SILENTARMY v5: Zcash miner, 115 sol/s on R9 Nano, 70 sol/s on GTX 1070

[toptek]

Is the NV support being done in OpenCL or porting kernel to cuda?

sure windows support is needed, but NV should have priority, tested linux fee miner it goes 50+ on 1070 so I guess a lot more optimization is possible

I think NV support is V4 and windows is V5 which is better . i use AMD but also want a stable version for windows i hope is fee free i will donate mine every day for week or 2 20 mins a day freely if so . to show my ty

[Hotmetal]

any body know how to set fanspeed gpu at ubuntu 16.40??
cause at sensor, speed just run max 30% with temperature 78-79C,
make throttling..
i try to modded bios, but not make change..
im appreciated for solutions...

For my 6 GPUs:

echo 150 > /sys/devices/pci0000:00/0000:00:01.0/0000:01:00.0/hwmon/hwmon1/pwm1
echo 150 > /sys/devices/pci0000:00/0000:00:1c.0/0000:02:00.0/hwmon/hwmon2/pwm1
echo 150 > /sys/devices/pci0000:00/0000:00:1c.1/0000:03:00.0/hwmon/hwmon3/pwm1
echo 150 > /sys/devices/pci0000:00/0000:00:1c.2/0000:04:00.0/hwmon/hwmon4/pwm1
echo 150 > /sys/devices/pci0000:00/0000:00:1c.4/0000:06:00.0/hwmon/hwmon5/pwm1
echo 150 > /sys/devices/pci0000:00/0000:00:1c.5/0000:07:00.0/hwmon/hwmon6/pwm1

[mrb]

Is the NV support being done in OpenCL or porting kernel to cuda?

sure windows support is needed, but NV should have priority, tested linux fee miner it goes 50+ on 1070 so I guess a lot more optimization is possible

It's going to be in OpenCL. It's only going to be a <200-line patch to silentarmy. Way simpler than rewriting everything in CUDA.

[krnlx]

Is the NV support being done in OpenCL or porting kernel to cuda?

sure windows support is needed, but NV should have priority, tested linux fee miner it goes 50+ on 1070 so I guess a lot more optimization is possible

It's going to be in OpenCL. It's only going to be a <200-line patch to silentarmy. Way simpler than rewriting everything in CUDA.

What about performance ? with eXtremal's SA version on nvidia i got 30 s/s on one card, and 22 per card on 6-gpu rig. The bottleneck was high cpu usage.

[frozn0000]

I've done some googling but can't find anything that works to fix this. :/  First time trying ubuntu was a few days ago!

When trying to mine:

mine@mine:~/silentarmy$ ./silentarmy --use 0,1,2,3,4 -c stratum+tcp://zec-eu.suprnova.cc:2142 -u ***** -p *****
Connecting to zec-eu.suprnova.cc:2142
Solver 0.0: unexpected banner "Cannot get OpenCL platforms (-1001)"
Solver 0.1: unexpected banner "Cannot get OpenCL platforms (-1001)"
Solver 1.0: unexpected banner "Cannot get OpenCL platforms (-1001)"
Solver 1.1: unexpected banner "Cannot get OpenCL platforms (-1001)"
Solver 2.0: unexpected banner "Cannot get OpenCL platforms (-1001)"
Solver 2.1: unexpected banner "Cannot get OpenCL platforms (-1001)"
Solver 3.0: unexpected banner "Cannot get OpenCL platforms (-1001)"
Solver 3.1: unexpected banner "Cannot get OpenCL platforms (-1001)"
Solver 4.0: unexpected banner "Cannot get OpenCL platforms (-1001)"
Solver 4.1: unexpected banner "Cannot get OpenCL platforms (-1001)"
Stratum server sent us the first job
Mining on 5 devices

and also:

mine@mine:~/silentarmy$ ./silentarmy --list
Cannot get OpenCL platforms (-1001)
mine@mine:~/silentarmy$

[mrb]

frozn0000: try "clinfo" - does it report your GPUs?

[mrb]

What about performance ? with eXtremal's SA version on nvidia i got 30 s/s on one card, and 22 per card on 6-gpu rig. The bottleneck was high cpu usage.

