I've been mining on p2pool for well over a year without any problems.
Then came the hard upgrade and I've been having problems ever since

I've upgraded to bitcoind to 0.10.2
I've upgraded to p2pool 14.0

BUT . . .

my local p2pool status says:

P2Pool: 11391 shares in chain (2750 verified/11391 total) Peers: 7 (1 incoming)
 Local: 4285GH/s in last 10.0 minutes Local dead on arrival: ~0.0% (0-1%) Expected time to share: 50.9 minutes
 Shares: 18 (0 orphan, 0 dead) Stale rate: ~0.0% (0-18%) Efficiency: ~112.5% (92-113%) Current payout: (0.0365)=0.0365 BTC
 Pool: 689TH/s Stale rate: 11.1% Expected time to block: 3.6 days


but it's WRONG

Coin Cadence says my payout is now zero, and that the Pool rate is 1.5PH/s

*** I deleted the data directory yesterday, either the shares download has failed, or it's taking forever ***
*** the local p2pool global pool rate started higher and has been slowly going down **
*** the local p2pool payout started low and has been going up, while the Coin Cadence payout has been doing the opposite, it started high and worked it's way down to zero! ***

suggestions?
What have I done to break it and how can I fix it?

ADVthaAnksNCE

Thank you for this research and posting it.
This reinforces my impression that the sweet-spot or elbow of the speed vs power is around or slightly less than 1300Gh/s

I managed to get hftool working on a Win8.1 box, had to comment out some lines of code, still got errors, but it worked enough to list the die speed/voltages, and to write back out to flash.

You're probably SOL for the firmware upgrade.

Has anyone managed to get hftool working on a system with more than one Habanero board?

Does anyone have any perl code, or can point me to code, that will let me manipulate the speed and voltage of the dies on a Hab board?

I'm trying to dynamically adjust the speed of the boards as the temperature changes.

I'm not using any A/C, only outside air, and over the course of the day the temp goes up and down and the boards need to be adjusted in turn.

Right now I'm modifying the *.conf files and then restarting cgminer, I'd like to do something a little more elegant.

ADVthaAnksNCE

damm!  keep blowing the breaker.

DUUUHHH!!    120vac * 15A = 1800W

can't put more than 2 haps per circuit!

amazing amount of heat

https://farm4.staticflickr.com/3798/14338282423_56600662c8_z_d.jpg

BTW - I wouldn't run chilies and hasps on the same instance of cgminer.  I got a significant jump by turning off the chilies

How do you get it that low?  I've never had p2pool GetBlockTemplate average less that 1 second, and that's on a SSD, but over DSL.

ADVthaAnksNCE

info of my experiences so far

Corsair h100 with stock fans

standard MinePeon Version 0.2.4.6 image running on RiPi

) had to switch the miner from bfgminer to cgminer 4.0.1 to get it to work, but it's not a version that supports all the hashfast features.  I gather I can get an enhanced version somewhere?

) evenly tightened the cooler nuts, tight, but not too tight.  waiting for release of the tool to fine tune the temperature/nuts.

) 4: HFB 0: 650MHz  93C 100% 0.87V  | 603.4G/629.0Gh/s | A:294297 R:14772 HW:1301 WU: 8751.2/m

) couldn't get 4 chilies and one haspanero to be recognized and running by minepeon.  finally figured out I had to sequence the power up.  turn on the chilies and wait for them to finish their boot.  turn on the haspanero and wait for their boot.  THEN turn on the RiPi.

What temperature should these be running at?  is 93C at 650Mhz way hot?
What components other than the ASICs need to be kept cool for max speed? i.e. is it worth it blowing air on top or bottom of the board?

ADVthaAnksNCE

uugg!   for some reason mine was shipped to my billing addr instead of shipto addr, no one home to sign for it so no delivery.
no new plaything for tonight  

I'll have to pick it up in the morn,  playtime starts tomorrow night

going to try starting out with a stock corsair h100 for cooling and one DPS-1200FB per Hab
mods to follow as needed

Am I correct in thinking that the payout an addresses will get, is directly related to the number of shares in the chain that were found by that address?

Since shares are constantly dropping off the chain, you have to keep finding shares, otherwise as your shares drop off the chain, your payout will also drop?

If you have a dry spell your payout will drop, because more shares are dropping off than are being found.
Likewise if you have a hot spell your payout will increase as more shares are added than dropped.

How do dead and orphaned shares fit into this?

Is this a reasonable accurate understanding?  

ADVthaAnksNCE

How much current will each die be pulling?  i.e. a power supply has to be able to supply at least XXA or XXXW, or a multiple there of . . .

ADVthaAnksNCE

Interesting, but @750W 62A it seems to be just shy of having enough current/watts with a reasonable safety factor, unless you do a redundant 3 per 2 Habs.

Does it need to be a PC power supply?
Or will a simple 12V 100A supply work with the right cable/connectors?

ADVthaAnksNCE