XMR-stak-JK 2.10.7 Compiled with no devfee (Updated: 2019-8-18) - page 15.

Shnikes101

full member

Activity: 254

Merit: 109

Quote from: RIGED on April 11, 2018, 11:29:25 AM

can you please post the code and placement of the command to stop cpu mining

I find it in the cpu.txt file but cant figure out which * to remove and where to place null (with "null" or without) comma , at the end or not
lol its all very confusing trying to understand the script backwards Shocked

He did so on post #5. You could also remove everything between the [ ] or use null.

Quote from: JaredKaragen on April 06, 2018, 05:49:17 PM

or another example if you dont want to mine on your CPU (cpu.txt):

This:

Code:

"cpu_threads_conf" :
[
    { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 0 },
    { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 2 },

],

Becomes this:

Code:

"cpu_threads_conf" :
[
   /* { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 0 },
    { "low_power_mode" : false, "no_prefetch" : true, "affine_to_cpu" : 2 },
  */
  //  or you can comment out a line one at a time like this line
],

Not sure why you closed the last thread. Wouldn't call my post useless spam.
As you mentioned, the instructions to compile and change the DevFee are clearly outlined on github.
There is always a risk in downloading unknown files. The safest option if you want to change the DevFee is to compile the program yourself - I'd encourage people to at least try.
Jared seems like a nice enough guy and I'm sure would help users out if they have issues but again, safest option is to compile yourself. Trust the code - beware of people.

RIGED

member

Activity: 182

Merit: 12

can you please post the code and placement of the command to stop cpu mining

I find it in the cpu.txt file but cant figure out which * to remove and where to place null (with "null" or without) comma , at the end or not
lol its all very confusing trying to understand the script backwards Shocked

rs1x

full member

Activity: 284

Merit: 102

nice work @JaredKaragen

JaredKaragen

legendary

Activity: 1848

Merit: 1166

My AR-15 ID's itself as a toaster. Want breakfast?

Updated original post with a download for alternate AMD and nvidia DLL's. They have given better performance on some older AMD cards, so if you had low results, might wanna give them a try.

JaredKaragen

legendary

Activity: 1848

Merit: 1166

My AR-15 ID's itself as a toaster. Want breakfast?

Quote from: cpryd on April 10, 2018, 11:32:38 AM

Why not read the manual:

Code:

"gpu_thread_num" : 2,
"gpu_threads_conf" : [
  // gpu: gfx901 memory:3920
  // compute units: 56
  { "index" : 0, "intensity" : 1932, "worksize" : 8, "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2, "comp_mode" : false
  },
  { "index" : 0, "intensity" : 1932, "worksize" : 8, "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2, "comp_mode" : false
  },

],

That looks like it should work at first glance; but if I were to fix it, I would be adding it to be automatically assigned multiple threads when it senses a vega... assuming I can understand the code to that point to make such a change.

A similar issue arises with i3-6100/i3-7100 CPU's. Its a 2 core 4 thread, but the app only detects and sets it up as a 2 core 2 thread for some reason. my i7-920, and many other CPU's detect and assign all threads fine. not sure what causes this. *edit* normal behavior as this is HyperThreading taking effect.

cpryd

newbie

Activity: 1

Merit: 0

Why not read the manual:

Code:

"gpu_thread_num" : 2,
"gpu_threads_conf" : [
  // gpu: gfx901 memory:3920
  // compute units: 56
  { "index" : 0, "intensity" : 1932, "worksize" : 8, "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2, "comp_mode" : false
  },
  { "index" : 0, "intensity" : 1932, "worksize" : 8, "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2, "comp_mode" : false
  },

],

JaredKaragen

legendary

Activity: 1848

Merit: 1166

My AR-15 ID's itself as a toaster. Want breakfast?

Quote from: RaidoP2 on April 09, 2018, 07:49:20 AM

Thanks for compiling this! Has anyone tried with R9 380 or R9 380x cards? What are the best settings?

Not sure about AMD, I only have an RX560 on hand for testing. Outside input is greatly appreciated, but since it seems to support the old HD cards, so I think the other R series cards should be supported.

