[ANN][X11/X13] X11 (Darkcoin)/X13 (Marucoin) miner (based on sph-sgminer)

[bumpy]

guys using smos/bamt, try using this, it is modified script from smos homepage

Code:

#!/bin/sh
mine stop
wget http://smos-linux.org/upgrade/sgminer.pl
wget http://smos-linux.org/upgrade/sgminer2.pl
rm -rf /opt/bamt/gpumon
mv sgminer2.pl /opt/bamt/gpumon
rm -rf /opt/bamt/common.pl
mv sgminer.pl /opt/bamt/common.pl
sleep 5
cd /opt/miners/
rm -rf cgminer
git clone https://github.com/lasybear/sph-sgminer_x11mod
mv /opt/miners/sph-sgminer_x11mod /opt/miners/sgminer
cp /opt/ADL/include/* /opt/miners/sgminer/ADL_SDK/
cd /opt/miners/sgminer/
./autogen.sh
sleep 5
make clean
sleep 5
CFLAGS="-O2 -Wall -march=native -I /opt/AMDAPP/include/" LDFLAGS="-L/opt/AMDAPP/lib/x86" ./configure --enable-scrypt --enable-darkcoin --enable-opencl --bindir="/opt/miners/sgminer" --prefix="/usr/local/bin"
sleep 5
make
sleep 5
rm -rf /root/sgminer-sph.sh
clear
echo "sgminer sph"
echo "SMOS-Linux.org"
echo "Now reboot your rig, and start mining"

[1714855500]

1714855500

[1714855500]

1714855500

[1714855500]

1714855500

[aznboy84]

ASUSR9280X-DC2-3GD5

Code:

sgminer.exe -k x11mod -o <host> -u <user> -p <pass> -I 19 -g 2 --gpu-memclock 1500 --gpu-engine 1090 --gpu-fan 70 --gpu-powertune 20 --thread-concurrency 8192 --lookup-gap 2

Only got ~2.5MH with lastest driver & rom 

[Elun]

Elun/all: thanks for the great work.

The latest version really helped for my 290 but my 5970 (2 gpu) went from 2*1.687Mh to 2*150Kh which more than lost out the gains on my 290. Is it possible to fix this? Same config, I tried a few tweaks, also running the 5970 at I 19 with no luck.

Did you try to change worksize?

Hi,
when running several testings and looking at sgminer logs, I noticed something very strange with sgminer :
GPUs are reordered in REVERSE by the miner !
GPU0 is on PCI5
GPU1 is on PCI4
GPU2 is on PCI3
GPU3 is on PCI2
GPU4 is on PCI1
GPU5 is on PCI0

I tried to set "gpu-reorder : false" but it didn't work
I strongly suggest you all to record miner's logs, using

Code:

sgminer $OPTIONS 2>> $LOG

note: this if for linux
then check your logs for

Code:

[14:10:34] [b]Mapping device[/b] 0 to GPU 5 according to Bus Number order
[14:10:34] Mapping device 1 to GPU 4 according to Bus Number order
[14:10:34] Mapping device 2 to GPU 3 according to Bus Number order
[14:10:34] Mapping device 3 to GPU 2 according to Bus Number order
[14:10:34] Mapping device 4 to GPU 1 according to Bus Number order
[14:10:34] Mapping device 5 to GPU 0 according to Bus Number order

Motherboard : ASRock H81ProBTC
OS : ubuntu 14.4

So, since idk when, I thought my Asus R920X was running on a 20% fan speed, but as a matter of facts, it was an SPH TRIX.....

edit:
stranger than weird, in the config file, I had to reorder the shaders/clocks in reverse. but I did not have to reorder FAN settings !

Try to use "gpu-map" configuration option. If i understand your post correctly.

[bullus]

nice job..
anyone tested with 6990 / 6970 / 5970 ?

