[ANN] Bitfury ASIC sales in EU and Europe

[ShadesOfMarble]

so is it a good deal ?

Do you want to mine? If yes, it is.
If you are for the quick buck, probably not. Buy BTC instead.

[dani]

What is wrong with my cards? It sometimes drops speed from 60-62 to 50 and sometimes even lower (right now its down to 40). Everything seemed fine, I switched everything back to autotune because the results were more or less the same. Stop/start doesnt help. Before restart I saw in putstat log instead of 0 qeued work numbers between 1 and 10, switchen positions randomly. After a restart it's back to:

0 1343 47 47

• http://127.0.0.1:8332/

0 1323 39 39 [1]http://127.0.0.1:8333/
0 1330 51 51 [2]http://127.0.0.1:8334/

any thoughts?

[Isokivi]

What is wrong with my cards? It sometimes drops speed from 60-62 to 50 and sometimes even lower (right now its down to 40). Everything seemed fine, I switched everything back to autotune because the results were more or less the same. Stop/start doesnt help. Before restart I saw in putstat log instead of 0 qeued work numbers between 1 and 10, switchen positions randomly. After a restart it's back to:

0 1343 47 47

• http://127.0.0.1:8332/

0 1323 39 39 [1]http://127.0.0.1:8333/
0 1330 51 51 [2]http://127.0.0.1:8334/

any thoughts?

Please note that your looking at a 5-minute sample, seems like normal variance to me. Does you pool have any stats you could use, with longer intervals ?

[dani]

What is wrong with my cards? It sometimes drops speed from 60-62 to 50 and sometimes even lower (right now its down to 40). Everything seemed fine, I switched everything back to autotune because the results were more or less the same. Stop/start doesnt help. Before restart I saw in putstat log instead of 0 qeued work numbers between 1 and 10, switchen positions randomly. After a restart it's back to:

0 1343 47 47

• http://127.0.0.1:8332/

0 1323 39 39 [1]http://127.0.0.1:8333/
0 1330 51 51 [2]http://127.0.0.1:8334/

any thoughts?

Please note that your looking at a 5-minute sample, seems like normal variance to me. Does you pool have any stats you could use, with longer intervals ?

You are right, but I also look at pool stats. here you can see my units hashing. http://eligius.st/~wizkid057/newstats/userstats.php/14cbUKDLpZPJUMfr3DcKWZXmiaQRYpaZbd
the speed right now is clearly too low. same happend yesterday evening and for a shorter period yesterday afternoon, as you can see. The spike going to 80+ was a 2nd unit hashing on the same adress, so you can ignore that

Edit:
I set the last 60gh-setting to best.cnf and rebooted, for now its at 58+ slowly reaching 60 as before. I don't see why it performs like this sometimes, there are no chips turned off (besides the ones that are always off, dead chips) at the time of bad slow hashing. It's just that every card starts to perform bad/slow

[marto74]

Hi,
Niko
I've send you a few Emails asking about bulk chip order prices.
Is there any problem ?

[Micky25]

dani, did you try other pools/difficulties?

[dani]

dani, did you try other pools/difficulties?

I tried eligius with diff 1 for the first hour or so, went "okay", changed diff to 32. Went like it did. I was on slush's pool with diff suggested 32, but it seemed not to work (hashrate <3gh according to pool). My friend's unit hashes with difficulty 1, tonight it slowed down to 2.9gh, restarted it and it's working fine again. Any recommendations for pools and difficulty?

[spiccioli]

dani, did you try other pools/difficulties?

I tried eligius with diff 1 for the first hour or so, went "okay", changed diff to 32. Went like it did. I was on slush's pool with diff suggested 32, but it seemed not to work (hashrate <3gh according to pool). My friend's unit hashes with difficulty 1, tonight it slowed down to 2.9gh, restarted it and it's working fine again. Any recommendations for pools and difficulty?

My recomendations for a pool is a few messages back, but difficulty depends on your hashing speed.

At 40 GH you can use 128-256, at 400 you shoud use 1024 or higher.

My 400 GHs kit, which is running at just a little less than 400 GH, mostly because of that partially broken card, is running at 1024 difficulty right now.

You should let it run for a day, at least, before assessing average speed.

spiccioli

[Micky25]

I get 27 on slushs with diff. 16 and 38 on eligius at the same time. I'm new to eligius and don't even know where to change difficulty

For me its like the less I touch, the better it gets. Seems to be a pretty self containing ecosystem, I like that, I'm a lazy guy  

[LainZ]

Received miner(starter kit) yesterday, everything is looking good @ 47GH/S!

Setup was insanely fast!

Thanks bitfury & punin !

[darkfriend77]

I have done 2 test runs ....

temperature was measured from the PCB.

without fans:

PCB @ 60° ... avg. Hashingrate around 41.5 GH

with fans:

PCB @ 41° ... avg. Hashingrate around 42.0 GH

So my conclusion is ... that they do work more or less the same cooled or uncooled ....

... maybee we should add some more information to the web gui ...

[Isokivi]

I have done 2 test runs ....

temperature was measured from the PCB.

without fans:

PCB @ 60° ... avg. Hashingrate around 41.5 GH

with fans:

PCB @ 41° ... avg. Hashingrate around 42.0 GH

So my conclusion is ... that they do work more or less the same cooled or uncooled ....

... maybee we should add some more information to the web gui ...

What would be awesome is if someone whipped up a script that can collect larger samples of data by picking up each individual 5 min log and compiling them in to a longer one with averages. I'd personally prefer a one that shows info on every single chip.

