EWBF's Cuda Equihash Miner 0.6

[exquisitess]

Using SMOS here with latest kernel 4.17

hardware:
motherboard: asrock h110 pro btc (13gpu)
memory : 8gb
cpu g4560
video cards: 12 gtx 1060 palit super jetstream


issue:

gpu gpu 11 and gpu 12 always gets stuck and stops working during gpu detection. Is there a way where you can build the miner to detect all gpus at the very start? similar to how dtsm does it? I noticed getting miners detected one by one i s very slow process and it could be that there is a specific timeout in your program that causes the gpu 11-12 to get stuck in my case.

restarting the miner will cause this issue to be intermittent. sometimes only gpu 12 gets stuck, sometimes both gpu 11 and 12 gets stuck.

please help resolve this. thanks and more power to you ewbf!

[1713904808]

1713904808

[1713904808]

1713904808

[CarlOrff]

@EWBF
Thank you for the v0.3 release and to maintain a Cuda 8 version (Linux).
My hashrate improvement with the 0.3 release is more than 10% :-).
(Just little tested with GTX1050Ti/Windows10/Cuda8 for the moment : will test with 1070, 1080Ti and 960/2GB next week-end).

[daz59]

Using SMOS here with latest kernel 4.17

hardware:
motherboard: asrock h110 pro btc (13gpu)
memory : 8gb
cpu g4560
video cards: 12 gtx 1060 palit super jetstream


issue:

gpu gpu 11 and gpu 12 always gets stuck and stops working during gpu detection. Is there a way where you can build the miner to detect all gpus at the very start? similar to how dtsm does it? I noticed getting miners detected one by one i s very slow process and it could be that there is a specific timeout in your program that causes the gpu 11-12 to get stuck in my case.

restarting the miner will cause this issue to be intermittent. sometimes only gpu 12 gets stuck, sometimes both gpu 11 and 12 gets stuck.

please help resolve this. thanks and more power to you ewbf!

I have this same issue, running 13 cards, the last two get stuck and wont run, sometimes the 3rd from last also gets stuck and it is very very slow to detect all cards, 10+minutes.

[McMCrypto]

Hi guys i am getting 20-22 sol/s   , with my 1060 6GB (x6),  Core +70 / Mem -350 , mining BTG (144_5  --pers BgoldPoW)

That Core/Mem setup  is suitable ?  thanks !

https://i.imgur.com/PYIyipW.png

[shmocs]

guys.. it says 144_5 takes now 1.99 but 1050 2GB cards are still not working

tail: /var/run/miner.output: file truncated
+-------------------------------------------------+
|         EWBF's Eqihash CUDA miner. v0.3         |
+-------------------------------------------------+
02:53:27 INFO: Used config file: /var/run/ethos/ewbf-zcash.conf
02:53:27 INFO: Current pool: eu.voidr.net:3032
02:53:27 INFO: Selected pools: 1
02:53:27 INFO: Algorithm: 144_5 "sngemPoW"
02:53:27 INFO: Devices: User defined.
02:53:27 INFO: Temperature limit: 85
02:53:27 INFO: Api: Listen on 127.0.0.1:42002
---------------------------------------------------
02:53:28 INFO: Target: 00a0000000000000...
02:53:28 INFO: Detected new work: e35d
02:53:29 CUDA Device: 0 GeForce GTX 1050, not enough memory, Available: 2012 MB, Required: 2031 MB
02:53:29 CUDA: Device: 1 GeForce GTX 1050, 1999 MB i:64
02:53:29 CUDA: Device: 2 GeForce GTX 1050, 1999 MB i:64
02:53:30 CUDA: Device: 3 GeForce GTX 1050, 1999 MB i:64
02:53:31 CUDA: Device: 4 GeForce GTX 1050, 1999 MB i:64
02:53:32 CUDA: Device: 1 Thread exited with message: "out of memory"
02:53:33 CUDA: Device: 5 GeForce GTX 1050, 1999 MB i:64
02:53:33 CUDA: Device: 2 Thread exited with message: "out of memory"
02:53:33 CUDA: Device: 3 Thread exited with message: "out of memory"
02:53:33 CUDA: Device: 4 Thread exited with message: "out of memory"
02:53:34 CUDA: Device: 5 Thread exited with message: "out of memory"
02:53:38 INFO: Detected new work: e35e
02:53:40 INFO: Detected new work: e35f
02:53:40 WARNING: Looks like GPU1 are stopped. Restarting...
02:53:40 INFO: GPU1 are restarted.
02:53:40 WARNING: Looks like GPU2 are stopped. Restarting...
02:53:40 INFO: GPU2 are restarted.
02:53:40 WARNING: Looks like GPU3 are stopped. Restarting...
02:53:40 INFO: GPU3 are restarted.
02:53:41 WARNING: Looks like GPU4 are stopped. Restarting...
02:53:41 INFO: GPU4 are restarted.
02:53:41 WARNING: Looks like GPU5 are stopped. Restarting...
02:53:41 INFO: GPU5 are restarted.
02:53:43 CUDA: Device: 1 Thread exited with message: "out of memory"
02:53:43 CUDA: Device: 2 Thread exited with message: "out of memory"
02:53:43 CUDA: Device: 3 Thread exited with message: "out of memory"
02:53:44 CUDA: Device: 4 Thread exited with message: "out of memory"
02:53:44 CUDA: Device: 5 Thread exited with message: "out of memory"

[TAKCuCT]

Still no fixes for rigs more than 12 cards.