6970 from 1600 to 1975kh/s

Code:

"kernel" : "x11mod",
"api-allow" : "W:127.0.0.1",
"api-listen" : true,
"api-mcast-port" : "4028",
"api-port" : "4028",
"expiry" : "1",
"failover-only" : true,
"kernel-path" : "/usr/local/bin",
"log" : "5",
"queue" : "1",
"scan-time" : "5",
"scrypt" : true,
"shares" : "0",
"failover-switch-delay" : "1",
"gpu-threads" : "2",
"gpu-dyninterval" : "7",
"gpu-engine" : "0",
"gpu-fan" : "60",
"gpu-platform" : "0",
"gpu-memclock" : "0",
"gpu-memdiff" : "0",
"gpu-powertune" : "8",
"gpu-vddc" : "0",
"intensity" : "15,",
"xintensity" : "0",
"temp-target" : "75",
"temp-overheat" : "85",
"temp-cutoff" : "95",
"temp-hysteresis" : "3",
"vectors" : "1",
"worksize" : "128",
"lookup-gap" : "0",
"shaders" : "1792,1536,1792,1792,1792",
"thread-concurrency" : "8192,8000,8192,8192,8192",
"no-pool-disable" : true,
"device" : "0,1,2,3,4"
}

second in config

[kocur]

ASUSR9280X-DC2-3GD5

Code:

sgminer.exe -k x11mod -o <host> -u <user> -p <pass> -I 19 -g 2 --gpu-memclock 1500 --gpu-engine 1090 --gpu-fan 70 --gpu-powertune 20 --thread-concurrency 8192 --lookup-gap 2

Only got ~2.5MH with lastest driver & rom 

try just

Code:

GPU_MAX_ALLOC_PERCENT=100
sgminer -o <host> -u <user> -p <pass> -k x11mod -I 18 -g 2 -w 256 --thread-concurrency 8192 --gpu-engine 1100 --gpu-memclock 1425

[oktay50000]

for me in windows its ok
but on bamt afer working 20~60 min sgminer closes itself
i tried diffrent voltages and clocks but no go
any tip?

[SimkoMiner]

this error is because there is sgminer.conf with undefinied pool, just rename this file to sgminer.co or something else if you use your own .bat file to run.
btw this miner work fine for me, everything is ok, and with -I 18 I get 7970 = 3,350 Mh, 7950 = 2,950 Mh 

Thank you Kocurik, it helped. I forgot to delete sgminer.conf file :-)

[djds]

For those having problems with 280x you can try this for the x11mod:

-I 15 -g 1 -w 128 --lookup-gap 2 --gpu-threads 2 --thread-concurrency 8192 --auto-gpu --gpu-engine 1080 --gpu-memclock 1500 --gpu-powertune -20 --auto-fan --gpu-fan 30-100 --temp-target 72 --temp-overheat 85 --temp-cutoff 95 --expiry 1 --scan-time 1 --queue 1 --no-submit-stale -s 1

It gives me 2.8 MH/s on both Sapphire and XFX 280x cards. It's undervolted to 1087 for XFX and 1112 for sapphire, so the cards work cooler. If you want to go higher you can reset the voltage to the card's defaults with trixx and try this:

-I 15 -g 1 -w 128 --lookup-gap 2 --gpu-threads 2 --thread-concurrency 8192 --auto-gpu --gpu-engine 1120 --gpu-memclock 1500 --gpu-powertune 20 --auto-fan --gpu-fan 30-100 --temp-target 72 --temp-overheat 85 --temp-cutoff 95 --expiry 1 --scan-time 1 --queue 1 --no-submit-stale -s 1

That gives 2.9 - 3.0 MH/s and even more.

I've spent some hours tunning, if it works you can send some pennies:

DRK: XxjDVJLGwKsXWUtT8rozu1C2ySk9crdcy7
BTC: 1CAKdW9QVWgjwxoj2jSsQU68s8kGukVwNB

Thanks!

[platinum4]

I just ate my own crow and got a SICK card warning with the standard 60s build after 5hours uptime.

Trying 120s SICK build.  Issues are only occurring when hooking up to profit-switching stratums.

[oktay50000]

can do these optimization for scrypt-n ?
i pay 20$ for it as donation

[Elun]

I just ate my own crow and got a SICK card warning with the standard 60s build after 5hours uptime.

Trying 120s SICK build.  Issues are only occurring when hooking up to profit-switching stratums.

Sucks. I'm total noob in sgminer sources. debug output will be very helpful. Can you add this to the miner params: --text-only --debug --verbose
Maybe i, maybe someone else will fix it.
Thanks.

[lasybear]

I just ate my own crow and got a SICK card warning with the standard 60s build after 5hours uptime.

Trying 120s SICK build.  Issues are only occurring when hooking up to profit-switching stratums.

