Quote from: slush on March 09, 2011, 06:53:15 PM

(I simply cannot stay outside)

Yes, less load on server for +/- same total pool hashrate. But you (as a miner) are not motivated to have huge hashrate on server, you're motivated to have high OWN hashrate. Higher pool hash rate does not make your payouts higher (only more steady). So with long ask rate to server, you're cutting your effective hashrate and losing some valid shares because you work on stale jobs.

Maybe your pool wouldn't be in the situation it is in now if it had less network traffic against it.

Our version of poclbm does NOT lower the hashrate of anyone using it. Theoretically, or otherwise. It gains shares at the same speed, and as long as we're accepting the shares, it doesn't negatively impact the user in any way, but benefits us greatly.

Even if a small percentage (and it's small, < .55%) of shares are stale, we can still allow 10 new users for every 1 user that runs our miner due to the reduction in server load.

Quote from: xenon481 on March 09, 2011, 04:39:48 PM

To be fair, a more efficient pool server means that costs per miner are lower and those savings "can" (and currently are) be passed on from the pool operator to the miners.

The problem is that the way that this pool has implemented their efficiency gains comes at the cost of increased percentages of Stale (unpaid) wasted processing power. And to make it worse, it specifically causes slower miners to incur the worst of this increase of Stales.

We've been completely open about why we said it was more efficient. We've been completely open about what our definition of efficient was... Our definition just happened to be the same as the dictionary's definition of efficient.

Yes, it is for the benefit of the server. We charge NOTHING to participate however, so there is no benefit to the "operators" aside from being able to participate in a pool that functions the way we want it to...

And no, the slower miners don't see any negative impact, because we accept shares, and pay them for shares for (CurrentBlock) and (CurrentBlock - 1). If a block is solved every 600 seconds, it would take them 1200 seconds to have their new shares for a getwork become stale.

If 1200 seconds is too slow for you to submit I share, I apologize, but I might suggest upgrading from your PIII to save you some trouble.