Topic: [XPM] [ANN] Primecoin High Performance | HP14 released! - page 105. (Read 397616 times)

I have one PC running around 1700 and have gotten 2 blocks in 2 days...there's a lot of luck involved apparently because with the first release I got a block in 2 hours running 5-30 pps

Get out bro. I mined 3 blocks in the past 48 hours on 3800 pps.

To the guy asking the question. I expect you to find at least four blocks weekly. Not sure though. I haven't got enough data to do probability maths.

Hmmm... that makes me think it isn't worth it anymore 

Are there any other CPU coins?

Looks like mikhael's github repository has been DDOSed :-(


Onkel Paul

35.000 pps get me one block the last 20 hours, you do the math

Can someone help me out? 2 questions... 1 if pool mining with several CPU should I add a new worker for each CPU? Is it even possible to run several CPU off one worker? Is it better to just have 1 worker per CPU? 2. What do I need to do if I want to solo mine on more than one CPU but want all the coins to go to one wallet? Thanks

So I'm getting 1700 primes/sec.

How often should I expect to find a block?

(According to your experience) 

Botnets don't count.

On how much vps or AWS large instance ? Could you tell your global pps and sievsize used ?

I'm not too sure of command line. I added to my .conf file "sievesize=2000000" eg.

Is there a command line argument I can pass in to change the sieve size?

About 100 in the last 24 hours, but I'm guessing you didn't want to hear that.  If it's any consolation the rate has dropped off dramatically and I'm now getting less than 2/hour with the same hardware.  In another 12 hours it will no longer be profitable to mine using those VPS's without a big jump in miner performance (I'm looking at you mikaelh )

Me and my friend both found another block last night. Again, both of us have a single computer of i5 performance running the miner. How about you guys?

I don't like how much variance there is currently in the 5-chains/h numbers. An rpc command would probably need to include some averaging. When the numbers are dumped to debug.log, people can do the averaging manually.

Some nice numbers there still.

Here some data.



Machine A (32-core, 512MB RAM ) Sieve=4.000.000  
2013-07-16 15:37:54 primemeter  47798623 prime/h 336961584 test/h      3180 5-chains/h
2013-07-16 15:38:54 primemeter  46317188 prime/h 327397443 test/h      3600 5-chains/h
2013-07-16 15:39:54 primemeter  44566177 prime/h 311356231 test/h      3480 5-chains/h
2013-07-16 15:40:54 primemeter  44270022 prime/h 307570114 test/h      2880 5-chains/h

--> 102 5-chain/core/h


Machine B  (25-core, 512RAM  )   Sieve=8.000.000  
2013-07-16 15:30:44 primemeter  28974114 prime/h 198837632 test/h      2640 5-chains/h
2013-07-16 15:31:44 primemeter  29712630 prime/h 204013540 test/h      2220 5-chains/h
2013-07-16 15:32:44 primemeter  26133944 prime/h 178498885 test/h      1740 5-chains/h
2013-07-16 15:33:44 primemeter  26519858 prime/h 178976417 test/h      2520 5-chains/h
2013-07-16 15:34:44 primemeter  28332196 prime/h 191231866 test/h      2100 5-chains/h
2013-07-16 15:35:44 primemeter  28916158 prime/h 196550784 test/h      2220 5-chains/h

 --->90 5chain/core/h


Machine B (5min test)  Sieve=100.000
2013-07-16 15:50:06 primemeter  57670379 prime/h 534606270 test/h      1440 5-chains/h
2013-07-16 15:51:06 primemeter  56831053 prime/h 524772794 test/h      1200 5-chains/h
2013-07-16 15:52:06 primemeter  56350341 prime/h 521713925 test/h       900 5-chains/h
2013-07-16 15:53:06 primemeter  56199923 prime/h 517748591 test/h      1260 5-chains/h
2013-07-16 15:54:06 primemeter  56103505 prime/h 524118865 test/h       600 5-chains/h

---> 43 C5/core/h



Low Sievesize is not a good idea.
I did that manually. We need a program who calculate that with bigger sample.

Mickael, why not include a command to remplace getprimespersec?
I think 5chain are more reliable.
For information, i found 2 block on machine A and 1 on B. But it's too low to have a significance.

I would trust the 5-chains/h number.

The test/h number doesn't really have any useful meaning as far as mining is concerned. It's just the number of candidate multipliers produced by the sieve algorithm.

The primes/h number tells how many of the multipliers produced a probable prime chain of length 1 or greater.

The 5-chains/h number tells how many probable prime chains were at least of length 5. It's supposed to be the most reliable performance number as far as mining is concerned. A 5-chain is essentially a lot closer to an 8-chain than a 1-chain.

Yeah I keep getting that too.  Not out-of-sync error message but if I have mined a coin and don't see the number confirms to go move for awhile, I have to restart it.  I have even restarted it and then suddenly have a coin coming.  Never got the out-of-sync error except when at a restart of the wallet client.

I'm getting "syncronizing with network" every hour, after restarting primecoin-qt it synchronizes and works ok, but after some time it is "out of sync" again 

@mikhaelh, any comment on what to do if test/h and 5-chains/h don't agree in debug.log for hp4 ? For example when sievesize=2M, I have  50M test/h and 350 5-chains/h, when sievesize=4M, I have  36M test/h and 780 5-chains/h.

https://bitbucket.org/pankkake/mikaelh2-primecoin