Sucks. I'm total noob in sgminer sources. debug output will be very helpful. Can you add this to the miner params: --text-only --debug --verbose
Maybe i, maybe someone else will fix it.
Thanks.

+ redirect output to file.

[platinum4]

Yeah sure I don't mind doing that.

How do you redirect to output file?  I've tried to understand the piping thing in the readme.MD but failed.

Regardless, I think the most important stat is the WU/m.

A good average for me is WU: 0.057/m, anybody else seeing any improvements over +1/-1 of that over the course of a few hours?

[Remember remember the 5th of November]

Windows binaries will come soon of the latest git. Just have some work to take care of.

[cayars]

For those having problems with 280x you can try this for the x11mod:

-I 15 -g 1 -w 128 --lookup-gap 2 --gpu-threads 2 --thread-concurrency 8192 --auto-gpu --gpu-engine 1080 --gpu-memclock 1500 --gpu-powertune -20 --auto-fan --gpu-fan 30-100 --temp-target 72 --temp-overheat 85 --temp-cutoff 95 --expiry 1 --scan-time 1 --queue 1 --no-submit-stale -s 1

It gives me 2.8 MH/s on both Sapphire and XFX 280x cards. It's undervolted to 1087 for XFX and 1112 for sapphire, so the cards work cooler. If you want to go higher you can reset the voltage to the card's defaults with trixx and try this:

-I 15 -g 1 -w 128 --lookup-gap 2 --gpu-threads 2 --thread-concurrency 8192 --auto-gpu --gpu-engine 1120 --gpu-memclock 1500 --gpu-powertune 20 --auto-fan --gpu-fan 30-100 --temp-target 72 --temp-overheat 85 --temp-cutoff 95 --expiry 1 --scan-time 1 --queue 1 --no-submit-stale -s 1

That gives 2.9 - 3.0 MH/s and even more.

I've spent some hours tunning, if it works you can send some pennies:

DRK: XxjDVJLGwKsXWUtT8rozu1C2ySk9crdcy7
BTC: 1CAKdW9QVWgjwxoj2jSsQU68s8kGukVwNB

Thanks!

-I 18 --gpu-engine 1150 --gpu-memclock 1500 -g 2 -w 256 --lookup-gap 2 --thread-concurrency 8192

That gives me 3.345 Mh/s running in closed case at 70 to 72F

[n2nshad0w]

This is great!  I've noticed HUGE gains in my cards!

my 7770 and running over 1.1!  and my 6950's @ 1.9 Thanks!

[Elun]

Yeah sure I don't mind doing that.

How do you redirect to output file?  I've tried to understand the piping thing in the readme.MD but failed.

Regardless, I think the most important stat is the WU/m.

A good average for me is WU: 0.057/m, anybody else seeing any improvements over +1/-1 of that over the course of a few hours?

You can add 2>logfilename
Maybe Remember remember the 5th of November build will be better, without such problem.
Sorry for my bad english.

[platinum4]

Hey 5th, I know you posted a paste of the x11mod kernel before, but can you combine x11 mod with what you see different between marucoin.cl (http://pastebin.com/X97XdwU3) X13 algo and the standard darkcoin.cl?

The x11mod kernel is some 8000 lines of code the x13 kernel much much shorter, but less efficient.  I wanted to see if we could optimize 11 of the 13 algorithms for upcoming use.  X13 is basically X11 hamsi+fugue i think.  So can we just smash the hamsi+fugue portion to bottom of the 8000 lines or what.

[tobenary]

X

-I 18 --gpu-engine 1150 --gpu-memclock 1500 -g 2 -w 256 --lookup-gap 2 --thread-concurrency 8192

That gives me 3.345 Mh/s running in closed case at 70 to 72F

I think I read somewhere that if you put i=18 to 280x you will fry them.
( same as if you put more then i=20 to 290 cards )

everybody can correct me if I'm wrong.

[platinum4]

Only the 290X Tri-X OC Cards falling DEAD with this miner.  Nothing physically wrong with them.  Hard clocking to 1040/1300 [box-stock] will report in.

Maybe 5th can compile a better version, or a version compiled for SSEv2 or something above a Pentium MMX.  Error the miner gives is "suggest hard hang" then I don't have GPUs on restart because that usually always crashes the entire miner; I'd rather run at half or 2/3 capacity.