Topic: [1500 TH] p2pool: Decentralized, DoS-resistant, Hop-Proof pool - page 767. (Read 2591916 times)

1. I setup the bitcoin.conf and use the Windows GUI version 0.52 of the Bitcoin Client and forward TCP ports 6667 and 8333 to the Client.
2. I setup p2pool and forward TCP Port 9333 to the client.
3. I setup CGMINER to use -o http://127.0.0.1:9332 -u User_from_bitcoin.conf -p Password_from_bitcoin.conf.

CGMINER works, there are many LPs and a higher reject rate, than with a normal pool, which is okay from what I know.

But the question is now, how will the payouts work? Is there any good read or reference for this?
I'm asking, because in the p2pool window I see:


and


Will it take 1.4 hours to see an estimated payout amount and will that directly show up in my Bitcoin Client GUI?

Thanks,
Dia

Is the flag to change queue in cgminer -Q 1?

edit: just worked it out

Hmm I wonder what that is:
value :   0.00077170 BTC
scriptPubKey :   OP_IFDUP OP_IF OP_2SWAP OP_VERIFY OP_2OVER OP_DEPTH

also Your coinbase (7) : 062f503253482f (./P2SH/)

Just saw a payment that was too high and found this

That's my first block on p2pool.

There wouldn't be a lot of value to that.  The only reason for wanting to check the entire nonce range is to reduce load on the server but w/ p2pool you are the server so a couple extra getworks per minute isn't going to really create any load issues.  As GPU/FPGA continue to get faster the it becomes less of an issue anyways.  A 7970 for example will need to process 2 nonce ranges per LP on average.  So will BFL & Icarus FPGAs.

In the context of P2Pool, no it isn't particularly useful.

DeathAndTaxes,

on p2pool it would be usefull to have an option that makes cgminer split nonce ranges between GPUs so that the full 4 billion nonces can be tested during the available 10 seconds.

spiccioli.

I'm running a 6770 at around 200 MH/s with 1 queue and 1 thread.  Also I use dynamic intensity which usually sits around 4-5.
Sounds like my settings are right where they should be.

I'm thinking this discarded work metric isn't really that useful anyway.

Ah, oops, thought you were talking about rejects.. Anyway, it isn't harmful. Mine is about 80%.

Discarded work isn't rejected.  It is simply work cgminer requested and then never starting working before discarded due to LP so it doesn't "cost" you anything.

Either because of settings or a bug cgminer is requesting a lot more work from p2pool then it will be able to finish.  When next LP occurs that requested work becomes stale so cgminer discards it (without doing any work on it).  It then requests a crap ton of work it will never be able to finish and 10 seconds later the same thing happens.

If you have threads per GPU > 1 and/or queue > 1 then change them both to 1.  Also knock the intensity down 1 or 2 notches.  

With a "normal" pool I find the optimal settings for a 5970 to be:
Queue: 2
Threads per GPU: 2
Intensity: 9

For p2pool I find the optimal settings for a 5970 to be:
Queue: 1
Threads per GPU: 1
Intensity: 8

LP are so short that maintaining a large queue of work is mostly wasted.  With an average time between LP of 10 seconds you need to be complete > 4.3 million hashes or you will never complete more than 1 getwork per LP.   That works out to ~430 MH/s (per GPU).  Using higher queue and/or threads really only makes sense if someone if using a > 430 MH/s miner.  Not per rig, or card but per chip.  Using both queue and threads >1 only makes sense on some hypothetical  > 1 GH/s chip.

I wasn't referring to the rejected field but the "DW" field which is discarded work.  My reject rate is 0.9%.

The DW field increments every time cgminer shows a long poll but not an accept.

That's definitely not normal. You should get at most 2% of shares rejected, as displayed in cgminer. If you're getting anything more, upgrade cgminer and decrease its intensity.

What kind of discarded work rates are people getting with cgminer on p2pool?

I read that it's normal on p2pool to have a higher discarded work rate than on a traditional pool.  I'm just trying to figure out how much is normal to see if I have problems or not.
My discarded work is about 1.5x the accepted.

Garr255,

I've found out that using bitcoind 0.6.0 RC1 with latest p2pool.py lowers CPU usage _a lot_.

spiccioli

I'm getting the same error with an i7 cpu, there must be another cause...

Thanks to Holiday tutorial i'm now mining (700mh/s) for p2pool... cheers

Does someone know how this can happen?

The recently found block gave me a reward of about 0.004 BTC, but my p2pool instance says I should have gotten about 0.22 BTC. How's that possible? Total hashing power shows up as about 294 GH/s, so I shouldn't be on a fork. I'm running p2pool v0.9 btw.

Edit: Is it possible that someone else who's on a fork found the block?

Edit2: Mystery solved. The payout came from elligius, I mined a few shares there.

thanks a lot forrestv!

It's pretty easy - start namecoind, then run P2Pool with --merged http://NAMECOIN_USER:NAMECOIN_PASSWORD@127.0.0.1:NAMECOIN_RPC_PORT/

Hi
Is a p2pool merged mining howto avalable ?