Ill have to go over the code... I am still figuring out how the dev wrote the app; but I think ive figured out how the containers work and the different functions of the app work; If I can understand how to create a second thread for vegas; than im sure hashrate would possibly bump up. Don't get your hopes up on this; it is pretty complicated and im trying to better understand the code....

My grand idea in the end is to be able to incorporate all of the open source miner apps into one. Not expecting this to happen, but it is a goal.

RaidoP2

newbie

Activity: 61

Merit: 0

Thanks for compiling this! Has anyone tried with R9 380 or R9 380x cards? What are the best settings?

RIGED

member

Activity: 182

Merit: 12

Quote from: JaredKaragen on April 09, 2018, 12:32:54 AM

Quote from: RIGED on April 09, 2018, 12:07:06 AM

thnx,
the vega can run two simultaneous threads reaching 2000H/s but need a double string input in the amd.txt and for the life of me I cant get the coding correct (spacing? a comma? idunno) I'm sure I will get right by the end of the day by trial and error
I will come back and post it here

When you do; generate a fresh amd.txt by deleting it and running xmr-stak again. Post it here with the #code posting ability

Code:

like this

Ill have a peek and give some input =)

Ill be able to put the Rx560 I have back on xmr-stak off my other tests to generate a amd.txt and see what i see on my end as well. Im pretty sure the vega driver supports the version of openCL I compiled it with.

got it work with max intensity set at 1012
but only one thread

Code:

"gpu_threads_conf" : [
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 0,
    "intensity" : 1012, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 1,
    "intensity" : 1012, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 2,
    "intensity" : 1012, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 3,
    "intensity" : 1012, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 4,
    "intensity" : 1012, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 5,
    "intensity" : 1012, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },

],

/*
 * Platform index. This will be 0 unless you have different OpenCL platform - eg. AMD and Intel.
 */
"platform_index" : 1,

giving about the same hash results as the best I can get it with a double thread per card (yea it was the intensities that were giving the buffer error message)

Code:

"gpu_threads_conf" : [
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 0,
    "intensity" : 1000, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  { "index" : 0,
    "intensity" : 800, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 1,
    "intensity" : 1000, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  { "index" : 1,
    "intensity" : 800, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 2,
    "intensity" : 1000, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  { "index" : 2,
    "intensity" : 800, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 3,
    "intensity" : 1000, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  { "index" : 3,
    "intensity" : 800, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 4,
    "intensity" : 1000, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  { "index" : 4,
    "intensity" : 800, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  // gpu: gfx901 memory:3920
  // compute units: 64
  { "index" : 5,
    "intensity" : 1000, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },
  { "index" : 5,
    "intensity" : 800, "worksize" : 8,
    "affine_to_cpu" : false, "strided_index" : 1, "mem_chunk" : 2,
    "comp_mode" : true
  },

],

/*
 * Platform index. This will be 0 unless you have different OpenCL platform - eg. AMD and Intel.
 */
"platform_index" : 1,

still playing around with the intensities on the dual thread amd.txt file but it seems the new heavy algo is very demanding on the GPUs as i cant get anywhere near the previous algo 2000H/s

good to be on Cryptonight again

many thanx JK!!

Xapodat

newbie

Activity: 35

Merit: 0

This miner does not work so well on old cards like HD 7970. I only get about 400H/s. Anybody has similar result?

sicklyslick

brand new

Activity: 0

Merit: 0

[cuda] error gpu 0 something something :365

1080ti on windows 10 64 bit. 21gb virtual memory

JaredKaragen

legendary

Activity: 1848

Merit: 1166

My AR-15 ID's itself as a toaster. Want breakfast?

Quote from: sicklyslick on April 09, 2018, 12:38:30 AM

[cuda] error gpu 0 something something :365

1080ti on windows 10 64 bit. 21gb virtual memory

edit nvidia.txt

see notes at beginning of file about adjusting bfactor and bsleep. probably happening cause you have a monitor plugged into GPU #0.

JaredKaragen

legendary

Activity: 1848

Merit: 1166

My AR-15 ID's itself as a toaster. Want breakfast?