[shmocs]

split them up. Or, launch 2 ewbf miners (or more)

[xenonwon]

I have this problem all time:

04.07.2018 13:45:52 CUDA: Device: 1 Thread exited with message: "all CUDA-capable devices are busy or unavailable"
04.07.2018 13:45:52 CUDA: Device: 2 Thread exited with message: "all CUDA-capable devices are busy or unavailable"
04.07.2018 13:45:53 CUDA: Device: 3 Thread exited with message: "all CUDA-capable devices are busy or unavailable"
04.07.2018 13:45:53 CUDA: Device: 4 Thread exited with message: "all CUDA-capable devices are busy or unavailable"
04.07.2018 13:45:53 CUDA: Device: 5 Thread exited with message: "all CUDA-capable devices are busy or unavailable"
04.07.2018 13:45:53 WARNING: Looks like GPU0 are stopped. Restarting...
04.07.2018 13:45:53 INFO: GPU0 are restarted.
04.07.2018 13:45:53 WARNING: Looks like GPU1 are stopped. Restarting...
04.07.2018 13:45:53 INFO: GPU1 are restarted.
04.07.2018 13:45:54 WARNING: Looks like GPU2 are stopped. Restarting...
04.07.2018 13:45:54 INFO: GPU2 are restarted.
04.07.2018 13:45:54 WARNING: Looks like GPU3 are stopped. Restarting...
04.07.2018 13:45:54 INFO: GPU3 are restarted.
04.07.2018 13:45:54 WARNING: Looks like GPU4 are stopped. Restarting...
04.07.2018 13:45:54 INFO: GPU4 are restarted.
04.07.2018 13:45:54 CUDA: Device: 0 Thread exited with message: "all CUDA-capable devices are busy or unavailable"
04.07.2018 13:45:54 WARNING: Looks like GPU5 are stopped. Restarting...
04.07.2018 13:45:54 INFO: GPU5 are restarted.

I have 6x 1070 ti, the OC is ok i think, 65 TDP, 180 core, 550 memory, windows 10.
Thanks for helps.

Not enough info: 
pool?
size of virtual memory?
v0.2 or v.03 miner?
bat file settings?

suprnova.cc
20000 virtual
v0.3
miner  --algo 144_5 --pers BgoldPoW --server btg.suprnova.cc --port 8866 --user user.user --pass x --pec --log 2
pause

I have tried to remove all the OC and it works stable for more than 15 hours, so I guess it was too much OC, but I do not think I had much OC.

[gsanjin]

suprnova.cc
20000 virtual
v0.3
miner  --algo 144_5 --pers BgoldPoW --server btg.suprnova.cc --port 8866 --user user.user --pass x --pec --log 2
pause

I have tried to remove all the OC and it works stable for more than 15 hours, so I guess it was too much OC, but I do not think I had much OC.

Equihash 144,5 is demanding much more GPU memory than "old" equihash so mining it taxes more your gpu's and what was stable before, now maybe isn't. Maybe your mem was overclocked a little to high for "new" algo, so I would try with lower mem overclocks on your place.

[xenonwon]

suprnova.cc
20000 virtual
v0.3
miner  --algo 144_5 --pers BgoldPoW --server btg.suprnova.cc --port 8866 --user user.user --pass x --pec --log 2
pause

I have tried to remove all the OC and it works stable for more than 15 hours, so I guess it was too much OC, but I do not think I had much OC.

Equihash 144,5 is demanding much more GPU memory than "old" equihash so mining it taxes more your gpu's and what was stable before, now maybe isn't. Maybe your mem was overclocked a little to high for "new" algo, so I would try with lower mem overclocks on your place.

Now i'm tried 65 PL, 190 core, 400 memory. Thanks.

[TAKCuCT]

split them up. Or, launch 2 ewbf miners (or more)

HiveOS does not allow 2 same miners.

[EWBF_]

so in the image with red font. u can see difference in power between version .2 and .3  .. .3 is a tad bit better. but in all my pc's with different mobos. different video cards different PSU's, all have now jagged lines. why is this happening.... im not gonna use another miner. but does anyone have a fix to smooth these teeth out?

can we get pics of others power lines on these algo's cause my whole entire lot, just now has this issue. and i know it's either a tweak i need to do, or miner just needs time to mature ? any ideas.

again im working on BTG new fork 144_5

I'm also concerned about erratic power and GPU load with new EWBF on new equihash (BTG 144,5).
I tried both v0.2 and 0.3 and GTX 1070 at different clocks (including stocks) - the same result; as seen on the picture (right panel on the picture):


Older EWBF (0.3.4b) with older equihash is fine with the same cards (left panel on the picture).

