EWBF's Cuda Equihash Miner 0.6

[Overdrive5]

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

So just to be clear here, They would have to type out the miner command in the terminal each time they want to load a second instance for the separate 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

So just to be clear here, They would have to type out the miner command in the terminal each time they want to load a second instance for the separate set of cards?

yes, that param allows for the ability to run each card on a different pool, algo, miner, etc (i.e., you could pick card 6 to run alone on PoolY with Account1 and you could pick card 7 to run alone on PoolZ with Account2)
in this instance, you'd run 1/2 of the cards on 1 setup and the other 1/2 on a 2nd setup (even though both setups are the same except card selection)

[sir_blacks]

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.

Hello EWBF_
Spec:
Real Ram: 8g
GPU 0: Gigabyte Windforce 2 NVIDIA GeForce GTX 1060 3GB (3072 MB) = Full Dedicate to Mining OR Gaming, obvious not make bouth in same time
GPU 1: Gigabyte GeForce GTX 650 2GB (2048 MB) = used for my own necessitate, browsing, working, sometime gaming
Virtual Memory / Page: 12228 min default with max 16384 that is more that inaff for 2 video card & 8 g real ram, i was increase much more, for testing reason, not mater same thing ...
min 12228 max 18432 = ERROR: out of memory = not page is issue
min 12228 max 20480 = ERROR: out of memory = not page is issue
OS: Windows 7 x64 Service Pack 1
Video Driver: 397.31

With this setting i was play game on 650 & mine on 1060 ETH, Neoscript, old Equihash with out problem only now i can't use your miner because it say from start "ERROR: out of memory" with is something wrong

Code:

+-------------------------------------------------+
|         EWBF's Eqihash CUDA miner. v0.3         |
+-------------------------------------------------+
20:15:04 INFO: Current pool: btg.suprnova.cc:8866
20:15:04 INFO: Selected pools: 1
20:15:04 INFO: Algorithm: 144_5 "BgoldPoW"
20:15:04 INFO: Devices: User defined.
20:15:04 INFO: Temperature limit: 90
20:15:04 INFO: Api: Disabled
---------------------------------------------------
20:15:04 INFO: Target: 00f0f0f0f0f0f0f0...
20:15:04 INFO: Detected new work: 5b7
20:15:04 CUDA: Device: 0 GeForce GTX 1060 3GB, 3072 MB i:64
20:15:04 ERROR: out of memory

miner --cuda_devices 0 --algo 144_5 --pers BgoldPoW --pec --server btg.suprnova.cc --port 8866 --user xxx --pass x

[lolchina]

Am i the only one so far that dont have any kind of improvment with 0.3 version speed wise,on 144.5 algo my 1080tis are doing 55-60 sols wich is the same speed that they had on previous version...Am i missing out something or boost is for 1060/70 cards?

[gsanjin]

...
GPU 1: Gigabyte GeForce GTX 650 2GB (2048 MB) = used for my own necessitate, browsing, working, sometime gaming
...
With this setting i was play game on 650 & mine on 1060 ETH, Neoscript, old Equihash with out problem only now i can't use your miner because it say from start "ERROR: out of memory" with is something wrong

Bolded is your problem, 3GB cards are new required minimum. Your only option is to start miner only on GPU0 (3GB one), look few post above how to do that (in your case use parameter --cuda_devices 0 for GPU0).

[SouthWEBA]

Hello @EWBF_

hardware:

motherboard: asrock h110 pro btc (13gpu)
memory : 4gb
cpu Celeron 3930
UNIX - Fedora

Version 0.3.4b working super to ZEC

video cards: 13 gtx EVGA SC2 1080Ti

issue:

gpu 9 gpu 10 gpu 11 and gpu 12 always gets stuck and stops working during gpu detection.
May be build the miner to detect all gpus at the very start?

