Pages:
Author

Topic: [ANN] Miner Control 1.6.1 - Auto profit switching miner controller - page 16. (Read 164300 times)

full member
Activity: 202
Merit: 100
Since this software is being actively improved, is time to start a new thread something like "Miner Control - Giraudy mod". Giraudy deserves the recognition for the good work.
full member
Activity: 224
Merit: 100
CryptoLearner

Well that's the problem, you can't use minercontrol with additionnal scripts. hence why i asked to have such capabilities directly into the tool Smiley (at least until ccminer is fixed, and even then you have other miner that doesn't have any overclock features)

As far as I know CCminer supports now launching with special config file, as it was previously for AMD miners. Here are the samples
https://bitcointalk.org/index.php?topic=769239.msg14509384;topicseen#msg14509384 You can try to play with it.
Nevertheless in my opinion overclocking is a very tricky thing, just imagine that you have a "zoo" of cards on different Nvidia chips. How much miner instances have to be started to meet the best parameters for each hardware type? You have to parametrise a bundle of different miners for a wide range of devices. That is the thing, that developers of NicehashMiner are trying to implement for a long time. I can't remember any new version of NHM without a serious bugs.

Yeah that's why i posted this

https://bitcointalksearch.org/topic/m.17884458

this way you could setup per algo/per card/per rig setting easy without relying on miner overlock but external one.

When i meant script i was speaking about external bat files for example to overlock, not ccminer conf file Smiley
full member
Activity: 133
Merit: 100

Well that's the problem, you can't use minercontrol with additionnal scripts. hence why i asked to have such capabilities directly into the tool Smiley (at least until ccminer is fixed, and even then you have other miner that doesn't have any overclock features)

As far as I know CCminer supports now launching with special config file, as it was previously for AMD miners. Here are the samples
https://bitcointalk.org/index.php?topic=769239.msg14509384;topicseen#msg14509384 You can try to play with it.
Nevertheless in my opinion overclocking is a very tricky thing, just imagine that you have a "zoo" of cards on different Nvidia chips. How much miner instances have to be started to meet the best parameters for each hardware type? You have to parametrise a bundle of different miners for a wide range of devices. That is the thing, that developers of NicehashMiner are trying to implement for a long time. I can't remember any new version of NHM without a serious bugs.
full member
Activity: 133
Merit: 100
aaand seems like miningfields api doesn't work=(

Fixed. Pls, download .exe file again. As for WTM active parameter, it should be in this way
Code:
,"WTM-DMD-Groestl": {"account": "acc","worker": "1070", "pricemode": 0, "cweight": 1.0, "fee":0,"sparam1": "","sparam2": "",
        "algos": [{"active":false, "cname": "diamond", "tag":"dmd", "algo": "dmd-gr", "priceid": "43", "folder": "_APARAM1_", "command": "_APARAM2_", "arguments": "_APARAM3_ stratum+tcp://eu.miningfield.com:3378  -u _ACCOUNT_._WORKER_  _SPARAM2_ -p 123" }]
eeem... WTM-Myriad-Groestl for dmd-gr?right?
It is just a title (copied it from MYR-GR), can be named as you wish, but should start with WTM anyway. WTM unblocked all IPs.
sr. member
Activity: 455
Merit: 250
aaand seems like miningfields api doesn't work=(

Fixed. Pls, download .exe file again. As for WTM active parameter, it should be in this way
Code:
,"WTM-Myriad-Groestl": {"account": "acc","worker": "1070", "pricemode": 0, "cweight": 1.0, "fee":0,"sparam1": "","sparam2": "",
        "algos": [{"active":false, "cname": "diamond", "tag":"dmd", "algo": "dmd-gr", "priceid": "43", "folder": "_APARAM1_", "command": "_APARAM2_", "arguments": "_APARAM3_ stratum+tcp://eu.miningfield.com:3378  -u _ACCOUNT_._WORKER_  _SPARAM2_ -p 123" }]
eeem... WTM-Myriad-Groestl for dmd-gr?right?
full member
Activity: 224
Merit: 100
CryptoLearner
If overclocking isn't directly supported by 1.8.4, I am assuming it may not be directly supported by all of the other miners too.

How do I make this custom script to make the clocks change then?

Also, to make sure I understand, for:

COIN : ZCASH ZCL PASC FTC VTC SIB DGB GRS MONA XZC KMD
CORE+MEM / CORE +150 / MEM +595 / TDP 60%

I would do Core 2150 and mem 4599? (I just added the + values to the standard values you said of 2000 and 4004)

Yup exactly, depending of your TDP, of course this OC will not be reached. but this is the numbers you wish to reach.

to make scripts i use the cmdline tool nvidia-inspector, that allow you to make scripted overclock.

like this

Code:
nvidiaInspector.exe -setBaseClockOffset:0,0,150 -setMemoryClockOffset:0,0,595 -setPowerTarget:0,60 -setTempTarget:0,0,85

Where do I put that code though and how do I make miner control switch to different overclock profiles for different coins?

