Topic: ### A ChainWorks Industries (CWI) Project - CWIgm | Simple Powerful Stable - page 40. (Read 67732 times)

Trying a single 1070 gpu (+125core/+500mem/60%tdp) with denarius (tribus) on win7.

Immediately noticed a new command in the bat that doesn't exist in the bat for sigt  --mean-net-diff=1.1

Miner reports a speed of 53.  I have never mined this algo before, so nothing to personally compare it to, but i did check yiimp benchs- 53 is pretty damn good

After a few minutes i noticed something weird.  In the right bottom box, a yellow line saying net diff > mean net diff.  The netdiff field in the miner turned red, and the miner just stopped mining.  I could see the card idling in afterburner.  This went on for a few blocks, then the netdiff field turned green, and the miner resumed.   Rinse/repeat every 5-10 minutes or so, with each break lasting about 2-3 blocks.

I guessed that the --mean-net-diff=1.1 was causing this, took it out, and this behavior stopped.  Miner just kept on mining.

I'm not sure what --mean-net-diff=1.1 is meant to do, and it's probably there for a good reason. For me, however, it's pausing the miner and makes the gpu just sit there idling for a bit.  An idling mining gpu is a sin

update after 10h uptime, speed reduced to 323Mh (initially was 325Mh), rig working stable, no reboot, no crashes

same story here - my one GPU rig goes offline after failing due to diff.

my other 5 GPU rig, would go offline after 8 or so but will reconnect again.

have yet to run 24 hours stable.

agree. great hashing rate but without a share after mining for about 7+ hours

ok, some quick observations after running for about 8 hours or so.

Short version - miner reports great hashrate, better then any other i tried so far.  However, the pool likes to jack up the worker diff way too high, choking out smaller rigs.

Long version:

Trying on a total of 10x1070's spread across one main rig, and some random boxes.

box 1 - win 7, single gpu. (worker 1)
box 2 - win 10, single gpu. (worker 4)
box 3 - win 7 2xgpu (worker 2)
box 4 - win 10 6xgpu. (worker 3)

All gpu's are tuned with exactly the same settings - +125 core, +500 mem, 60%tdp (110w) in afterburner. These were my go to settings for zcash and lbry, so i left them alone for now  All get just over 29mh on skunk, according to the miner.

However, and this is especially true for the single and 2x gpu machines, the pool difficulty ends up being too high.  The single gpu's started out at pool diff of 64 (worker at 0.25), which from my previous experimenting is the sweet spot for a single 1070 gpu, and then eventually ended up at diff of 256 (worker at 1).  At 256 the card chokes.  When the nethash is nice and steady, the card is still able to submit work ever 2-3 sometimes 4 blocks.  When whatever is hitting this coin hits it hard (when you see the nethash double in like 5 minutes) the card can't keep up with fast moving blocks at all.  There was a time when i caught a glimpse at the miner and saw at least a good dozen blocks go by before the card was able to submit any work.  There were instance that the card was so chocked, the pool would show that a worker dropped out, meaning that the card submitted no shares in whatever time interval the pool uses to calculate hash.    In fact, as i'm typing this i can see the miner window on my single gpu boxes, and one of them just had about 10 blocks go by without any work submitted (diff 256 pool, 1 at worker).

It's a similar story for the 2xgpu box.  The diff started off at 128, which again i think is a sweetspot for this setup, but then climbed to 512.  At that diff, just like the single gpu setup, too many blocks go by before a share is submitted and the cards choke.

6xgpu rig seems to be not as affected by this.  It started off at 256, then eventually climbed to 640.  It too misses out on blocks, and hashrate reported at the pool jumps around wildly, but at least it's able to get some shares in every other block or so.

Overall, i am very happy with the miner.  The pool is finding blocks at a good clip, and with good efficiency, so maybe this approach of jacking up the worker diff works.  But it feels wrong to see this many bunches of blocks go by without a share submitted.  To me it means that my gpu's are not even competing to find a solution to a whole lot of blocks.  You can't solve a block if you don't at least compete to solve it.

As i was finishing typing this, another 6 blocks went by before my single gpu rig submitted a share.  That means that this particular gpu didn't even compete for 6/7th of the opportunities to find a block.

Its been almost 4 hours without the pool hitting a block. The pool should hit the block about every 70-80 minutes with the current hashrate. The last two hours before the last block the pool hit 5 blocks which was great, so its normal to be little dry now. For the last 24 hours it hit 25 blocks which was great because half of that time the pool was with almost twice less hashrate. I hope we will start hitting blocks again very soon. If it goes more than 6 -7 hours without a block we can speak about a problem but i doubt it will get to that.

Does anyone know which android app can be used to monitor pool with the QR code provided?

Not code, just strings. And I'm not claiming it as mine It was clearly stated it's a cwigm and all who want an original version were encouraged to participate here.
Yes, everyone who uses 3rd party versions should first accept the risks.

DNR mining:

1080 Ti Aorus 97-102 MH
1080 Aorus 70-75 MH

Core +100 Memory -500 TDP = 0.7

But mining of denarius is not profitable today. Speed need to be at least 20% more MH from cards like in "SP tribus MOD"

for first: sorry for my english.

gpu's: 6x GTX Titan Xp
drivers: 385.12 beta

default bat for sigt

first start was with PL 100%, core +150, mem -500 :  driver was crashed in few seconds

second start was with PL 100%, core stock, mem stock : rig was crashed in about 4 minutes

hmm... like as ewbf in past

ok, PL 87%, core stock, mem stock : rig stable about 10 min

next, PL87%, core +150, mem stock: rig stable

intensity as auto is 21

add -i 25 to bat and run - rig stable already 30 min

http://roenko.ru/img/miner.png

what is missing: remote monitoring, like as EWBF 0.3.4b.
 

hi all ...

seems like i am online again - and will catch up on the posts ...


#crysx

ok, some quick observations after running for about 8 hours or so.

Short version - miner reports great hashrate, better then any other i tried so far.  However, the pool likes to jack up the worker diff way too high, choking out smaller rigs