I don't know we will have to see. OpenCL code has a tendency to need to be tweaked to run well on a device. Current silentarmy runs really well on AMD, but it may not be as efficient on Nvidia. So I will have work on optimizing for Nvidia and this is going to be incremental improvements released over time.

[zawawa]

I just sent a donation to your address.
Thank you for the great work!

[rednoW]

What about performance ? with eXtremal's SA version on nvidia i got 30 s/s on one card, and 22 per card on 6-gpu rig. The bottleneck was high cpu usage.

I don't know we will have to see. OpenCL code has a tendency to need to be tweaked to run well on a device. Current silentarmy runs really well on AMD, but it may not be as efficient on Nvidia. So I will have work on optimizing for Nvidia and this is going to be incremental improvements released over time.

It is wrong way from the start. You'll never achieve CUDA speeds with OpenCL on nVidia.

[felixbrucker]

What about performance ? with eXtremal's SA version on nvidia i got 30 s/s on one card, and 22 per card on 6-gpu rig. The bottleneck was high cpu usage.

I don't know we will have to see. OpenCL code has a tendency to need to be tweaked to run well on a device. Current silentarmy runs really well on AMD, but it may not be as efficient on Nvidia. So I will have work on optimizing for Nvidia and this is going to be incremental improvements released over time.

It is wrong way from the start. You'll never achieve CUDA speeds with OpenCL on nVidia.

i suppose we will need someone else to port it to cuda, has anyone yet offered help?

[bronsky]

any body know how to set fanspeed gpu at ubuntu 16.40??
cause at sensor, speed just run max 30% with temperature 78-79C,
make throttling..
i try to modded bios, but not make change..
im appreciated for solutions...

For card 0 :

Go to /sys/class/drm/card0/device/hwmon/hwmon1
and look up the min and max values for fanspeed for your card (files pwm1_min and pwm1_max).
Say, the pwm1_max says 255 and you want to set fanspeed to ~40%

Enter these commands, either diretly in a bash shell or in a shell script.

export PTH=/sys/class/drm/card0/device/hwmon/hwmon1
cat $PTH/pwm1
sudo bash -c "echo 102 > $PTH/pwm1"
cat $PTH/pwm1

The cat command will show you the value before and after you set the speed.
Hopefully, you may hear a change in the fanspeed.

For other cards locate the corresponding files.
Cheers

thanks for reply, i will try but, just type at terminal or must download something app?

[bronsky]

any body know how to set fanspeed gpu at ubuntu 16.40??
cause at sensor, speed just run max 30% with temperature 78-79C,
make throttling..
i try to modded bios, but not make change..
im appreciated for solutions...

For my 6 GPUs:

echo 150 > /sys/devices/pci0000:00/0000:00:01.0/0000:01:00.0/hwmon/hwmon1/pwm1
echo 150 > /sys/devices/pci0000:00/0000:00:1c.0/0000:02:00.0/hwmon/hwmon2/pwm1
echo 150 > /sys/devices/pci0000:00/0000:00:1c.1/0000:03:00.0/hwmon/hwmon3/pwm1
echo 150 > /sys/devices/pci0000:00/0000:00:1c.2/0000:04:00.0/hwmon/hwmon4/pwm1
echo 150 > /sys/devices/pci0000:00/0000:00:1c.4/0000:06:00.0/hwmon/hwmon5/pwm1
echo 150 > /sys/devices/pci0000:00/0000:00:1c.5/0000:07:00.0/hwmon/hwmon6/pwm1

i just type command at terminal bro?

type or paste ur commandline and hit enter
$ echo 150 > /sys/devices/pci0000:00/0000:00:01.0/0000:01:00.0/hwmon/hwmon1/pwm1
then type or paste ur commandline hit enter
$ echo 150 > /sys/devices/pci0000:00/0000:00:1c.0/0000:02:00.0/hwmon/hwmon2/pwm1
then type or paste ur commandline hit enter
$ echo 150 > /sys/devices/pci0000:00/0000:00:1c.1/0000:03:00.0/hwmon/hwmon3/pwm1
then type or paste ur commandline hit enter
$ echo 150 > /sys/devices/pci0000:00/0000:00:1c.2/0000:04:00.0/hwmon/hwmon4/pwm1
then type or paste ur commandline hit enter
$ echo 150 > /sys/devices/pci0000:00/0000:00:1c.4/0000:06:00.0/hwmon/hwmon5/pwm1
then type or paste ur commandline hit enter
$ echo 150 > /sys/devices/pci0000:00/0000:00:1c.5/0000:07:00.0/hwmon/hwmon6/pwm1
then type or paste ur commandline hit enter

how if i use 7 gpu broo..
and what mean of echo 150? how percent of fanspeed ? thanks newbie here

[Amph]

What about performance ? with eXtremal's SA version on nvidia i got 30 s/s on one card, and 22 per card on 6-gpu rig. The bottleneck was high cpu usage.

