EWBF's CUDA Zcash miner

[sublimus]

I've been using AMD rigs for a while, and I thought I'd try my hand with some CUDA boards.

I have a rig with Windows 10x64, 378.66 drivers, and three Geforce 1080 boards. For some reasons, Windows keeps shutting down one of the boards. If I click on it and have it find the drivers, it install older drivers. All three boards do work then however with the older driver.

When I run EWBF, the hashrate starts out around what I would expect and then quickly decays.

My questions:
1. Are the 378.66 drivers required? If so, any clues on why one of the boards doesn't like it?
2. Why the decaying hashrate? I've not mucked with any settings yet. These are stock.
3. Are the hash rates from Linux comparable to Windows? I'm much more comfortable with Unix.

NOTE: I do have the BIOS set to integrated graphics chip. This I've found makes managing things much easier. The motherboard is a MSI Z97.

I would bet that if you switch the card with another that would indicate the culprit is the slot it is in.
I think that it indicates a conflict with the integrated graphic.
Tried Disabling it ?

[fatsailor]

The problem turned out to be a bad USB cable. I went ahead and installed Ubuntu 16.04 server as I was tired of playing with device manager.

All three cards are now chugging along at ~470 sols/s.

Anyway know how to tweak the cards under linux ala MSI Afterburner?

[induktor]

My questions:
1. Are the 378.66 drivers required? If so, any clues on why one of the boards doesn't like it?
2. Why the decaying hashrate? I've not mucked with any settings yet. These are stock.
3. Are the hash rates from Linux comparable to Windows? I'm much more comfortable with Unix.

NOTE: I do have the BIOS set to integrated graphics chip. This I've found makes managing things much easier. The motherboard is a MSI Z97.

Hi, no no need, i am using 367.44 (lubuntu 14.04 x64) and works perfectly on 1070, 1060 and 970, i don't have a 1080 but if 1070 works , 1080 should too, i upgraded one of the rigs to 378 just to see what changes ,and nothing does, stable as a rock, never 1 rig freezes, unloads, or reboot, 100% stable since the last mod about 7 months ago.

hashrate decay?, you mean profitability?, well price went down, difficulty went up, so ... we are fucked!

if you don't overclock then yes, hashrates at the same target power limit, are similar.

i don't recommend you use the onboard GPU, better disable it and use one of the GPUs, today gpus are powerful enough to handle graphical linux desktop and mine at 100% load without any trouble  

[PressEnterNoaw]

The problem turned out to be a bad USB cable. I went ahead and installed Ubuntu 16.04 server as I was tired of playing with device manager.

All three cards are now chugging along at ~470 sols/s.

Anyway know how to tweak the cards under linux ala MSI Afterburner?

Go ahead and install the last driver from Nvidia. I'm using ver. 375.26, but there is currently 1 new release 375.39 and a beta driver 378.09 available (haven't tested them out yet)

I don't think there is much to gain with the driver versions, most important is to find a working and stable one for your system.

2nd:

Go ahead and enable OC with:

sudo nvidia-xconfig -a --cool-bits=28 --allow-empty-initial-configuration

after a reboot you gain access to OC and fan control in X conf

Hope that helped you a little

[mirny]

The problem turned out to be a bad USB cable. I went ahead and installed Ubuntu 16.04 server as I was tired of playing with device manager.

All three cards are now chugging along at ~470 sols/s.

Anyway know how to tweak the cards under linux ala MSI Afterburner?

Go ahead and install the last driver from Nvidia. I'm using ver. 375.26, but there is currently 1 new release 375.39 and a beta driver 378.09 available (haven't tested them out yet)

I don't think there is much to gain with the driver versions, most important is to find a working and stable one for your system.

2nd:

Go ahead and enable OC with:

sudo nvidia-xconfig -a --cool-bits=28 --allow-empty-initial-configuration

after a reboot you gain access to OC and fan control in X conf

Hope that helped you a little

how you can enable OC, lets say for 3, or 4 cards from terminal?

