[~1000 GH/sec] BTC Guild - 0% Fee Pool, LP, SSL, Full Precision, and More

[eleuthria]

We are NOT on difficulty 2 shares.  That was planned a while ago, but it was due to some poor scaling on the database backend.  Difficulty 2 shares would've cut the DB load in half, but I was able to revise the DB schema and completely eliminate the need for it.


If you're receiving stales over 2%, odds are your miner is screwed up.  It is nearly impossible to have that rate of stales.  Nearly every user is reporting rates of 0.5% or less.

The most common reason people are seeing invalids is that their miner is sending shares to multiple servers at once.  PM me with your IP Xephen, there's one possible reason for the huge increase, but I will not announce it in the thread as its only affecting a very small subset of users and its for a very specific reason.

[1713427910]

1713427910

[1713427910]

1713427910

[1713427910]

1713427910

[eleuthria]

In the end, a stat delay is what we're going with.

+1

EDIT: Btw, I dont know if Im just having bad memory, but it feels like lately we are having more variance. 2'2 TH/s we should be around 45 minutes each block on average, but we are getting several very long blocks (4+ hours) and then several very very short blocks (less than 10 minutes). I dont remember so much variance before. Probability is a bitch I guess.

Individual block variance is always pretty large.  The point of having a high speed pool isn't that your variance per block is low, its that your variance over a period of time is low.  Looking at our luck over an entire difficulty shows our variance has been fairly low (the last two difficulties were less than 1% from average).  Even this difficulty we're within 10% of the expected average, and we were offline/slowed for a very large chunk of the difficulty.

[BookLover]

Really, I thought eleuthria said when he implemented difficulty two shares that each share show up as two on the site and the site is still showing odd numbers of shares. 
Yes I've restarted my miners.  I use a combination of a .bat file and task scheduler to restart my miners every half hour. 
No I haven't updated guiminer recently but this version had been working just fine.

I would upgrade guiminer for no other reason than the newer versions of poclbm are significantly faster than they used to be (depending on how old your version is).

Upgraded still having ~10% stale shares.  I wonder what the problem is which eleuthria won't announce?  I can't believe I'm the only one with this many stales. 

Here's some more information for those who are trying to help.  I'm running poclbm through the latest guiminer, I have two miners for each core(the miners are pointed to different servers and I've adjusted the -f setting to -f 30 and -f 65), I also have these settings(-v -w 128), I'm running multiple machines on the pool but have been unable to determine if all the machines are affected.  I made no changes to the settings near the time of increased shares except for possible a change in where I was pointing the miners, I'm running a windows7 64-bit operating system, and I restart my miners every half hour, and my cards are HD Radeon 5970's.

[eleuthria]

If you're using TOR, expect shitty rejected share rates.  Just an FYI.  The IP changing of TOR will mark your account as a potential botnet and give you last priority on LP updates/getwork processing.

[eleuthria]

Pool stats are now running on a 1 hour delay.  Current round stats/estimated reward have been removed to remove any way to scrape data from the website to determine when we have started a new round.

Anybody using the API should remove the following pieces from their code:
  estimate_reward, round_shares, round_stales, round_time

They will be taken out of the API in about a week.  Until then they have been filled in with dummy data to avoid breaking gadgets/parsers, as well as give me some good stats on how much speed is involved in pool hopping.

[Yatta99]

In the end, a stat delay is what we're going with. 

Not a problem, just throwing something against the wall. Sometimes it sticks, sometimes it doesn't 

IRT stales: running a stale rate of only 0.226% here so very good job on whatever you're doing 

[sirky]

I also have almost no stales here as well.

[bodhipraxis]

I do not like the removal of round stats. BTCGuild is my primary pool. I have paid 2.5% the ENTIRE time I have mined here. What is the distinction between "pool hopping" and having a system like SmartCoin that automatically compensates for pool downtime? Eleuthria, you had MAJOR downtime earlier this month, and I had to mine elsewhere. It is a real handicap not to know the round stats. Why don't you offer them only to 2.5% donors as one of your "perks"?

[ellipsis]

The API is returning invalid JSON if you have no current shares again.

If you're using TOR, expect shitty rejected share rates.  Just an FYI.  The IP changing of TOR will mark your account as a potential botnet and give you last priority on LP updates/getwork processing.

Weak.

[bodhipraxis]

yeah, its too bad. My mining group just brought up new machines. We were going to point them at BTCGuild. Now reconsidering - 13Gh/s to be redirect. I will advise that we keep 3 GH/s pointed at BTCGuild, until Eleuthria clarifies his stats policy.

[hugolp]

I was thinking, and the pool hoopers could know sometimes when a round starts. When the previous round was longer than 1 hour the round will appear and the pool hoopers will know a new round has started. Its a bit of a long shot, but I think the solution is quite easy. It could be solved by delaying 10 minutes the report of a block that was longer than 1 hour.

