Status as of Thursday, September 7, 2017

Discussion of development releases of Prohashing.
tabak6
Posts: 11
Joined: Wed Jun 28, 2017 1:53 am

Re: Status as of Thursday, September 7, 2017

Post by tabak6 » Thu Sep 07, 2017 10:29 am

Tell me how to set the range h from x1 to x2, or h is less than a certain value?
User avatar
Steve Sokolowski
Posts: 3907
Joined: Wed Aug 27, 2014 3:27 pm
Location: State College, PA

Re: Status as of Thursday, September 7, 2017

Post by Steve Sokolowski » Thu Sep 07, 2017 12:53 pm

tabak6 wrote:This problem is already 8 days, can we hope that you will solve it? Can I register a new account?
The stale shares problem is nothing new; it just seems worse because we fixed the problems with shares being lost, and now this is the next capacity bottleneck.

I did some investigation and determined that the only way we're going to be able to solve this is to parallelize coin assignment. I started on that already, but it will probably take about 20 days or so. This is the holy grail of performance improvements and will probably increase capacity from around 3.5 TH/s to over 10. In the meantime, Chris will try a few workarounds to see if he can turn off coins to reduce server load, and registrations will remain closed for the time being.
hashingpro
Posts: 207
Joined: Fri Aug 18, 2017 1:14 am

Re: Status as of Thursday, September 7, 2017

Post by hashingpro » Thu Sep 07, 2017 4:32 pm

So I noticed something today while running a test, not sure if it matters or not.

To check profits on some other sites, I moved sets of my miners off to gather some data.
Rather than shifting the pool orders around I just deleted the username from PH pool info so it would find it as DEAD and move down the order for testing.

So blanking out the username does in fact make PH show up as DEAD. HOWEVER...even though its DEAD, Ive gotten over 2X as many GetWork requests from PH as the site im actually doing the testing and mining on.

Does it take more cycles to validate the username/account to make sure you are sending work out to miners who can connect and perform it, or does it just take fewer cycles to send out work to ever miner on that is requesting it? Are my test rigs getting info sent that they have no way to ever fulfill? Does that take the profitability away from the miners ON the pool?

Image
User avatar
mrSnipes
Posts: 5
Joined: Sun Sep 03, 2017 6:12 am

Re: Status as of Thursday, September 7, 2017

Post by mrSnipes » Thu Sep 07, 2017 6:18 pm

Yeah, please open a EU Server, then i will put my l3+ back towards you :)
Post Reply