EWBF, please comment

Why do you think this is a problem?
This is because the solver is slow and the workload is not homogenous, probably over time this will be slightly improved.

Using SMOS here with latest kernel 4.17

hardware:
motherboard: asrock h110 pro btc (13gpu)
memory : 8gb
cpu g4560
video cards: 12 gtx 1060 palit super jetstream


issue:

gpu gpu 11 and gpu 12 always gets stuck and stops working during gpu detection. Is there a way where you can build the miner to detect all gpus at the very start? similar to how dtsm does it? I noticed getting miners detected one by one i s very slow process and it could be that there is a specific timeout in your program that causes the gpu 11-12 to get stuck in my case.

restarting the miner will cause this issue to be intermittent. sometimes only gpu 12 gets stuck, sometimes both gpu 11 and 12 gets stuck.

please help resolve this. thanks and more power to you ewbf!

Ok, i will check this out.

Hello @EWBF_

Spec:
GPU 0: Gigabyte Windforce 2 NVIDIA GeForce GTX 1060 3GB (3072 MB)
- GPU is together with other video card what i use for my own purpose
OS: Windows 7 x64 Service Pack 1
Video Driver: 397.31

INFO: Algorithm: 144_5
CUDA: Device: 0 GeForce GTX 1060 3GB, 3072 MB i:64
ERROR: out of memory

GPU-Z Report it is 103 Mb ram use so to a simple math 3072 - 103 = 2969 MB there are left
According to https://forum.bitcoingold.org/t/our-new-equihash-equihash-btg/1512

- Quote: Equihash-BTG: "<144,5> parameters require a minimum of 700 MB to run and use about 2.5 GB to run efficiently"
So 2,5G or 2,9G 1060 with 3G Vram it suppose to work but according to EWBF it is not have inaff memory with is ?!?!?! 

cheers

This can be the size of virtual memory, check your paging file.

[ermis1975]

I use this miner with 3 1070 cards. Everything was ok but today I notice that pool rejects all shares. I use miningpoolhub.

[CarlOrff]

guys.. it says 144_5 takes now 1.99 but 1050 2GB cards are still not working

Your screen was connected to one of the GPUs or to the motherboard (iGPU) ?
Someone else did some testing with 2 GB NVidia GPUs and release 0.3 of EWBF miner ?

[Overdrive5]

split them up. Or, launch 2 ewbf miners (or more)

HiveOS does not allow 2 same miners.

If this helps, I can give instructions to run  2 instances of EWBF on ZhashOS.

https://bitcointalk.org/index.php?topic=4481713.msg40261859#msg40261859

[Overdrive5]

ZhashOS Release 2.01 is now available.

-Updated to EWBF 0.3 for up to 10% faster hashrate depending on the card.

-Revised the 1streadme1st document to give further guidance

-added FAQ.PDf to instruct on setting up Power/Fan Speed/Overclock settings on GPU. Very Important!

-Made password change function (passwd) operational. All users setting up Remote Desktop are highly recomended to set thier own password for the rig.

-Shrunk the download even further. Now ~475MB.

-MD5: 8124074e6e0f321a10f309bea89cf9b9 ZhashOS_r201.img.zip

https://mega.nz/#!nq5WGSLY!XzbAkTm9k1MXEsccSk0xW_SPv9dLg_ro6eDTJKO6KLk

[shmocs]

guys.. it says 144_5 takes now 1.99 but 1050 2GB cards are still not working

Your screen was connected to one of the GPUs or to the motherboard (iGPU) ?
Someone else did some testing with 2 GB NVidia GPUs and release 0.3 of EWBF miner ?

@CarlOrff - No screen. Worked smooth 24/7 for 6 months already.

@ewbf - is there a reason why this small difference makes the difference ? (Available: 2012 MB, Required: 2031 MB). I hoped that this is the reason of the change, to make 2GB cards available again.

[Overdrive5]

split them up. Or, launch 2 ewbf miners (or more)

HiveOS does not allow 2 same miners.

If this helps, I can give instructions to run  2 instances of EWBF on ZhashOS.

https://bitcointalk.org/index.php?topic=4481713.msg40261859#msg40261859

Hive OS does allow 2 instances. Just open another terminal window and run a separate instance. Have been doing all along.

Out of curiosity, and to help TAKuCT, what do you have to type into the terminal to load the 2nd instance with the separate 2nd set of cards?

[ComputerGenie]

Out of curiosity, and to help TAKuCT, what do you have to type into the terminal to load the 2nd instance with the separate 2nd set of cards?

run the 1st with like

Code:

--cuda_devices 0 1 2 3 4

and the 2nd with like

Code:

--cuda_devices 5 6 7 8

[ripcurrent]

I have a 1080ti rig that wont run the miner

all CUDA-capable devices are busy or unavailable error.

I have tried down clocking and increasing VM to 40000 still no go.


Any other suggestions...