I don't know we will have to see. OpenCL code has a tendency to need to be tweaked to run well on a device. Current silentarmy runs really well on AMD, but it may not be as efficient on Nvidia. So I will have work on optimizing for Nvidia and this is going to be incremental improvements released over time.

It is wrong way from the start. You'll never achieve CUDA speeds with OpenCL on nVidia.

i suppose we will need someone else to port it to cuda, has anyone yet offered help?

i recall only a guy on the zcash forum that was going to port a cuda version that hash at 60 sol for a 1070, but he is still working on it, the bad thing is that his miner his private with 10% fee, and witht he recent dump i'm not sure anymore if it's worth it

[Streetlight9]

Is there a command in 16.04 to Hard-reboot Ubuntu? like in EthOS?

As dev2 some times drops from 50ish to 26... and it wont reboot(still runs) via ssh when I sudo reboot, have to go to the Rig to turn off.

http://i608.photobucket.com/albums/tt167/TheFlame1/2016-11-06_11-43_zpsw8knfrli.jpg

sudo Hard-reboot...

maybe?

[nwfella]

Figured I would go ahead and join the SilentArmy :p

Seeing as how I wanted to go ahead and check out this whole equihash deal figured this was as good a place as any.  Got 1x 280x and a 4gb 270x up and running at ~43sol/sec.

Is there a command in 16.04 to Hard-reboot Ubuntu? like in EthOS?

As dev2 some times drops from 50ish to 26... and it wont reboot(still runs) via ssh when I sudo reboot, have to go to the Rig to turn off.



sudo Hard-reboot...

maybe?

As for your question, personally I'm not aware of any 'hard-reboot' command other than sudo reboot.  Personally I've found just issuing a

Code:

pkill silentarmy

seems to do the trick.  Hope this helps

[krnlx]

As for your question, personally I'm not aware of any 'hard-reboot' command other than sudo reboot.  Personally I've found just issuing a

Code:

echo b > /proc/sysrq-trigger

As root. Not with sudo.

[nwfella]

As for your question, personally I'm not aware of any 'hard-reboot' command other than sudo reboot.  Personally I've found just issuing a

Code:

echo b > /proc/sysrq-trigger

As root. Not with sudo.

"I see said the blind man as he picked up the hammer and saw!"

No wonder I like linux so much, always a cool new trick to learn about!!

[frozn0000]

frozn0000: try "clinfo" - does it report your GPUs?

mine@mine:~$ clinfo
terminate called after throwing an instance of 'cl::Error'
  what():  clGetPlatformIDs
Aborted (core dumped)
mine@mine:~$

I also get a Ubuntu popup that says Sorry, the application clinfo has stopped unexpectedly.  If you notice further problems, try restarting the computer.

[frozn0000]

frozn0000: try "clinfo" - does it report your GPUs?

I went back and tried to install the official AMD drivers thinking that was the problem.

The following packages have unmet dependencies:
 unity-control-center : Depends: libcheese-gtk23 (>= 3.4.0) but it is not going to be installed
                        Depends: libcheese7 (>= 3.0.1) but it is not going to be installed
E: Error, pkgProblemResolver::Resolve generated breaks, this may be caused by held packages.
mine@mine:~/silentarmy$ 

I am on 14.04 using 5x 280x.  I was on 16.04 until I realized that they don't make drivers for the 280x for that version of Ubuntu so I downgraded.  Doesn't matter to me which version I am on.  This is strictly a mining machine.  Thanks for your time.

[eXtremal]

Can someone try this kernel with NVidia GPUs? Replace input cl file with this http://coinsforall.io/distr/input.cl