Quote from: RIGED on April 09, 2018, 12:07:06 AM

thnx,
the vega can run two simultaneous threads reaching 2000H/s but need a double string input in the amd.txt and for the life of me I cant get the coding correct (spacing? a comma? idunno) I'm sure I will get right by the end of the day by trial and error
I will come back and post it here

When you do; generate a fresh amd.txt by deleting it and running xmr-stak again. Post it here with the #code posting ability

Code:

like this

Ill have a peek and give some input =)

Ill be able to put the Rx560 I have back on xmr-stak off my other tests to generate a amd.txt and see what i see on my end as well. Im pretty sure the vega driver supports the version of openCL I compiled it with.

RIGED

member

Activity: 182

Merit: 12

Quote from: JaredKaragen on April 08, 2018, 09:39:56 PM

Quote from: RIGED on April 08, 2018, 09:34:56 PM

thnx!!

I couldn't get it to run two threads on the vega64
could you please post the code for 6 gpus if you know

thnx again for the effort

Not sure what you mean.

If you have multiple GPU, you will see a config file like this for nvidia:

You can see each GPU listed with it's configs. There are notes that I did not cut and paste about modifying 'bfactor' and 'bsleep' to 8 and 100 respectively. This has helped me in my issues in the past. the AMD file should contain a similar description of its settings. I am thinking this is probably your issue? You should see all system GPU's listed in the AMD.txt file.

I don't have any vegas on hand to test with or anything, so I can't get a config from a vega to show a prime example as the app creates the configs itself... so give it a shot; assuming I understood your initial question properly Wink

Also; if you add more GPU's to the system, or remove some, or change the CPU, delete the appropriate file (amd.txt, cpu.txt or nvidia.txt) and re-run xmr-stak.exe. it will re-assess the system and rebuild the configs for the current hardware set.

thnx,
the vega can run two simultaneous threads reaching 2000H/s but need a double string input in the amd.txt and for the life of me I cant get the coding correct (spacing? a comma? idunno) I'm sure I will get right by the end of the day by trial and error
I will come back and post it here

minahan24

newbie

Activity: 23

Merit: 0

Quote from: JaredKaragen on April 08, 2018, 10:12:52 PM

Quote from: minahan24 on April 08, 2018, 09:56:56 PM

OP, what settings do I need to tweak to increase my speed. I have 6x rx570 and I've run SRBMiner and getting 5700h/s, while I'm only getting 4000h/s with this miner. Maybe you can point me to the settings I need to play around.

I believe it has something to do with the threads and blocks variables; the .txt file should have an explanation of each one and what it does.

does SRBMiner apply an overclock to the cards when it launches? that may be the difference in hashrate.

Im not sure is xmr-stak is the quickest miner app overall TBH... but, I do know so far it has a wide range of compatibility with devices compared to most other apps i've tried.

The SRBMiner doesnt apply any overclock, I have to manually apply it myself. But I'm using the same overclock I've used on the SRBMiner on this miner but getting lower speed so I think I need to play around with the settings. I will update.

JaredKaragen

legendary

Activity: 1848

Merit: 1166

My AR-15 ID's itself as a toaster. Want breakfast?

Quote from: minahan24 on April 08, 2018, 09:56:56 PM

OP, what settings do I need to tweak to increase my speed. I have 6x rx570 and I've run SRBMiner and getting 5700h/s, while I'm only getting 4000h/s with this miner. Maybe you can point me to the settings I need to play around.

I believe it has something to do with the threads and blocks variables; the .txt file should have an explanation of each one and what it does.

does SRBMiner apply an overclock to the cards when it launches? that may be the difference in hashrate.

Im not sure is xmr-stak is the quickest miner app overall TBH... but, I do know so far it has a wide range of compatibility with devices compared to most other apps i've tried.

minahan24

newbie

Activity: 23

Merit: 0

OP, what settings do I need to tweak to increase my speed. I have 6x rx570 and I've run SRBMiner and getting 5700h/s, while I'm only getting 4000h/s with this miner. Maybe you can point me to the settings I need to play around.

JaredKaragen

legendary

Activity: 1848

Merit: 1166

My AR-15 ID's itself as a toaster. Want breakfast?

Quote from: RIGED on April 08, 2018, 09:34:56 PM

thnx!!

I couldn't get it to run two threads on the vega64
could you please post the code for 6 gpus if you know

thnx again for the effort

Not sure what you mean.

