because they are flashed with 290x bios

All my stuff has stock bios, but my 290's with hynix, outwork my 290x's with elpida, by about 10%..  The 290's get 333k, while the 290x get 301k.

Still haven't gotten the last wolf release to run, need a compiled release.

[go6ooo1212]

290 should get around 350 kH/s , I think.

[jch9678]

I have a couple problems with my miners that maybe someone has some ideas on how to fix them.  The one that I want to tackle first is I don't think sgminer is making the correct bins, if that makes any sense.  For testing purposes I'm only mining x11 with this config.

Code:

{
  "pools": [
    {
      "name": "NiceHash x11 Multi Failover",
      "url": "stratum+tcp://stratum.westhash.com:3336",
      "user": "x",
      "pass": "x",
      "profile": "x11"
    }
  ],
  "profiles": [
    {
      "name": "x11",
      "algorithm": "darkcoin-mod",
      "xintensity" : "744",
      "worksize": "256",
      "gpu-engine": "0",
      "gpu-memclock": "0",
      "gpu-fan": "30-100"
    },
    {
      "name": "x13",
      "algorithm": "marucoin-mod",
      "worksize": "256",
      "gpu-engine": "0",
      "gpu-memclock": "0",
      "gpu-fan": "30-100"
    },
    {
      "name": "x14",
      "algorithm": "x14",
      "worksize": "256",
      "gpu-engine": "0",
      "gpu-memclock": "0",
      "gpu-fan": "30-100"
    },
    {
      "name": "x15",
      "algorithm": "bitblock",
      "worksize": "256",
      "gpu-engine": "0",
      "gpu-memclock": "0",
      "gpu-fan": "30-100"
    },
    {
      "name": "nist5",
      "algorithm": "talkcoin-mod",
      "worksize": "256",
      "gpu-engine": "0",
      "gpu-memclock": "0",
      "gpu-fan": "30-100"
    },
    {
      "name": "qubit",
      "algorithm": "qubitcoin",
      "worksize": "256",
      "gpu-engine": "0",
      "gpu-memclock": "0",
      "gpu-fan": "30-100"
    },
    {
      "name": "scrypt",
      "algorithm": "alexkarnew",
      "lookup-gap": "2",
      "thread-concurrency": "27400",
      "worksize": "256",
      "gpu-engine": "1000",
      "gpu-memclock": "1250",
      "gpu-fan": "30-100"
    },
    {
      "name": "nscrypt",
      "algorithm": "alexkarnew",
      "nfactor": "11",
      "lookup-gap": "2",
      "thread-concurrency": "22500",
      "worksize": "256",
      "gpu-engine": "1000",
      "gpu-memclock": "1250",
      "gpu-fan": "30-100"
    },
    {
      "name": "keccak",
      "algorithm": "maxcoin",
      "lookup-gap": "2",
      "thread-concurrency": "32765",
      "worksize": "256",
      "gpu-engine": "0",
      "gpu-memclock": "0",
      "gpu-fan": "30-100"
    }
  ],
  "failover-only": true,
  "device": "all",
  "gpu-threads": "1",
  "temp-cutoff": "95,95,95,95,95,95",
  "temp-overheat": "85,85,85,85,85,85",
  "temp-target": "78,78,78,78,78,78",
  "gpu-reorder": true,
  "gpu-memdiff": "0,0,0,0,0,0",
  "shares": "0",
  "kernel-path": "/usr/i586-mingw32msvc/bin",
  "api-allow": "W:127.0.0.1",
  "api-listen": true,
  "api-mcast-port": "4028",
  "api-port": "4028",
  "auto-fan": true,
  "blake-compact": true,
  "expiry": "1",
  "failover-switch-delay": "30",
  "gpu-dyninterval": "7",
  "gpu-platform": "1",
  "hamsi-expand-big": "7",
  "hamsi-short": true,
  "keccak-unroll": "8",
  "luffa-parallel": true,
  "log": "5",
  "no-pool-disable": true,
  "no-client-reconnect": true,
  "no-restart": true,
  "queue": "1",
  "scan-time": "1",
  "show-coindiff": true,
  "tcp-keepalive": "30",
  "temp-hysteresis": "3"
}

With the lastest sgminer from nicehash I can get 5.6 Mh/s with a reference r9 290x on Stilt's bios with stock clocks.  The bin that is created is called darkcoin-modHawaiigw256l4ku8bclp.bin
This happens with Wolf0's and Badman64's builds as well.  In the name you can see ku8, bc, and lp but where is big7 and hs?  I know those abbreviations get appended on the name because I have a test machine that produced this bin, marucoin-modHawaiigw256l4big1hs.bin  I tried putting "hamsi-expand-big": "7" and "hamsi-short": true, in the x11 profile but there was no change.

Am I right in assuming that sgminer is not building a bin with the 2 hamsi parameters?  Or does sgminer just not include all parameters in the file name?  I just need to figure this one out first before I work on the second problem, which is for the life of me I cannot get 2 threads to work.  Thanks in advance.

[jch9678]

X11 does not use Hamsi - it's not going to do anything when given the options for it! Therefore, for X11, those options have no effect and are omitted in the binary name.

Thanks for the answer.  Man, I feel like I should've figured that out.  Do you have any ideas why I'm having such a hard time trying 2 gpu threads.  I use what seem to be the de facto standard: worksize 64 and xintensity 64 but I get nothing but constant driver crashes.  I was able to get cards hashing with a worksize of 48 but the hash was terrible.  Are you running your cards with Stilt's bios, just curious if some of the things Stilt disabled would have an effect.

Incidentally, a few posts back I think you wrote that you don't attach a monitor to your 290 card, do you mean you are using onboard graphics or is it just attached to another card and that's why "gpu-platform" : "-1" is used.  Never seen the -1 .  Anyway I appreciate the work you put into all this even if I can't afford it, yet.

[jch9678]

X11 does not use Hamsi - it's not going to do anything when given the options for it! Therefore, for X11, those options have no effect and are omitted in the binary name.

Thanks for the answer.  Man, I feel like I should've figured that out.  Do you have any ideas why I'm having such a hard time trying 2 gpu threads.  I use what seem to be the de facto standard: worksize 64 and xintensity 64 but I get nothing but constant driver crashes.  I was able to get cards hashing with a worksize of 48 but the hash was terrible.  Are you running your cards with Stilt's bios, just curious if some of the things Stilt disabled would have an effect.

Incidentally, a few posts back I think you wrote that you don't attach a monitor to your 290 card, do you mean you are using onboard graphics or is it just attached to another card and that's why "gpu-platform" : "-1" is used.  Never seen the -1 .  Anyway I appreciate the work you put into all this even if I can't afford it, yet.

I don't use -1, but actually, several of my rigs are headless - they have no monitor at all, and I connect to them via SSH. I also don't run Stilt's BIOSes, currently. Running on Windows? What driver version?

Unrelated, I've worked on my x13 & x15 some more, broke 5MH/s on x13 for 280X 

Screenshot (nsfw): https://ottrbutt.com/miner/x13localrigwolf-11242014-2.png

Yeah, Windows 8.1 I've tried 14.6 rc2 and 14.7, I think I even tried 13.12 when I merge mined vtc and zerovert for a day.  Congrats on the x13 speed, it's going to be a real bummer when your miner cl's let you mine x13 faster than I can mine x11.  Hahaha.  Btw, my girlfriend thinks the chick on your left monitor is hotter than the one on the right.

[jch9678]

X11 does not use Hamsi - it's not going to do anything when given the options for it! Therefore, for X11, those options have no effect and are omitted in the binary name.

Thanks for the answer.  Man, I feel like I should've figured that out.  Do you have any ideas why I'm having such a hard time trying 2 gpu threads.  I use what seem to be the de facto standard: worksize 64 and xintensity 64 but I get nothing but constant driver crashes.  I was able to get cards hashing with a worksize of 48 but the hash was terrible.  Are you running your cards with Stilt's bios, just curious if some of the things Stilt disabled would have an effect.

Incidentally, a few posts back I think you wrote that you don't attach a monitor to your 290 card, do you mean you are using onboard graphics or is it just attached to another card and that's why "gpu-platform" : "-1" is used.  Never seen the -1 .  Anyway I appreciate the work you put into all this even if I can't afford it, yet.

I don't use -1, but actually, several of my rigs are headless - they have no monitor at all, and I connect to them via SSH. I also don't run Stilt's BIOSes, currently. Running on Windows? What driver version?

Unrelated, I've worked on my x13 & x15 some more, broke 5MH/s on x13 for 280X 

Screenshot (nsfw): https://ottrbutt.com/miner/x13localrigwolf-11242014-2.png

Yeah, Windows 8.1 I've tried 14.6 rc2 and 14.7, I think I even tried 13.12 when I merge mined vtc and zerovert for a day.  Congrats on the x13 speed, it's going to be a real bummer when your miner cl's let you mine x13 faster than I can mine x11.  Hahaha.  Btw, my girlfriend thinks the chick on your left monitor is hotter than the one on the right.

Hm... I know the Win drivers tend to be unstable - perhaps try lowering clocks? Also, it's surprising you can mine x11 at 5MH/s with a 280X - I thought around 4.5MH/s was the limit.

She has pretty good taste; I agree. 

Ah sorry I meant it'll be an even bigger bummer when you can mine x13 with a 280x faster than I can mine x11 with a 290x.  So sad to write that.  I've been running at stock clocks and i prefer to use afterburner, maybe it's time to start increasing the powertune and voltage and/or flip the bios switch.  So many combinations, makes my head hurt.  Do you have reference cards?