[jlsminingcorp]

What would be awesome is if someone whipped up a script that can collect larger samples of data by picking up each individual 5 min log and compiling them in to a longer one with averages. I'd personally prefer a one that shows info on every single chip.

Hi Isokivi, I'd be happy to give this a go over the weekend (we do a reasonable amount of parsing of text files in bash, so it should be straightforward). However, we don't have any bitfury hardware yet (October can't come round soon enough ), so would need some sample log files and some info on the file locations etc. Let me know if you're interested - somebody else may have done it already of course and may be able to share more quickly.

[spiccioli]

What would be awesome is if someone whipped up a script that can collect larger samples of data by picking up each individual 5 min log and compiling them in to a longer one with averages. I'd personally prefer a one that shows info on every single chip.

Either this or an official statement about the need of fans and/or the temperature range which chips can withstand.

spiccioli

[Isokivi]

What would be awesome is if someone whipped up a script that can collect larger samples of data by picking up each individual 5 min log and compiling them in to a longer one with averages. I'd personally prefer a one that shows info on every single chip.

Hi Isokivi, I'd be happy to give this a go over the weekend (we do a reasonable amount of parsing of text files in bash, so it should be straightforward). However, we don't have any bitfury hardware yet (October can't come round soon enough ), so would need some sample log files and some info on the file locations etc. Let me know if you're interested - somebody else may have done it already of course and may be able to share more quickly.

The file we want is /run/smh/.stat.log

This is what they look like: http://pastebin.com/xN7t9WaH

Please note that: pastebin screws up the line lenght, all the info regarding a single chip is actually on one line.
And secondly, this is a sample of that has two h-boards, the maximum is sixteen.

Let me know if theres something else I can help with.
[edit] btw since theres propably no nice way to time the data collection exactly when the new logs are generated (or perhaps there is, Im out of familiar waters here) I would suggest that you have the script pick up data every 330 seconds (5min 30 sec).

[Ollie]

One of my boards drops to ~10ghs
The chips turn off...anyone got any suggestions on getting a bit more out of it ?

Is it worth turning auto-tune off on those chips and setting them to 52 or something ?

I am having the same problem. Initially I thought that it was a problem with Bus A, so moved my cards to buses B and C, but the problem repeated. I experimented with turning autotune off and hand tuning frequencies, and it has been hashing ok for about an hour now.

[darkfriend77]

What would be awesome is if someone whipped up a script that can collect larger samples of data by picking up each individual 5 min log and compiling them in to a longer one with averages. I'd personally prefer a one that shows info on every single chip.

Hi Isokivi, I'd be happy to give this a go over the weekend (we do a reasonable amount of parsing of text files in bash, so it should be straightforward). However, we don't have any bitfury hardware yet (October can't come round soon enough ), so would need some sample log files and some info on the file locations etc. Let me know if you're interested - somebody else may have done it already of course and may be able to share more quickly.

The file we want is /run/smh/.stat.log

This is what they look like: http://pastebin.com/xN7t9WaH

Please note that: pastebin screws up the line lenght, all the info regarding a single chip is actually on one line.
And secondly, this is a sample of that has two h-boards, the maximum is sixteen.

Let me know if theres something else I can help with.
[edit] btw since theres propably no nice way to time the data collection exactly when the new logs are generated (or perhaps there is, Im out of familiar waters here) I would suggest that you have the script pick up data every 330 seconds (5min 30 sec).

Maybee we should just try to add our stuff into ... the chainminer ...

https://github.com/bfsb/chainminer

We could do a other view ... in the webGui which shows each chip and his performace ... with his stats ...

... information is already been collected in the software for each chip ...

[jlsminingcorp]

The file we want is /run/smh/.stat.log

This is what they look like: http://pastebin.com/xN7t9WaH

Please note that: pastebin screws up the line lenght, all the info regarding a single chip is actually on one line.
And secondly, this is a sample of that has two h-boards, the maximum is sixteen.

Let me know if theres something else I can help with.
[edit] btw since theres propably no nice way to time the data collection exactly when the new logs are generated (or perhaps there is, Im out of familiar waters here) I would suggest that you have the script pick up data every 330 seconds (5min 30 sec).

OK, looks good. When copied over to one of my linux boxes it looks like this (http://s15.postimg.org/z1kuze2l7/bfminer_log.jpg). I presume that there are no lines of whitespace at the beginning or end of the file. Does that look right?

I was thinking that it might be better to have the script running as a cron job rather than building in a certain waiting period before sampling into the script itself, but we can do it the other way if you like. I can get the script to check the timestamp of the log file and sleep for 30 secs or so if it's the same as the previous data point in any case, so that data isn't double counted.

Any preference on file format (was thinking comma separated text for easy input into excel - xml should be possible, but might take longer to get working)? Do you have any particular data headings in mind for the consolidated output?

[Isokivi]

What would be awesome is if someone whipped up a script that can collect larger samples of data by picking up each individual 5 min log and compiling them in to a longer one with averages. I'd personally prefer a one that shows info on every single chip.

Either this or an official statement about the need of fans and/or the temperature range which chips can withstand.

spiccioli

I have officially (punin said this) heard over irc that a chip was tested to hash untill the solder melted and it fell. The chip in question was heated with a soldering iron, while hashing.

[jlsminingcorp]

Maybee we should just try to add our stuff into ... the chainminer ...

https://github.com/bfsb/chainminer

We could do a other view ... in the webGui which shows each chip and his performace ... with his stats ...

... information is already been collected in the software for each chip ...

Yes, that's definitely a better option in the long term. It would certainly take me a lot longer than a quick text-file-parsing bash script, but there are much better coders than me in bitcoinland, so they may be able to implement this more quickly.