I think a valid question is whether AM is structured to have sufficient capital to invest in next generation devices without raising funds. For now, I assume they do, but there really isn't much information available to us to verify that.

Also, mining dividends are quite unlikely to *suddenly* disappear except for loss of the farm (catastrophe, natural disaster, sabotage, seizure). Otherwise, if AM are unable to maintain their percentage of the total hash rate, I would expect it would be evident in a gradual decline over a few months. To me at least, that's a little different from the BTC run-up, because BTC don't pay dividends. Hardware sales revenue is subject to much faster winds than hash rate, but AM has some structural and scale advantages from playing both sides.

So,, you must find a bitcoin block? How does that work on P2Pool? Isn't the difficulty on bitcoin
a lot higher than the others?

Maybe this is a dumb question, but if you are merge mining do you only get blocks in the merged coins when you find a bitcoin block? Or are the rewards independent?

ASICminer were one of the first to market and did quite well. They built a nice hashing farm and were able to sell devices with quite large markups. But they were pretty much the only clowns in the town. These questions remain:

1. Can anyone else produce ASICs? (yes)

2. Will they put pressure on ASICminer's markups? (yes)

3. Can ASICminer maintain their share of the network? (probably)

4. Can ASICminer maintain the margins of the hardware they sell? (no)

5. Do superior ASICs exist? (yes)

7. Has ASICminer demonstrated an ASIC competitive with what the competition has demonstrated? (no)

8. Can ASICminer produce a more efficient chip without stumbling? (unknown)

ASICminer had a brilliant strategy to get to market first by relying on outdated manufacturing and idle factories. They did that well, but now they need to move to the next stage where the others have been toiling. ASICminer is a good hare, but given how close the tortoises appear to be, ASICminer better not trip.

I can understand that shares decrease in value as the total number of shares in the block increases (hence the overall downward sloping of the flat regions). But what explains the jumps up and down? I only have one orphan, and 9 valid shares in this block... (two found during the time plotted there). The jumps up only correspond with one found share... (the orphan was before the start of the plot).

Or are the big jumps down because someone found a very high value share?

Who cares if AM or BFL was a better investment last year? It has jack to do about the future and purchase decisions today.

Why does this happen?



After two days I have 10 shares and 1 orphan and 2.4% DOA? Am I doing something wrong?

Isn't this JSON-RPC thing for getblocktemplate completely stupid? Why aren't we using a shared memory segment with bitcoind?

Thank you for the pointers!

I actually had read those guides, but I think I feel victim to pre-mature optimization and applying without testing. Particularly in light of gyverlb's advice linked by GrapeApe which describes actually methodically testing and measuring the results of tuning to see what's working (sacrilege!). So... I disabled all the tweaks for now in bitcoin.conf and restarted bitcoind... But it turns out there was also a second "mistake" I made.

After restarting bitcoind without tweaks, getblocktemplate latency started out low but then would gradually and steadily build up to ~2sec(!) over the course of 30 minutes! So I decided to open 8333 for bitcoind to maybe get more connections. That didn't do much, either... That's when I noticed that I had been too clever and set "ufw limit" on the bitcoind and p2pool ports (linux kernel firewall rate limiting). When I deleted those rules and replaced them with the corresponding "ufw allow" unlimited rules my rate dropped considerably and seems to have stayed low over night (0.1-0.3s for the last six hours). So, I'll now work on tweaking more methodically.

Thanks for getting me pointed in the right direction!

The profit margin on the USB eruptors is less than 50% what it was. There will be no windfall from those, even if anyone bites. And honestly the group buy buzz seems quite lackluster, certainly not anywhere near 2x previous volume. We'll see how btcguild's direct sales go.

Hi, sorry if this is the wrong place to ask. I'm new to p2pool.

I've got it setup and running (debian) and I just want to make sure things are looking OK. I have 330MH/s (AM USB eruptor). I've been running p2pool for 15 hrs, expected shares are ~5hrs, but I've only found 1 share. Is that something to be concerned about yet? Or is this amount of variance normal and I should sit tight? I don't have an issue letting it run longer waiting, but if it's already apparent that my setup is doing something wrong, that would be good to know if I need to roll my sleeves up. My getwork latency is ~1s. I apparently got 1 share so something is working...

Version: unknown 7032706f6f6c
Pool rate: 621GH/s (20% DOA+orphan) Share difficulty: 1190
Node uptime: 0.620 days Peers: 7 out, 0 in
Local rate: 308MH/s (9.3% DOA) Expected time to share: 4.60 hours
Shares: 1 total (0 orphaned, 0 dead) Efficiency: 124.7%
Payout if a block were found NOW: 0.00262778 BTC to xxx. Expected after mining for 24 hours: 0.0124 BTC per block.
Current block value: 25.28591 BTC Expected time to block: 37.2 hours

Also, I'm running the 11.4 release--is that recommended or do most people run the master branch from github?

Is peers in at 0 a problem? I can telnet to port 9333 from remote. Do I need more open ports?

You're counting orders, not devices. Even that one lab_rat order in your screenshot was for two devices...

Yes, that's the short view.

The long view is that each of these devices poisons ASICminer's customers. They burn every customer in more ways than one. You can only fool your sheep customers a few times before the herd learns to not come back and graze. Friedcat's already said we won't have new chips before October (I'm not sure of the language barrier but the phrasing leads me to believe that he's leaving the door wide open for well after October as a definite possibility--which means they're playing catchup).

If ASICminer comes up with new chips and prices them competitively then it's probably fine, but they will be much, much tighter margins than we're used to as stockholders going forward. If ASICminer thinks that their track record alone commands a premium, they better not stumble or swerve at all. It's inconcievable that every single competitor will goof as badly as BFL has--and even BFL looks to be in good shape for an October price war.

The present: ASICminer is selling nothing except 1btc trinkets that are not worth 1btc. (They're fun trinkets though--but trinkets aren't serious mining hardware)

Relying on buyers to make irrational purchases isn't a long term or sustainable business plan. That's a collapse waiting to happen and an opportunity for your competitors. ASICminer enables the competition to exist because they cannot shut them out.

That's rich when promoting a company that sells devices at prices that will never return investment. You can't sell ASICs at GPU prices forever and if ASICminer actually relies on those sorts of margins they will be strangled. Friedcat seems like a shrewed guy and I bet he has plenty of tricks up his sleeves, but y'all act like Titantic passengers.

"Interleaved works, but chips have a problem; proceeding with one-job-per-chip from this point"

https://github.com/luke-jr/BitForce_SC/commit/26984f0a5aeae68e0d164dc38934385f474c8d4f

I was discussing yearly revenue (which bitfair had ballparked at 4.2btc/share a few posts before the one I finally replied to).

AM dividends are already impossible to exceed 0.063btc/week. Of course, it's also the case for AM to sell hardware at prices that are very unlikely to ever mine their price (c.f. USB Eruptor). There is a finite supply of suckers though, so that can't last forever particularly in the face of a price war which seems on the near horizon.

Your conservative revenue numbers are fantasyland. If AM mined 100% of the blockchain for an entire year (either by mining or through hardware sales priced to be premined for 1 year), that comes to only:

(25*6*24*365.25)/400000 = 3.28725 BTC/share yearly revenue

Nah. It will take them two weeks to make it that far into their email box. In fact, if you request a refund now, they might just find out by early July.