[eleuthria]

yeah, its too bad. My mining group just brought up new machines. We were going to point them at BTCGuild. Now reconsidering - 13Gh/s to be redirect. I will advise that we keep 3 GH/s pointed at BTCGuild, until Eleuthria clarifies his stats policy.

What clarification?  Stats are delayed by 1 hour, exactly the same as Deepbit.  The alternative is you have pool hoppers that DO impact your rewards per day.

[flower1024]

i still prefer "the user transparent" way too.

just start with a stats page how many people earn more than others through hopping (i already offered you writing the db statements, just need your schema)

THEN if ALL know whats going on, we can make a decision.
maybe: top 30% of people who profted from entering/leaving too much get a cut.

if you state that clearly one your site i don't see a problem. as it would only affects rounds which the miner left early

[bodhipraxis]

So Eleuthria -- PLEASE CLARIFY:

You are no longer showing "unconfirmed rewards", right?

And you are delaying "confirmed rewards" as well?

So...basically we don't know if we are getting our block reward until....?

'cuz that's how it looks right now in the Dashboard!

[eleuthria]

So Eleuthria -- PLEASE CLARIFY:

You are no longer showing "unconfirmed rewards", right?

And you are delaying "confirmed rewards" as well?

So...basically we don't know if we are getting our block reward until....?

'cuz that's how it looks right now in the Dashboard!

Unconfirmed/Confirmed rewards do not increase until 1 hour after the block was found.  For 2.5% donators, the 1 hour delay means your rewards don't go up until after that hour has passed.  Otherwise its trivial to track when a new round starts.

The unconfirmed rewards will reach confirmed status in the exact same amount of time.  The delay is on publishing a new block's stats, not updating confirmations on blocks that have already been published.

[Peao]

Pool stats are now running on a 1 hour delay.  Current round stats/estimated reward have been removed to remove any way to scrape data from the website to determine when we have started a new round.

Thank you, eleuthria!

[bodhipraxis]

How exactly does it negatively impact me if others pool-hop, and how does that impact compare with not knowing how my miners are doing in your pool? (looking at one of my miners right now, and it sure likes a bunch of new work was pushed a couple of minutes ago, but no update in "confirmed rewards")
Now we can't have an estimate of how many shares we are contributing?
Eleuthria, some of that drop in your hashrate is due to people who prefer BTCguild pointing their clusters away from you.
My three machines in my office (all that be thermodynamically sustained with the cooling I have available) are pointed at BTCGuild -- BUT...
I work in a data-center rich envr. and I know for a fact that in this building, there are at least 14 other miners, most with 2-5 GH/s on site...we just talked in the elevator about you, since most of us have preferred your pool for QUITE a while.
I know that you work, work, work...that's why I tell everyone in my mining group to DONATE, DONATE...

[burp]

I do not like the removal of round stats. BTCGuild is my primary pool. I have paid 2.5% the ENTIRE time I have mined here. What is the distinction between "pool hopping" and having a system like SmartCoin that automatically compensates for pool downtime? Eleuthria, you had MAJOR downtime earlier this month, and I had to mine elsewhere. It is a real handicap not to know the round stats. Why don't you offer them only to 2.5% donors as one of your "perks"?

Sounds good to me, when you donate 2.5% you may pool hop

[eleuthria]

How exactly does it negatively impact me if others pool-hop, and how does that impact compare with not knowing how my miners are doing in your pool? (looking at one of my miners right now, and it sure likes a bunch of new work was pushed a couple of minutes ago, but no update in "confirmed rewards")
Now we can't have an estimate of how many shares we are contributing?
Eleuthria, some of that drop in your hashrate is due to people who prefer BTCguild pointing their clusters away from you.
My three machines in my office (all that be thermodynamically sustained with the cooling I have available) are pointed at BTCGuild -- BUT...
I work in a data-center rich envr. and I know for a fact that in this building, there are at least 14 other miners, most with 2-5 GH/s on site...we just talked in the elevator about you, since most of us have preferred your pool for QUITE a while.
I know that you work, work, work...that's why I tell everyone in my mining group to DONATE, DONATE...

When I first implemented the stat delay, I immediatley changed the pool stats API to show the current round at 1 second with 100 shares.  Within 15 minutes the pool speed increased over 200 GH/sec.  We are still higher now than before I put in that delay.

Reset stats are almost fully restored (just adding the reset buttons now).  That will let you monitor miner performance when testing new configurations.  Last share times are also about to be restored, which will be almost immediately followed by miner idle settings being visible again.

[gentakin]

Within 15 minutes the pool speed increased over 200 GH/sec.

So there's at least 200ghash/s worth of pool hopping power right now. Interesting!