Topic: [CLOSED] BTC Guild - Pays TxFees+NMC, Stratum, VarDiff, Private Servers - page 376. (Read 903163 times)

Sorry folks. I should've said poclbm HAS support and cgminer WILL SOON have support as eleuthria said in a previous post.

Nobody seemed inclined to comment on my previous posts, but it's kind of bugging me.

Why does stratum require all lower case worker names in the command line?  I'm using poclbm w/native stratum support for testing.
Thanks,
Sam

It doesn't, I think allinvain misread this:

What?  cgminer supports stratum now?  I haven't seen a new release since 2.7.5.

Or are you commenting on the fact that ckolivas said that he would consider supporting it?
Sam

What does your WU: say?
Sam

U=3.1 for 860Mh/s at BTC Stratum and  11 on a few pools when using Stratum. Rebooted all systems and they settled into 3's overnight. Ran elsewhere and get U of 10-11 in a minute and that is where its been for near 1/2 a day already.
Had been running great until difficulty changed then it took a crap all over and I don't see if the Diff is 1 or higher but I am looking
 
Where would that info be.. I am still  looking for my dynamic difficulty

Nice to see that poclbm and cgminer have native support. I'm waiting on phoenix next and MPBM (fpgas). However I've got a little bit over 5 machines so I suppose it would be more efficient to have one central proxy for all the miners, but the bandwidth cost would be negligible even on the most basic dsl or cable connection. What was it that each stratum connection consumes, something like 10 kB/minute?

It would be cool, but it would be "pointless".  poclbm now has native support, and cgminer is adding native support soon.  Between these two mining clients, that's roughly 80% of the speed on most pools.  Once the miners have native support, you can just run regular backup pools instead of the proxy, or use 'load balancing' in cgminer to split work among multiple pools.

The proxy is mostly a short-term solution to get the protocol out there and usable for old mining software.  It may see some long term use though for large farms (more than 5 separate machines).  The overhead on stratum is basically static per connection.  So running 5 machines with 5 separate connections will use 5 times the data, whereas using a single proxy for 5 mining rigs uses the same as a single machine would (at least with dynamic difficulty adjustments).

Thanks for the clarification and keep up the great work. So far I'm loving the improved accepted % of the new stratum mining protocol.

One thing that would be cool is if we could run just one mining proxy for multiple pools - ie if the mining proxy had some sort of config file or could bind to multiple ports on the machine running the software.

The Stratum protocol can merged mine namecoins, if the pool supports it.  Currently neither slush or myself have implemented merged mining into the stratum pools.  As of right now, NMC is worth so little that is just hasn't been a high priority compared to the other work going into Stratum.

Not sure if this has been asked before, but with the new beta STRATUM mining protocol do we still mine on the namecoin blockchain as well? I don't see any rewards  for NMC.

Fixed the redirect problem, thank you for reporting it.

Any reason why my browser is giving me the basic authentication prompt when I point it to btcguild.com?

Edit: Seems to be a redirect problem. If I type the full HTTPS URL to login.php I get in fine.

I have made some large changes to the DNS again.  If your miners are experiencing extremely high rejects, please restart the mining software.  If it continues, you may need to reboot the machine.  I've still not figured out why some computers seem to completely ignore DNS changes, or bounce between new and old DNS entries repeatedly.

This does not affect Stratum users.

I am seeing this as well. Switched to the 0.8.3 windows proxy and my 4 rigs running BAMT 0.5 (Phoenix/Phatk2) stopped having issues so far. It's been about an hour where before they would drop to another pool within 4-5 min. I know could switch to cgminer/bfgminer on them, but I'm waiting for BAMT 0.6, since they have been stable for over six months, I'd rather not mess with them. 
If anyone is looking for the latest windows proxy version, located here: https://github.com/slush0/stratum-mining-proxy/downloads

For clarification on how dynamic difficulty works, from start to finish:


1) When you first connect, you are submitting shares of diff=1 (or greater).
2) If you submit shares very quickly, you will eventually be adjusted to a higher difficulty.
3) When you submit a higher difficulty share, on the BTC Guild website your 'Accepted' share count will increase by your difficulty.


Example:

A 4.2 GH/s miner connects and submits 1 share per second at diff=1.  These shares get posted as 1 accepted share per valid submission.
They get adjusted to diff=2, and begin submitting 1 share per 2 seconds at diff=2.  These shares get posted as 2 accepted shares per valid submission.
They get adjusted to diff=4, and are now in the target zone, submitting 1 share per 4 seconds.  These shares get posted as 4 accepted shares per submission.


When multiple workers are connected through the same proxy, this same logic occurs, but it is adjusted for your entire farm, rather than just one worker.  Let's assume you still have 4.2 GH/s, but it is split over 5 miners (each miner is 840 MH/s):

At first connect, each miner is submitting shares approximately once per 5 seconds on average) adding up to 1 share per second in aggregate.  They get posted as 1 accepted share per submission.
They get adjusted to diff=2, and now each miner is submitting a share once per 10 seconds on average.  They get posted as 2 accepted shares per submission, to the miner that found the share.
They finally adjust to diff=4, with each miner submitting a share once per 20 seconds on average.  Once again, it's posted as 4 accepted shares to the miner that solves a valid share.



Similar to solo mining or pool luck, you will always have some variance on individual share times, but at a target rate of once per 3-10 seconds, you should see very consistent share submissions.  The big exception here is if your farm has a large number of workers.  An individual worker in the above example averages 1 share per 20 seconds.  However, they may get lucky and submit 3-4 shares in 20 seconds.  If they're unlucky, they might take 3-4 minutes.  But your entire mining farm as a whole will see very consistent submissions over a 30-60 minute period.  Over a 24 hour period, the variance is almost non-existent.

Found that the hardware errors were stemming from a restart of my stratum proxy and Diablominer on that slow worker freaked out.

I can't say why you would get hardware errors.  Nothing the pool does could cause hardware errors as far as I'm aware.

With difficulty adjustments, your difficulty as adjusted per connection (per proxy).  Your entire farm will submit one share per 3-6 seconds on average, depending on which particular worker was "lucky".  Your overall farm variance will not be affected, just the individual worker variance.

If you have a mining farm with very large discrepancies between the speed of rigs, you may want to run two proxies, one for slow rigs and one for fast rigs.  This is especially true for somebody with a BFL mini-rig.  However, don't think that running two proxies so your individual workers submit work more frequently will have a noticeable effect on your earnings.  It's mostly a "feel good" measure so you don't have individual workers with large gaps between submits.

1 and 2:  When difficulty is changed, the message is sent with a longpoll telling your miners to abort previous work.  ANY shares submitted from prior work will be rejected, otherwise the system is open to exploiting by fast miners withholding shares to submit after difficulty has been adjusted to change their rewards.

Same
running 0.8.3 and working good.