[PallMall]

Does any one know how to setup automatic rig reboot if one of GPU stoped?

[alrami]

Does any one know how to setup automatic rig reboot if one of GPU stoped?

Try to use following:

---------------------------------------------
:start
miner.exe --server eu1-zcash.flypool.org --user YourWallet.YourWorker --pass YourPass --port 3333 --eexit 1
ping 127.0.0.1 > nul
goto start
---------------------------------------------

[dtawom]

I couldn't find anything in the help about about it, but is there switch to set a GPU usage percentage with EWBF?  I have two 1070's in my personal rig with 3 monitors.  I watch a lot of streaming video online and when the miner is running I get a lot of skipping and stuttering.  It would be nice to be able to set 5% or so aside for my general computer usage rather than killing mining on a whole card just so I can watch netflix.

[PallMall]

Does any one know how to setup automatic rig reboot if one of GPU stoped?

Try to use following:

---------------------------------------------
:start
miner.exe --server eu1-zcash.flypool.org --user YourWallet.YourWorker --pass YourPass --port 3333 --eexit 1
ping 127.0.0.1 > nul
goto start
---------------------------------------------

Thank you,
This will restart miner itself but I'm looking for windows reboot.

[V_I_T_A_L_Y]

EWBF_ when will the new release

[EWBF_]

EWBF_ when will the new release

Now i try "squeeze" all what are possible from current algorithm, unfortunately improvements for one card can reduce speed for another. So when I find a way to maximize the improvement of various cards in the same time, I will publish it, but at this point there is no ETA.

[pintycar]

EWBF_ when will the new release

Now i try "squeeze" all what are possible from current algorithm, unfortunately improvements for one card can reduce speed for another. So when I find a way to maximize the improvement of various cards in the same time, I will publish it, but at this point there is no ETA.

Thanks for the update EWBF_ ; did you get a chance to have a look at the ZERO project? It'd be great to have a GPU-miner from a respected dev like you, as all there is right now is CPU based. Thanks!

[EWBF_]

EWBF_ when will the new release

Now i try "squeeze" all what are possible from current algorithm, unfortunately improvements for one card can reduce speed for another. So when I find a way to maximize the improvement of various cards in the same time, I will publish it, but at this point there is no ETA.

Thanks for the update EWBF_ ; did you get a chance to have a look at the ZERO project? It'd be great to have a GPU-miner from a respected dev like you, as all there is right now is CPU based. Thanks!

Yes. This is simply equihash with a different parameters. But at this moment zcash miner more important. In the future, I will create a miner with these parameters.

[pintycar]

Thanks for the update EWBF_ ; did you get a chance to have a look at the ZERO project? It'd be great to have a GPU-miner from a respected dev like you, as all there is right now is CPU based. Thanks!

Yes. This is simply equihash with a different parameters. But at this moment zcash miner more important. In the future, I will create a miner with these parameters.

OK, thanks for replying man. It's fair enough, I understand your position. Maybe in a near future if ZERO becomes popular then you'll jump on it. Thanks again.

[ioglnx]

EWBF_ when will the new release

Now i try "squeeze" all what are possible from current algorithm, unfortunately improvements for one card can reduce speed for another. So when I find a way to maximize the improvement of various cards in the same time, I will publish it, but at this point there is no ETA.

Thanks for the update EWBF_ ; did you get a chance to have a look at the ZERO project? It'd be great to have a GPU-miner from a respected dev like you, as all there is right now is CPU based. Thanks!

Yes. This is simply equihash with a different parameters. But at this moment zcash miner more important. In the future, I will create a miner with these parameters.

Why not make this params init during call..

Like miner.exe --equiinitparam=x,x which would be easy to make

[disman]

EWBF_ when will the new release

Now i try "squeeze" all what are possible from current algorithm, unfortunately improvements for one card can reduce speed for another. So when I find a way to maximize the improvement of various cards in the same time, I will publish it, but at this point there is no ETA.