I tested to start 2 miner

cuda_devices 0 1 2 3 4 5 6 and 7 8 9 10 11 12

and is very very very slow process detected card :-( and crashed EWBF

Please help me ASAP i have +- 3000 card

13x1070 card working normal

[Overdrive5]

...
GPU 1: Gigabyte GeForce GTX 650 2GB (2048 MB) = used for my own necessitate, browsing, working, sometime gaming
...
With this setting i was play game on 650 & mine on 1060 ETH, Neoscript, old Equihash with out problem only now i can't use your miner because it say from start "ERROR: out of memory" with is something wrong

Bolded is your problem, 3GB cards are new required minimum. Your only option is to start miner only on GPU0 (3GB one), look few post above how to do that (in your case use parameter --cuda_devices 0 for GPU0).

And to add more info to this, EWBF requires compute 5 or higher.

Unfortunately, a GTX 650 is only compute 3.0 . Maybe use the 650 to mine something else?

[Overdrive5]

Hello @EWBF_

hardware:

motherboard: asrock h110 pro btc (13gpu)
memory : 4gb
cpu Celeron 3930

video cards: 13 gtx EVGA SC2 1080Ti

issue:

gpu 9 gpu 10 gpu 11 and gpu 12 always gets stuck and stops working during gpu detection.
May be build the miner to detect all gpus at the very start?

I started 2 miner

cuda_devices 0 1 2 3 4 5 6 and 7 8 9 10 11 12

and is very very very slow process detected card :-( and crashed EWBF

Please help me ASAP i have +- 3000 card

13x1070 card working normal

SouthWEBA, Which O/S?

[sir_blacks]

...
GPU 1: Gigabyte GeForce GTX 650 2GB (2048 MB) = used for my own necessitate, browsing, working, sometime gaming
...
With this setting i was play game on 650 & mine on 1060 ETH, Neoscript, old Equihash with out problem only now i can't use your miner because it say from start "ERROR: out of memory" with is something wrong

Bolded is your problem, 3GB cards are new required minimum. Your only option is to start miner only on GPU0 (3GB one), look few post above how to do that (in your case use parameter --cuda_devices 0 for GPU0).

gsanjin if you was read i say GP0 is use only for mining aka always in any mining application i enable only GP0 aka in EWBF case i have "--cuda_devices 0" used, so what you was say is not issue, thx anyway

@Overdrive5 you to was not read all what i write 650 GPU = "used for my own necessitate, browsing, working, sometime gaming" is NOT use for mining ... only 1060 read full post what done on previos page aka here https://bitcointalk.org/index.php?topic=4466962.msg41579162#msg41579162

[SouthWEBA]

UNIX - Fedora

Version 0.3.4b working super to ZEC

Hello @EWBF_

hardware:

motherboard: asrock h110 pro btc (13gpu)
memory : 4gb
cpu Celeron 3930

video cards: 13 gtx EVGA SC2 1080Ti

issue:

gpu 9 gpu 10 gpu 11 and gpu 12 always gets stuck and stops working during gpu detection.
May be build the miner to detect all gpus at the very start?

I started 2 miner

cuda_devices 0 1 2 3 4 5 6 and 7 8 9 10 11 12

and is very very very slow process detected card :-( and crashed EWBF

Please help me ASAP i have +- 3000 card

13x1070 card working normal

SouthWEBA, Which O/S?

[Overdrive5]

UNIX - Fedora

Version 0.3.4b working super to ZEC

Hello @EWBF_

hardware:

motherboard: asrock h110 pro btc (13gpu)
memory : 4gb
cpu Celeron 3930

video cards: 13 gtx EVGA SC2 1080Ti

issue:

gpu 9 gpu 10 gpu 11 and gpu 12 always gets stuck and stops working during gpu detection.
May be build the miner to detect all gpus at the very start?

I started 2 miner

cuda_devices 0 1 2 3 4 5 6 and 7 8 9 10 11 12

and is very very very slow process detected card :-( and crashed EWBF

Please help me ASAP i have +- 3000 card

13x1070 card working normal

SouthWEBA, Which O/S?

How much swap are you running?

[baraday]

Am i the only one so far that dont have any kind of improvment with 0.3 version speed wise,on 144.5 algo my 1080tis are doing 55-60 sols wich is the same speed that they had on previous version...Am i missing out something or boost is for 1060/70 cards?

I also have 0 improvement on my 1080ti, so perhaps with the next version 

[Coin++]

Am i the only one so far that dont have any kind of improvment with 0.3 version speed wise,on 144.5 algo my 1080tis are doing 55-60 sols wich is the same speed that they had on previous version...Am i missing out something or boost is for 1060/70 cards?

For me : 1050Ti, and 15% boost with v0.3 :-)

[EWBF_]

Hello @EWBF_

hardware:

motherboard: asrock h110 pro btc (13gpu)
memory : 4gb
cpu Celeron 3930
UNIX - Fedora

Version 0.3.4b working super to ZEC

video cards: 13 gtx EVGA SC2 1080Ti

issue:

gpu 9 gpu 10 gpu 11 and gpu 12 always gets stuck and stops working during gpu detection.
May be build the miner to detect all gpus at the very start?

I tested to start 2 miner

cuda_devices 0 1 2 3 4 5 6 and 7 8 9 10 11 12

and is very very very slow process detected card :-( and crashed EWBF

Please help me ASAP i have +- 3000 card

13x1070 card working normal

Hello! check this out: https://mega.nz/#F!uhQR0IRa!jZ_tdULPoUf3ib4O6-1BoQ this is a special version, basically the same v0.3, but with a modified device launch system. Please note that if you do not have a problem with starting up devices, do not use this version!

[gsanjin]

...
GPU 1: Gigabyte GeForce GTX 650 2GB (2048 MB) = used for my own necessitate, browsing, working, sometime gaming
...
With this setting i was play game on 650 & mine on 1060 ETH, Neoscript, old Equihash with out problem only now i can't use your miner because it say from start "ERROR: out of memory" with is something wrong

Bolded is your problem, 3GB cards are new required minimum. Your only option is to start miner only on GPU0 (3GB one), look few post above how to do that (in your case use parameter --cuda_devices 0 for GPU0).

gsanjin if you was read i say GP0 is use only for mining aka always in any mining application i enable only GP0 aka in EWBF case i have "--cuda_devices 0" used, so what you was say is not issue, thx anyway

@Overdrive5 you to was not read all what i write 650 GPU = "used for my own necessitate, browsing, working, sometime gaming" is NOT use for mining ... only 1060 read full post what done on previos page aka here https://bitcointalk.org/index.php?topic=4466962.msg41579162#msg41579162

Yes, on second read You're using v0.3? There were problems with 3GB on first versions of miner on Win 10 because of way Win 10 addresses GPU memory, but I see you use Win 7. Have you tried using --cuda_devices 1 maybe gpu ordering is wrong? Any difference if you try mining with just one card in system?

[Hurtman]

EWBF_
is there any news about the normal support of Maxwell? (GTX 970/980)

[SouthWEBA]

Hello @EWBF_

hardware:

motherboard: asrock h110 pro btc (13gpu)
memory : 4gb
cpu Celeron 3930
UNIX - Fedora

Version 0.3.4b working super to ZEC

video cards: 13 gtx EVGA SC2 1080Ti

issue:

gpu 9 gpu 10 gpu 11 and gpu 12 always gets stuck and stops working during gpu detection.
May be build the miner to detect all gpus at the very start?

I tested to start 2 miner

cuda_devices 0 1 2 3 4 5 6 and 7 8 9 10 11 12

and is very very very slow process detected card :-( and crashed EWBF

Please help me ASAP i have +- 3000 card

13x1070 card working normal

Hello! check this out: https://mega.nz/#F!uhQR0IRa!jZ_tdULPoUf3ib4O6-1BoQ this is a special version, basically the same v0.3, but with a modified device launch system. Please note that if you do not have a problem with starting up devices, do not use this version!

This version not working :-(

I use CUDA 8

06.07.2018 00:19:09 ERROR: CUDA driver version is insufficient for CUDA runtime version 

[sir_blacks]

Hello! check this out: https://mega.nz/#F!uhQR0IRa!jZ_tdULPoUf3ib4O6-1BoQ this is a special version, basically the same v0.3, but with a modified device launch system. Please note that if you do not have a problem with starting up devices, do not use this version!

@EWBF_ hmm after i take this new test version miner it start running what it make me think in other version when it use --cuda_devices 0 & there are 2,3 gpu in reality miner it was not deselect other one coz check this:

--cuda_devices 1 (2 GPU) - v0.3 normal version

Code:

+-------------------------------------------------+
|         EWBF's Eqihash CUDA miner. v0.3         |
+-------------------------------------------------+
23:55:57 INFO: Current pool: btg.suprnova.cc:8866
23:55:57 INFO: Selected pools: 1
23:55:57 INFO: Algorithm: 144_5 "BgoldPoW"
23:55:57 INFO: Devices: User defined.
23:55:57 INFO: Temperature limit: 90
23:55:57 INFO: Api: Disabled
---------------------------------------------------
23:55:57 INFO: Target: 00f0f0f0f0f0f0f0...
23:55:57 INFO: Detected new work: 62f
23:55:58 CUDA: Device: 0 GeForce GTX 1060 3GB Not selected.
23:55:58 ERROR: out of memory

--cuda_devices 0 (2 GPU) - v0.3 normal version

Code:

+-------------------------------------------------+
|         EWBF's Eqihash CUDA miner. v0.3         |
+-------------------------------------------------+
20:15:04 INFO: Current pool: btg.suprnova.cc:8866
20:15:04 INFO: Selected pools: 1
20:15:04 INFO: Algorithm: 144_5 "BgoldPoW"
20:15:04 INFO: Devices: User defined.
20:15:04 INFO: Temperature limit: 90
20:15:04 INFO: Api: Disabled
---------------------------------------------------
20:15:04 INFO: Target: 00f0f0f0f0f0f0f0...
20:15:04 INFO: Detected new work: 5b7
20:15:04 CUDA: Device: 0 GeForce GTX 1060 3GB, 3072 MB i:64
20:15:04 ERROR: out of memory

--cuda_devices 0 (2 GPU) - new test version functional one

Code:

+-------------------------------------------------+
|         EWBF's Eqihash CUDA miner. v0.3         |
+-------------------------------------------------+
00:11:37 INFO: Current pool: btg.suprnova.cc:8866
00:11:37 INFO: Selected pools: 1
00:11:37 INFO: Algorithm: 144_5 "BgoldPoW"
00:11:37 INFO: Devices: User defined.
00:11:37 INFO: Temperature limit: 90
00:11:37 INFO: Api: Disabled
---------------------------------------------------
00:11:38 INFO: Target: 00f0f0f0f0f0f0f0...
00:11:38 INFO: Detected new work: 636
00:11:38 Device detection:
00:11:38 CUDA: Device: 0 GeForce GTX 1060 3GB, 3072 MB i:64
00:11:38 CUDA: Device: 1 GeForce GTX 650 Not selected.
00:11:38 Running device threads:
00:11:38 CUDA: Device: GeForce GTX 1060 3GB 0000:01:00.0 started...

Power Limit: 74%, Core Clock: +164, Memory: +500 (Samsung Vram)

Code:

00:20:46 Temp: GPU0: 63C
00:20:46 GPU0: 24 Sol/s
00:20:46 Total speed: 24 Sol/s
00:20:46 -----------+--------------+
| GPU | Power usage |  Efficiency  |
+-----+-------------+--------------+
|  0  |     82W     |  0.29 Sol/W  |
+-----+-------------+--------------+

so conclusion new test version it fix using device correctly (cuda 9.1) ....

[cirlama]

for me, mining zero with gtx 1080 ti, efficiency is better with high intensity (--intensity 90 --algo 192_7) and reduced power (like 65 to 70%), 33 sol/s (+400 mem, +80 core, 65% power)

similar results for MNX coin (--intensity 90 --algo 96_5)

with base intensity, I need more power in Watt to achieve the same hashrate

[gsanjin]

so conclusion new test version it fix using device correctly (cuda 9.1) ....

Glad you found out solution, after all it was bug in device selection code.