New Miner Needing Help with HD7870 settings on NSGminer?

[RIPPEDDRAGON]

Have you tried different drivers or reinstalling your current ones? Have you tried with a fresh nsgminer and config? Have you checked GPU memory usage while mining? There are so many things is could be, you are going to have to try a bunch of stuff to find out whats going wrong. Welcome to being a miner!

I have seen so many crazy bugs over the last 3+ years of building and tearing rigs down, once you get beyond simple config issues I find it best to nuke it and rebuild from scratch.

[greenuser]

@RIPPEDDRAGON Thank you, I will have a play tommorrow, I am looking at it hash away now mining ethereum. It works perfect for days, totaly unattended.
AMD Catalyst 15.12 drivers are the best for mining eth, Later drivers yeald lower hasrates. I will try a driver update tommorrow while nursing my hangover. :smirk:

These are hard faults to find. It will hash away all night on feathercoin and then just stop and display “wait”. It has mined 127 ftc over the last two days but it’s no good to me if i have to reset it every few hours.

As i say, i used to have an Antminer U3 like this. It would work for days and then just go “ZOMBIE”. I guess “WAIT” in NSGminer is the same as “ZOMBIE” cgminer!
I was so glad to see the back of those ASICs

[ghostlander]

@greenuser --gpu-engine and --gpu-memclock require arguments. For example, --gpu-engine 1000 --gpu-memclock 1500 set the GPU core to 1000MHz and GPU memory to 1500MHz (6000MHz effective).

Don’t set --gpu-vddc to zero. It’s no good.

WAIT happens when the miner waits for work supplied by the pool. SICK, DEAD and HW errors are signs of GPU issues.

[greenuser]

@ghostlander Thanks, will try tommorow and post results. What value should i set --gpu-vddc ? or should i remove that line from the conf

[ghostlander]

@greenuser You better remove it to keep the defaults.

[greenuser]

@ghostlander i asume it is the default value. All i did was put this line in a bat file

nsgminer --neoscrypt - -I 13 -o http://p2pool.neoscrypt.de:19327 -O 6wTKzQovBDnMrhxtcR44KyLz3xWAPx15Ap

I then ran the bat file and then once running hit [S]ettings then [W]rite to make a config file.
The above post “My current nsgminer.conf file for 4 off HD7870 looks like this…”
is the un-edited conf. I just changed fan speeds to make all cards run at the same temp (cards in the middle of the rack run hotter than the outside ones)
The --gpu-vddc value is what is set as default.

[ghostlander]

@greenuser -I 13 or --intensity 13 are both correct. Note the number of hyphens.

[greenuser]

yes for sure. Once a basic bat file has been made and exicuted and… you have created a conf file by hitting [S]ettings then [W]rite while running, you no longer need the bat file. You can get rid of it.

You can then just run the nsgminer.exe to start mining and arguments will be read from the conf. This way i get to see all the defaults and make changes as needed.

Tommorrow i will edit the “nsgminer.conf” file with the GPU core 1000MHz and GPU memory 1500MHz arguments and post results.

Thank you for the arguments values :pray:

[greenuser]

I will edit the conf like this

{
“pools” : [
{
“url” : “http://p2pool.neoscrypt.de:19327”,
“user” : “6wTKzQovBDnMrhxtcR44KyLz3xWAPx15Ap”,
“pass” : “”,
“pool-priority” : “0”
}
]
,
“temp-cutoff” : “90,90,90,90”,
“temp-target” : “75,75,75,75”,
“intensity” : “13,13,13,13”,
“vectors” : “1,1,1,1”,
“worksize” : “256,256,256,256”,
“kernel” : “neoscrypt,neoscrypt,neoscrypt,neoscrypt”,
“lookup-gap” : “0,0,0,0”,
“thread-concurrency” : “0,0,0,0”,
“shaders” : “0,0,0,0”,
“gpu-engine” : “0-1000,0-1000,0-1000,0-1000”,
“gpu-fan” : “0-55,0-78,0-100,0-100”,
“gpu-memclock” : “1500,1500,1500,1500”,
“gpu-memdiff” : “0,0,0,0”,
“gpu-powertune” : “0,0,0,0”,
“gpu-vddc” : “0.000,0.000,0.000,0.000”,
“temp-overheat” : “85,85,85,85”,
“api-port” : “4028”,
“expiry” : “120”,
“expiry-lp” : “3600”,
“gpu-dyninterval” : “7”,
“gpu-platform” : “0”,
“gpu-threads” : “2”,
“log” : “5”,
“no-pool-disable” : true,
“queue” : “1”,
“scan-time” : “60”,
“neoscrypt” : true,
“skip-security-checks” : “0”,
“submit-stale” : true,
“temp-hysteresis” : “3”,
“shares” : “0”,
“kernel-path” : “/usr/local/bin”
}
0

[greenuser]

Ok this rig has run fine mining Ethereum for the last 4 days
Today i tried to mine ftc

i tryed these bat files commands

nsgminer --neoscrypt -I 13 -o http://p2pool.neoscrypt.de:19327 -O 6wTKzQovBDnMrhxtcR44KyLz3xWAPx15Ap

Then:

nsgminer --neoscrypt -I 12 -o http://p2pool.neoscrypt.de:19327 -O 6wTKzQovBDnMrhxtcR44KyLz3xWAPx15Ap

Then:

nsgminer --neoscrypt –-gpu-engine 1000 --gpu-memclock 1500 -I 13 -o http://p2pool.neoscrypt.de:19327 -O 6wTKzQovBDnMrhxtcR44KyLz3xWAPx15Ap

Then:

nsgminer --neoscrypt –-gpu-engine 1000 --gpu-memclock 1500 -I 13 -o http://p2pool.neoscrypt.de:19327 -O 6wTKzQovBDnMrhxtcR44KyLz3xWAPx15Ap

The command window opend on each time and i saw some writing but it went blue screen with white vertical stripes before i could read anything. Had to reboot to get rid of the stripes

So then i tried sgminer 5 with

sgminer -o http://p2pool.neoscrypt.de:19327 -O 6gTcFMDZeJRL4tgJbnHQDjXZe8Gm7kKaT8

and got the same blue screen with white vertical stripes, again needed to boot to get out.