If you have multiple GPU, you will see a config file like this for nvidia:

Code:

"gpu_threads_conf" :
[
  // gpu: GeForce GTX 1070 architecture: 61
  //      memory: 6789/8192 MiB
  //      smx: 15
  { "index" : 0,
    "threads" : 28, "blocks" : 45,
    "bfactor" : 8, "bsleep" :  100,
    "affine_to_cpu" : false, "sync_mode" : 3,
  },
  // gpu: GeForce GTX 980 architecture: 52
  //      memory: 3370/4096 MiB
  //      smx: 16
  { "index" : 1,
    "threads" : 12, "blocks" : 48,
    "bfactor" : 6, "bsleep" :  25,
    "affine_to_cpu" : false, "sync_mode" : 3,
  },

],

You can see each GPU listed with it's configs. There are notes that I did not cut and paste about modifying 'bfactor' and 'bsleep' to 8 and 100 respectively. This has helped me in my issues in the past. the AMD file should contain a similar description of its settings. I am thinking this is probably your issue? You should see all system GPU's listed in the AMD.txt file.

I don't have any vegas on hand to test with or anything, so I can't get a config from a vega to show a prime example as the app creates the configs itself... so give it a shot; assuming I understood your initial question properly Wink

Also; if you add more GPU's to the system, or remove some, or change the CPU, delete the appropriate file (amd.txt, cpu.txt or nvidia.txt) and re-run xmr-stak.exe. it will re-assess the system and rebuild the configs for the current hardware set.

RIGED

member

Activity: 182

Merit: 12

thnx!!

I couldn't get it to run two threads on the vega64
could you please post the code for 6 gpus if you know

thnx again for the effort

JaredKaragen

legendary

Activity: 1848

Merit: 1166

My AR-15 ID's itself as a toaster. Want breakfast?

Quote from: Sgsg666 on April 07, 2018, 12:01:58 PM

Yay! Thanks this is awesome!
Other cryptonight miners you might want to configure:

https://bitcointalksearch.org/topic/cryptonight-miner-designed-for-amd-built-in-overdriven-and-hw-monitoring-3058014
https://github.com/xmrig/xmrig
https://github.com/Bendr0id/xmrigCC

is there a reason to recompile more than one XMR app? I mean; I just wanted to provide a legitimate outlet to a no devfee miner for my miner batch/packages I am assembling. I liked XMR-stak because it incorporated Nvidia as well as AMD cards; as well as supporting all my pre-CUDA7 devices. (470, 550, 550ti, 650, 650ti, etc etc.)

Something I am mainly on the look out for is: an all-around good version of SGMiner that handles everything from X11->X16R and everything in between for AMD cards. I don't have my hands on a stable and reliable "AMD 'other algo' miner" as of yet. Same for equihash and ETH miners. ETH is bottom of my list as it is its own animal; and not a coin/platform I ever believed in. I only use ETH on exchanges for selling otherwise unsellable coins at a profit. If I can not make a profit on DOGE markets; usually I can squeeze a profit out of going from ALT->ETH->BTC. Since ETH doesn't scale as small as DOGE, its easier to pinch out a modest profit.

I already have the nvidia side almost handled; I just want to find a few more no fee sources like for equihash and possibly one for ETH.

Quote from: Documan on April 07, 2018, 02:32:49 AM

I wonder if it works with RX 550 cards. The claymore miner does not work with that.

I've already tested with a rx560, so I don't see why it wouldn't work; the algo isn't super dependent on memory size AFAIK.

Side note: With this new no-devfee recompiled version; it has been maintaining a more steady hashrate report at the pool. fewer wild swings, etc. as well as a better monthly income as per nanopool's estimate. I think it is due in part to the miner app disconnecting; and reconnecting to the pool and getting auto-diff reset when a block change happens during devfee mining. The first few lower diff shares tend to get a fake hashrate bump etc. I see it now. Devfee mining really hurt the miners profits in the end.... Maybe you guys will see it eventually.

I wish people treated mining apps as people treated software in the early 90's on BBS's...... We would be a better society as a whole.

FYI: sorry for the damaged first upload. I had no clue until something was mentioned. If anyone had downloaded when I first uploaded; I suggest replacing it with the current posted version at the top of this thread.

Topic: XMR-stak-JK 2.10.7 Compiled with no devfee (Updated: 2019-8-18) - page 15. (Read 28650 times)