Topic: [NemosMiner] multi algo profit switching NVIDIA/CPU miner - page 208. (Read 289447 times)

Well, since you did such a good job at de-mystifying zpools inner workings, maybe you should ask tpruvot about connection between algo intensity and CPU speeds.
Anyway, we need more Sherlocks like you on-board.

changing the intensity is a band-aid approach at getting it to run on your computer i think, i actually use this still as i learn to adjust my settings better... but in my research and testing on multiple mining rigs, intensity is really really dependent on your CPU to GPU combo it seems.. because as you increase the intensity level, the CPU is taking larger blocks of data from the pool server you mine from to compute and passing it to the GPU, sometimes the block size overwhelms the GPU causing delays and rejects to occur or hard crash... each algo has a default intensity and different block sizes, thus why one intensity number doesnt work on all algo's....

For me my Ryzen 7 rig can nearly max out all algo intensity levels, but my dell workstation mining rig with 2.0gh xeon cpu has to use much lower intensity levels to not crash the system on the same brand GTX1080ti's... It very well could not be totally up to the CPU and maybe has something to do with the bandwidth size between the 2, but in my testings, it seems to matter the most... and more so the more GPU's you run in the system... so settings that barely ran stable at 1 single GPU, i couldnt for the life of me getting working on 2 or more.. changing the intensity as little as 0.10 would bring the miner back into stability with 2 GPU's on a certain algo

If others can add to this, maybe i can learn more from it all as well... im just going off of things i have witnessed in my testings...

Those custom intensities introduced for Alexis in 1.7.6 did wonders for one of my 1060 cards that was constantly refusing to work with previous version.
What is the default intensity for Alexis anyway?
I read somewhere that its default intensities are optimized for 1070. True that?
Since only this one problematic 1060 needs custom intensities, would it be advisable to continue using default intensities on other 1060s that work fine on 1.7.5. and use 1.7.6 only on this problematic 1060?
I guess running them on default intensities instead on lower ones would improve hashrate?
Or did your benchmarks prove that the new custom intensities do not affect hashrate much, and also provide more stability?
And thanks again for your work and good luck buying that Asus you linked earlier, but at half the price

My experience with Nemosminer and zpool so far, all algo enabled, with two 1070:

Earnings are about 0.004 btc daily, with nicehash calculator two 1070 should earn 0.0044.
But nicehash has 4% fee so the profit is very similar.

Let's see what happens with some REMs

I read on SPccminer thread that  the newest version has hard-coded dev fee.
But I also read much bullshit on that thread, so do not take this for granted.

Hi, Nemo

using your integrated miner for a while, thanks a lot.

Any plan to replace EWBF 0.33b by 0.34b, I believe it's slightly faster.

I dropped lyra2z from the pool.

Hi NEMO

1st of all thanx for the miner most awesome.

I have 1 issue and would like youR opinion,  i am using 2 X 1060 SC and 1 x1080 running 2 bat files for 1.7.5/1.7.6 and after about 2 hours or so i get a system crash, looks like a memory dump.  I can run 1.7.4  with all 3 cards and no problems at all.  Any Ideas?

Can you post the results of the test when it's ended?

Almost all algos other than equihash do better with Mem UV as much as possible.

Currently running lyra2v2, skein, c11, x17 and blake2s exclusively.

Testing to see if cutting down on the multiple switching will help profitability. Seems that certain algos are tougher on clocks as well. These seem to allow me to overclock even more for about 5% improvements on hash so I'm interested in seeing if that compensates for the missed opportunities with the other algos.

Anyone else experimented with this before? I'm using a 7x1070 rig and a 5x 1080 rig.

Also rather than using + values as each manufacturer has different base rates I've been using 1740 base core on 80TDP for 1070's at 127W and 1820 base core on 85 TDP for 1080s at 165W

Understood. (sorry caps locks was stuck on)..

i have it disabled myself just REM it... or remove it from bat completely..

ask crackfoo to remove it from zpool..  

Understood.
But I really think that lyra2z should be completely removed from Nemo and zpool alltogether.

CARDS ARE OC MODEL I MONITOR FREQUENCY THEY RUNNING AROUND 1900-1975MHZ THE MEM OC IS MORE FOR EWBF.. BEEN A GOOD DAY ON EQUIHASH 2DAY...

How come you are using such low core OC on 1060. I use 150 plus 80% PL.
And how come you are overclocking memory? My understanding is that ccminer algos are not memory-intensive, so some people even lower memory clocks when using it.

https://www.computerlounge.co.nz/components/componentview.asp?partid=28830

I had a look at both versions... I see on the zpool one you have dedicated bat files specifically per GPU. I'm just about to build a 5x1080 rig... Any plans on making a 1080 bat file?

Thanks!

1. yes thats how it works

2.yes for my 1060 and 1070 i use the same oc +55mhz core +596mhz mem  1070 are set to 70tdp (125watt) 1060 are @80%tdp (89watt) ...

It's the first time I use Nemos Zpool, at the beginning it writes:

"Terminating workio thread" every time a new algo start

untill it started mining bitcore (more profitable algo I suppose)
Is everything allright?

Second question, I'm using Afterburner to set up clocks but different algo requires different set up,
are you using an average set up to maximize the profit?

1.7.3-mph may need a update 2moro my test rig seems to have a hard time switching from equish algo to other algos..