Post
Topic
Board Mining (Altcoins)
Re: [ANN] dstm's ZCash Nvidia Miner v0.5 (Linux)
by
kryztoval
on 03/10/2017, 10:07:10 UTC
Quote
Are your systems on wifi? If yes, you could try using a cable if possible.
You might test running zm on less GPUs using the '--dev' option so you'll have less connections per system.
                                                                                                                     
Systems are wired.

Quote
Not sure what you mean by 'build up of connections'. It opens one socket per GPU there should be no 'build up of connections'.
Imagine a memory leak, but made with open sockets. That is what this feels like. could be or not, I am just trying to explain it somehow. Like, if you have a disconect, do you close the socket or does garbage collection close it? do you reuse the sockets? what if the sockets are left alive and for some reason my installation doesn't do garbage collection until network crashes? again, this is pure guesswork. this is a shot in the dark and may not even be possible. Just saying what it "feels" like.

Quote
I have a 10GPU system running stable, no disconnects, on flypool. I'm pretty sure they have no issues handling much more connections. If we run out of ideas I'll write a test for this Smiley
Seems weird. I imaged the system as ubuntu 17.04. I can get the image online if you want to try it or I can get you access to one of the rigs too if this proves too troublesome.

Quote
Complete output till it disconnects (pls include also the 'connection closed by server' message - if you get it).
I'll make it with -dev, give me a couple of hours to let it run and I'll send you those.

Quote
Edit:
I'm pretty sure there should be no issues with the number of connections on the pool side. So let's say you have 1000 rigs mining the same ZEC address, the pool must be able to handle all 1000 connections mining the same ZEC address.
Yeah, I am pretty sure this is something that should not happen, this is weird.