Currently mining Ethereum again.

[greenuser]

Ok its working!
I reinstalled all GPU drivers and noted clock and core values in GPU-Z while mining Ethereum
Core 1000Mhz
Clock 1200Mhz

nsgminer --neoscrypt –-gpu-engine 1000 --gpu-memclock 1200 -I 13 -o http://p2pool.neoscrypt.de:19327 -O 6wTKzQovBDnMrhxtcR44KyLz3xWAPx15Ap

[Wellenreiter]

Are you using this address: 6gTcFMDZeJRL4tgJbnHQDjXZe8Gm7kKaT8?

After about 3 hours of mining our stale rate went up and is at about 84% at the moment.
Looks like you set your GPU parameters a bit to aggressive

[wrapper]

@greenuser Did you try nsgminer with no setting? I’d be interested to see what Engine settings NSGminer sets on it’s own. It poles the GPU for it what it thinks its settings are.

Also, I found one of my cards would not run stable over Engine 980 it might be worth trying 990. You can check by optimizing the engine speed to reduce the number of rejects.

Sometimes a lower engine can give the highest hash, as its a complex relation between memory and speed which is different for every hashing algorithm. Also, the highest hash is not necessary the most energy efficient.

[greenuser]

I get the same results with no settings, when i reduce values, hashpower goes down.

But now that i am hashing, why do i get 50%+ stale hashrate?
shutting down as i am using power but not making coin.

[greenuser]

@Wellenreiter said:

Are you using this address: 6gTcFMDZeJRL4tgJbnHQDjXZe8Gm7kKaT8?

After about 3 hours of mining our stale rate went up and is at about 84% at the moment.
Looks like you set your GPU parameters a bit to aggressive

@Wellenreiter Sorry i only just saw your post. Yes my address, I will reduce -I

Others here seem to have Stale hashrate 20%+ too. 6f1q2U6TuR8FijX96na9dHDko4HYAu3XPr

[greenuser]

@Wellenreiter

Current conf file

{
“pools” : [
{
“url” : “http://p2pool.neoscrypt.de:19327”,
“user” : “6gTcFMDZeJRL4tgJbnHQDjXZe8Gm7kKaT8”,
“pass” : “”,
“pool-priority” : “0”
}
]
,
“temp-cutoff” : “95,95,95,95”,
“temp-target” : “75,75,75,75”,
“intensity” : “11,11,11,11”,
“vectors” : “1,1,1,1”,
“worksize” : “256,256,256,256”,
“kernel” : “neoscrypt,neoscrypt,neoscrypt,neoscrypt”,
“lookup-gap” : “0,0,0,0”,
“thread-concurrency” : “0,0,0,0”,
“shaders” : “0,0,0,0”,
“gpu-engine” : “970,970,970,970”,
“gpu-fan” : “0-85,0-85,0-90,0-85”,
“gpu-memclock” : “1300,1300,1300,1300”,
“gpu-memdiff” : “0,0,0,0”,
“gpu-powertune” : “0,0,0,0”,
“gpu-vddc” : “0.000,0.000,0.000,0.000”,
“temp-overheat” : “85,85,85,85”,
“api-port” : “4028”,
“expiry” : “120”,
“expiry-lp” : “3600”,
“gpu-dyninterval” : “7”,
“gpu-platform” : “0”,
“gpu-threads” : “2”,
“log” : “5”,
“no-pool-disable” : true,
“queue” : “1”,
“scan-time” : “60”,
“neoscrypt” : true,
“skip-security-checks” : “0”,
“submit-stale” : true,
“temp-hysteresis” : “3”,
“shares” : “0”,
“kernel-path” : “/usr/local/bin”
}

So i turned up Mem clock to 1300 and turned down Core clock to 970.
Mem Clock shows correct values in GPU-Z but Core clock (gpu-engine) still reads 1000Mhz although set at 970Mhz, not sure why.
Ints set down at 11

I will watch for Stale Shares

[wrapper]

@greenuser said:

I get the same results with no settings, when i reduce values, hashpower goes down.

But now that i am hashing, why do i get 50%+ stale hashrate?
shutting down as i am using power but not making coin.

The fact that other miners are suffering, means it is with the server or communications to it. There is trouble with connections the further you are away, for instance.

Not having a good pool connection : it’s one good thing about running a local p2pool node, it acts as a buffer, collecting your shares then passing them round the network.

Also, it would be interesting to save a ping in a 5 min chron script and do a chart of connection variability from the output.

[greenuser]

So i need to find a closer pool to me here in Cornwall UK?

[wrapper]

@greenuser stales can be due to connections

I say you need to run a p2pool node like me, it’s pretty easy and low power. ie there are few in UK on about cornwall

How much are the cpus doing now on your miner?

[greenuser]

@wrapper Low on hard drive space, how big are the files?

Relatively small, it clears over a day as stratum is reached and block go by. I think it is python, I’ll see if I can see