Thanks for the update EWBF_ ; did you get a chance to have a look at the ZERO project? It'd be great to have a GPU-miner from a respected dev like you, as all there is right now is CPU based. Thanks!

Yes. This is simply equihash with a different parameters. But at this moment zcash miner more important. In the future, I will create a miner with these parameters.

Why not make this params init during call..

Like miner.exe --equiinitparam=x,x which would be easy to make

Yes. You can also make different versions of miners to different cards.

[EWBF_]

EWBF_ when will the new release

Now i try "squeeze" all what are possible from current algorithm, unfortunately improvements for one card can reduce speed for another. So when I find a way to maximize the improvement of various cards in the same time, I will publish it, but at this point there is no ETA.

Thanks for the update EWBF_ ; did you get a chance to have a look at the ZERO project? It'd be great to have a GPU-miner from a respected dev like you, as all there is right now is CPU based. Thanks!

Yes. This is simply equihash with a different parameters. But at this moment zcash miner more important. In the future, I will create a miner with these parameters.

Why not make this params init during call..

Like miner.exe --equiinitparam=x,x which would be easy to make

This miner uses "unrolled" memory layout, it increases speed but require more memory. With this memory layout and parameters N 192, K 7 miner will request something around 10Gb of memory. So as you can see this not an option and in this case needed original memory layout that will use around 3Gb of memory.

[french]

Trying to run on EC2 G2 instance. Cuda is installed. Any ideas?
ubuntu@ip:~$ ./miner --server zec-us-west1.nanopool.org --user Lt1d9mqRud5EHou4HXWKVPQBQtLQTXrjzNr.hey1 --pass z --port 6666
+-------------------------------------------------+
|         EWBF's Zcash CUDA miner. 0.3.0b         |
+-------------------------------------------------+
INFO: Server: zec-us-west1.nanopool.org:6666
INFO: Solver Auto.
INFO: Devices: All.
INFO: Temperature limit: 90
INFO: Api: Disabled
---------------------------------------------------
INFO: Target: 0004189300000000...
INFO: Detected new work: 1488291559
ERROR: Cannot run workers.
ubuntu@ip:~$

[EWBF_]

Trying to run on EC2 G2 instance. Cuda is installed. Any ideas?
ubuntu@ip:~$ ./miner --server zec-us-west1.nanopool.org --user Lt1d9mqRud5EHou4HXWKVPQBQtLQTXrjzNr.hey1 --pass z --port 6666
+-------------------------------------------------+
|         EWBF's Zcash CUDA miner. 0.3.0b         |
+-------------------------------------------------+
INFO: Server: zec-us-west1.nanopool.org:6666
INFO: Solver Auto.
INFO: Devices: All.
INFO: Temperature limit: 90
INFO: Api: Disabled
---------------------------------------------------
INFO: Target: 0004189300000000...
INFO: Detected new work: 1488291559
ERROR: Cannot run workers.
ubuntu@ip:~$

Cuda 8?

[french]

Trying to run on EC2 G2 instance. Cuda is installed. Any ideas?
ubuntu@ip:~$ ./miner --server zec-us-west1.nanopool.org --user Lt1d9mqRud5EHou4HXWKVPQBQtLQTXrjzNr.hey1 --pass z --port 6666
+-------------------------------------------------+
|         EWBF's Zcash CUDA miner. 0.3.0b         |
+-------------------------------------------------+
INFO: Server: zec-us-west1.nanopool.org:6666
INFO: Solver Auto.
INFO: Devices: All.
INFO: Temperature limit: 90
INFO: Api: Disabled
---------------------------------------------------
INFO: Target: 0004189300000000...
INFO: Detected new work: 1488291559
ERROR: Cannot run workers.
ubuntu@ip:~$

Cuda 8?

CUDA Version 7.5.18 actually