Well that's the problem, you can't use minercontrol with additionnal scripts. hence why i asked to have such capabilities directly into the tool Smiley (at least until ccminer is fixed, and even then you have other miner that doesn't have any overclock features)
member
Activity: 108
Merit: 10
If overclocking isn't directly supported by 1.8.4, I am assuming it may not be directly supported by all of the other miners too.

How do I make this custom script to make the clocks change then?

Also, to make sure I understand, for:

COIN : ZCASH ZCL PASC FTC VTC SIB DGB GRS MONA XZC KMD
CORE+MEM / CORE +150 / MEM +595 / TDP 60%

I would do Core 2150 and mem 4599? (I just added the + values to the standard values you said of 2000 and 4004)

Yup exactly, depending of your TDP, of course this OC will not be reached. but this is the numbers you wish to reach.

to make scripts i use the cmdline tool nvidia-inspector, that allow you to make scripted overclock.

like this

Code:
nvidiaInspector.exe -setBaseClockOffset:0,0,150 -setMemoryClockOffset:0,0,595 -setPowerTarget:0,60 -setTempTarget:0,0,85

Where do I put that code though and how do I make miner control switch to different overclock profiles for different coins?
full member
Activity: 224
Merit: 100
CryptoLearner
If overclocking isn't directly supported by 1.8.4, I am assuming it may not be directly supported by all of the other miners too.

How do I make this custom script to make the clocks change then?

Also, to make sure I understand, for:

COIN : ZCASH ZCL PASC FTC VTC SIB DGB GRS MONA XZC KMD
CORE+MEM / CORE +150 / MEM +595 / TDP 60%

I would do Core 2150 and mem 4599? (I just added the + values to the standard values you said of 2000 and 4004)

Yup exactly, depending of your TDP, of course this OC will not be reached. but this is the numbers you wish to reach.

to make scripts i use the cmdline tool nvidia-inspector, that allow you to make scripted overclock.

like this

Code:
nvidiaInspector.exe -setBaseClockOffset:0,0,150 -setMemoryClockOffset:0,0,595 -setPowerTarget:0,60 -setTempTarget:0,0,85
sr. member
Activity: 455
Merit: 250
hmmm...suprnova is working now...for how long...
member
Activity: 108
Merit: 10
If overclocking isn't directly supported by 1.8.4, I am assuming it may not be directly supported by all of the other miners too.

How do I make this custom script to make the clocks change then?

Also, to make sure I understand, for:

COIN : ZCASH ZCL PASC FTC VTC SIB DGB GRS MONA XZC KMD
CORE+MEM / CORE +150 / MEM +595 / TDP 60%

I would do Core 2150 and mem 4599? (I just added the + values to the standard values you said of 2000 and 4004)
full member
Activity: 224
Merit: 100
CryptoLearner
Can I somehow just add that to the code to automatically switch to those settings when miner control switches to a different algorithm?

For the moment existed solution. Check the miner possible OC keys. Here is for ccminer 1.8.4

Code:
*** ccminer 1.8.4 for nVidia GPUs by tpruvot@github ***
    Built with VC++ 2013 and nVidia CUDA SDK 8.0

      --mem-clock=3505  Set the gpu memory max clock (346.72+ driver)
      --gpu-clock=1150  Set the gpu engine max clock (346.72+ driver)
      --pstate=0[,2]    Set the gpu power state (352.21+ driver)
      --plimit=100W     Set the gpu power limit (352.21+ driver)

Add wished values in "algorithms" section to "aparam3", for example

{ "name": "lbry", "display": "Lbry", "hashrate": 300000, "mu": "kH/s","power": 50, "aparam1": "Miners\\Nvidia\\Ccminer\\Tpruvot", "aparam2": "ccminer_tpruvot.exe", "aparam3": "-a lbry --mem-clock=3505 --gpu-clock=1150  -o" }

Check the result in applications console.



Problem is overlock isn't possible in 1.8.4, there is a bug with the NVLM that Epsylon fixed in 2.0 but it's not  released yet, hence why i had to rely on external scripts.

Does anyone have any recommended settings for the GTX 1070 with this miner? Should I reduce the power limit %? Should I increase the mem and core clock? Any suggestions on which settings to maximize profits would be appreciated

Depend on the algos, every of them have different settings. It's to you to test out each algos to find the good stuff. From my experience there is 3 kinds of algos, the one needed GPU core, the one needing Memory, and the one needing both.

with a 1070 i have 3 settings but of course you'll have to find your sweet spot because it differ from your rig, card model and such.

here is what i use. Coin are market with their symbols, you can find those on Whattomine.

