stales and what to do to get rid of 'em

[2weiX]

Hi Boards,


I am running three 6950s at core 840 and mem 800, clocked with the ClockTweak commandline tool.
temps: 74/76/72°C
Hashrates: 330 to 340 each.
GUIminer 2011-07-11 with  -v -w128 -f60

I am mining with Mineco.in since I like the payout model there, especially how the transaction fees stay within the pool.

However, I am getting 2 to 5% stales
right now I have 2,46%  2,47% 4,02%


Is there anything you can tell me how to get rid of these? Is is the software, the clocking? the wrong pool, even? Other members of the same pool get 0,5% and less(!) stales.


Any ideas?

[1713484109]

1713484109

[1713484109]

1713484109

[1713484109]

1713484109

[1713484109]

1713484109

[hugolp]

Stales is a problem of syncronization between the pool and you. Its not an issue of your card.

I tried Mine.co and saw huge stales as well. Other pools have better software.

[2weiX]

So I have tried two pools against each other, namely bitp.it and mineco.in


I have three workers (let's call them #1, #2, #3)
On each worker, I started two miners, one for each pool.



This shows that whatever the pool, I get 2% stales on average.

Could this also be because I am on a WiFi? Is there anything (like port forwarding) that could be done to improve this?

I tried Slush as well, same number.

[DrHaribo]

I'm getting 0.048% stale in my own pool with my own miner software. Example: 13 stales and 27046 accepted proofs-of-work.

I couldn't tell you if it's my miner or pool backend. I haven't tried running a different miner for a long time to see if there is a difference.

But I would imagine it depends mostly on the miner software, the pool backend software, and the load on the pool server.

You could try my miner and compare with other miners/pools you tried, see if it helps: http://bitminter.com

Edit: I wrote both the miner and pool backend programs myself - it's not pushpoold like most pools.

[JoelKatz]

Some pool report every invalid share as a stale. Some mining programs don't double-check that a share reported by the GPU miner is valid. This can be a symptom of too-aggressive overclocking.

[2weiX]

Some pool report every invalid share as a stale. Some mining programs don't double-check that a share reported by the GPU miner is valid. This can be a symptom of too-aggressive overclocking.

tried that as well (the not-overklocking).
stock 6950s on 840core 800mem, so all within BIOS parameters (default ist 800core 1250mem).

[compro01]

might give btcguild a try

my long term stale rate is sitting at 1.38% (599 stale out of 42931)

i suspect the extremely low stale rate is one of the reasons why so many people are staying with deepbit.  i got 0.71% stale (out of 160840 shares) when i mined there until i switched.

[phelix]

might give btcguild a try

my long term stale rate is sitting at 1.38% (599 stale out of 42931)

i suspect the extremely low stale rate is one of the reasons why so many people are staying with deepbit.  i got 0.71% stale (out of 160840 shares) when i mined there until i switched.

btcguild helped me with high stales, too (after some update they did a couple of weeks ago)

you might also want to try http://bitp.it

[anodyne]

Could this also be because I am on a WiFi? Is there anything (like port forwarding) that could be done to improve this?

The wireless is probably the culprit here. Mining doesn't take much bandwidth, but latency can be a big problem. Not sure if there are any settings you can use to improve it. Encryption and authentication probably adds quite a few unnecessary milliseconds, especially with normal consumer adapters and routers, but turning that off is of course limited by what else the network is used for.

In any case, if you can move the rig to test it with a physical connection I'd recommend trying it out.