COIN : ZCASH ZCL PASC FTC VTC SIB DGB GRS MONA XZC KMD
CORE+MEM / CORE +150 / MEM +595 / TDP 60% (about 110-112w in nvidia-smi, you would have to see what's the sweet spot for you, each model have their own TDP limit)

COIN : LBC DCR DGB-SKEIN DMD
++CORE -MEM / CORE +185 / MEM -500 / TDP 60%

COIN : ETH ETC EXP
--CORE ++MEM / CORE -400 / MEM +795 / TDP 60%

Are those + and - values based off the original core/mem clock or the default boosted core/mem clock? I have EVGA 08G-P4-5173-KR

Based on the normal clock of the card. (2000/4004)
member
Activity: 108
Merit: 10
Does anyone have any recommended settings for the GTX 1070 with this miner? Should I reduce the power limit %? Should I increase the mem and core clock? Any suggestions on which settings to maximize profits would be appreciated

Depend on the algos, every of them have different settings. It's to you to test out each algos to find the good stuff. From my experience there is 3 kinds of algos, the one needed GPU core, the one needing Memory, and the one needing both.

with a 1070 i have 3 settings but of course you'll have to find your sweet spot because it differ from your rig, card model and such.

here is what i use. Coin are market with their symbols, you can find those on Whattomine.

COIN : ZCASH ZCL PASC FTC VTC SIB DGB GRS MONA XZC KMD
CORE+MEM / CORE +150 / MEM +595 / TDP 60% (about 110-112w in nvidia-smi, you would have to see what's the sweet spot for you, each model have their own TDP limit)

COIN : LBC DCR DGB-SKEIN DMD
++CORE -MEM / CORE +185 / MEM -500 / TDP 60%

COIN : ETH ETC EXP
--CORE ++MEM / CORE -400 / MEM +795 / TDP 60%

Are those + and - values based off the original core/mem clock or the default boosted core/mem clock? I have EVGA 08G-P4-5173-KR
sr. member
Activity: 455
Merit: 250
mkay...need to switch back to yiimp i guess for lbry and sib..gbye mona...=(
sr. member
Activity: 455
Merit: 250
Here is response from WTM  Wink

Quote

Hey,

I have blocked some IPs yesterday due to many errors spawned by accessing `/coins/.json` path.
I guess you tried to code access for a coin that is not added to WTM.
Please use valid URLs.

WTM

I've tested another solution to reduce amount of requests. This is a buffer array for WTM data. For the moment each pool requests necessary data from WTM independently.
With array it will get all info from there, and array itself will be syncronised with WTM in a certain period of time. It will be the first step to deal with the situation.

but it would be much... slower i guess.if we get this array once in a hour...then we can't switch with difficulty drops.we can only switch with average profitability(1 hour in this example)
newbie
Activity: 12
Merit: 0
That JSON requests are evil. All theese sites are so afraid of ddos and stuff... And it would be much more beneficial for all automation soft to use data subscriptions like FIX whatever possible.
full member
Activity: 133
Merit: 100
Here is response from WTM  Wink

Quote

Hey,

I have blocked some IPs yesterday due to many errors spawned by accessing `/coins/.json` path.
I guess you tried to code access for a coin that is not added to WTM.
Please use valid URLs.

WTM

I've tested another solution to reduce amount of requests. This is a buffer array for WTM data. For the moment each pool requests necessary data from WTM independently.
With array it will get all info from there, and array itself will be syncronised with WTM in a certain period of time. It will be the first step to deal with the situation.
sr. member
Activity: 455
Merit: 250
sooo...any way to fix that for masses?=)
newbie
Activity: 12
Merit: 0
Seen theese blank responses today and belive that number of connections per minute is the case. 1 main worker + 1 control computer = data outage.
sr. member
Activity: 455
Merit: 250
so, this is suprnova or what?why i can see all the data, but can't see time on suprnova?
my brain is blowing=)


Tryied from another IP address, everything is working fine. I think, it is WTM blacklist for IPs, too many API requests from the same IP. It is a kind of DDOS protection. Still waiting an answer from them.

yes, that was a problem with original MC, if you have more rigs, say more than 5, they pull pools' API like every 10 seconds, so you get blank response most of the time.

is it possible to make one interface that pulls API info and all MCs to connect to it via local network.

i have 1 rig only...
hero member
Activity: 698
Merit: 500
so, this is suprnova or what?why i can see all the data, but can't see time on suprnova?
my brain is blowing=)


Tryied from another IP address, everything is working fine. I think, it is WTM blacklist for IPs, too many API requests from the same IP. It is a kind of DDOS protection. Still waiting an answer from them.

yes, that was a problem with original MC, if you have more rigs, say more than 5, they pull pools' API like every 10 seconds, so you get blank response most of the time.

is it possible to make one interface that pulls API info and all MCs to connect to it via local network.
---------------
there were some questions how to make MC jump on every spike, set those:
        "mintime": 1,
        "switchtime": 0,
        "delay": 1,
        "exittime": 0,
        "minprice": "0BTC",
        "minebyaverage": false,
        "dynamicswitching": true,
        "dynamicswitchpower": 2,
        "ignoreoutliers": false,
   "dynamicswitchpivot":1.05,
   "dynamicswitchoffset":0,
        "iqrmultiplier": 2.2,

set "minprofit": 1 to nicehash as it is pay per share, others are estimate, so give them at least "minprofit": 1,01
Pages:
Jump to: