Blue Fury Support Thread.

[monsieurmarc]

Hi Beastly,

As mentioned earlier mine is pootling around at about 1.9ghash

I suspect the variation is all the outputs seen is the tolerance on your resistors.

The 223's I believe typically have a 5% tolerance

Regards
 Marc

[Beastlymac]

Hi Beastly,

As mentioned earlier mine is pootling around at about 1.9ghash

I suspect the variation is all the outputs seen is the tolerance on your resistors.

The 223's I believe typically have a 5% tolerance

Regards
 Marc

It could be a resistor issue. We are considering changing the resistor for the next batch. But we wanted to keep the units in a "safe zone" so as to prolong the life of the device and stop and potential issues that may be caused by straining the chip.

What mining software have you tried and what OS is your host computer?

[monsieurmarc]

At the moment Win7 and have tried bfg 3.2 and all later versions and see no hw errors on 3.2 and a lot on 3.6
Tonight I will put it onto my Pi at home and see how it does.

Zero hw errors is a nice thing to see :-)

[HellDiverUK]

Change the resistor and put a decent heat sink on.  

The one supplied with my two units are pathetic, and even with a fan rammed up against them, they're running at holyomgshithot temps at 2.4GH.  The ones on the RedFurys looks like they could cope, but the heat sink on my Blue Fury is more like something I'd put on a 7805.

[Beastlymac]

Change the resistor and put a decent heat sink on.  

The one supplied with my two units are pathetic, and even with a fan rammed up against them, they're running at holyomgshithot temps at 2.4GH.  The ones on the RedFurys looks like they could cope, but the heat sink on my Blue Fury is more like something I'd put on a 7805.

Thanks we will take that into consideration for the next batch. It is my understanding that the heatsinks where sourced at a rush because we didn't want to delay shipping any further.

At the moment Win7 and have tried bfg 3.2 and all later versions and see no hw errors on 3.2 and a lot on 3.6
Tonight I will put it onto my Pi at home and see how it does.

Zero hw errors is a nice thing to see :-)

Please try 3.6 as it gives a better understanding of the error rate. 3.2 ignores the errors and doesn't display them.

Can you please define lots

Thanks

[HellDiverUK]

Blah, feel free to feel the way you feel about what I post. In all honesty, I couldn't care less. It's not helping anyone either, but I hope you got your daily portion of "let's be negative at random forum people"
You're better off trying to help all the people who are not getting what they've paid for.

I think if you read my posts, you'll find a LOT of my posts are indeed help.  Either directly to other users, or with developers in bug reports.  Only a small proportion are 'STFU dumbass' posts.  Those posts are reserved for a select few, FYI.

[HellDiverUK]

Thanks we will take that into consideration for the next batch. It is my understanding that the heatsinks where sourced at a rush because we didn't want to delay shipping any further.

I just remembered I have a temp probe.  Would you like me to measure the temps and report back?  I could do that later on this evening if you'd like.

[monsieurmarc]

Change the resistor and put a decent heat sink on.  

The one supplied with my two units are pathetic, and even with a fan rammed up against them, they're running at holyomgshithot temps at 2.4GH.  The ones on the RedFurys looks like they could cope, but the heat sink on my Blue Fury is more like something I'd put on a 7805.

Thanks we will take that into consideration for the next batch. It is my understanding that the heatsinks where sourced at a rush because we didn't want to delay shipping any further.

At the moment Win7 and have tried bfg 3.2 and all later versions and see no hw errors on 3.2 and a lot on 3.6
Tonight I will put it onto my Pi at home and see how it does.

Zero hw errors is a nice thing to see :-)

Please try 3.6 as it gives a better understanding of the error rate. 3.2 ignores the errors and doesn't display them.

Can you please define lots

Thanks

Ok thanks I will do that

[Mudbankkeith]

My Pi is giving an average of 2.33 on MinePeon, with an average of 2.24 at the pool. No special mods, no special drivers, just:-    -S BPM:all

Reject rate is 0.81%    HW Error rate is 59.23%

simple maths says this is not a true error rate.

2.33Gh less 0.81% =  2.311Gh    The pool shows 2.24Gh so corrected HW error rate must be 3.07%                 

The extra 56.16% is probably down to running a 5Gh chip at less than 2.5Gh.

So with 2.24Gh at the pool I am quite happy for now.


In short ignore the HW error rate, Its not true.  The actual figure reported at the pool is the one that pays you.

[Beastlymac]

Thanks we will take that into consideration for the next batch. It is my understanding that the heatsinks where sourced at a rush because we didn't want to delay shipping any further.

I just remembered I have a temp probe.  Would you like me to measure the temps and report back?  I could do that later on this evening if you'd like.

I don't have the equipment to accurately measure temperature, sorry.

My Pi is giving an average of 2.33 on MinePeon, with an average of 2.24 at the pool. No special mods, no special drivers, just:-    -S BPM:all

Reject rate is 0.81%    HW Error rate is 59.23%

simple maths says this is not a true error rate.

2.33Gh less 0.81% =  2.311Gh    The pool shows 2.24Gh so corrected HW error rate must be 3.07%                 

The extra 56.16% is probably down to running a 5Gh chip at less than 2.5Gh.

So with 2.24Gh at the pool I am quite happy for now.


In short ignore the HW error rate, Its not true.  The actual figure reported at the pool is the one that pays you.

Sounds like a old version of bfgminer?
Try a git pull

[HellDiverUK]

Thanks we will take that into consideration for the next batch. It is my understanding that the heatsinks where sourced at a rush because we didn't want to delay shipping any further.

I just remembered I have a temp probe.  Would you like me to measure the temps and report back?  I could do that later on this evening if you'd like.

I don't have the equipment to accurately measure temperature, sorry.

Read what I said again...

I'm offering to report temps to you.  So you can see what temps real world units are reaching when overclocking using the bent piece of foil heat sink.

[Mudbankkeith]

Thanks we will take that into consideration for the next batch. It is my understanding that the heatsinks where sourced at a rush because we didn't want to delay shipping any further.

I just remembered I have a temp probe.  Would you like me to measure the temps and report back?  I could do that later on this evening if you'd like.

I don't have the equipment to accurately measure temperature, sorry.

My Pi is giving an average of 2.33 on MinePeon, with an average of 2.24 at the pool. No special mods, no special drivers, just:-    -S BPM:all

Reject rate is 0.81%    HW Error rate is 59.23%

simple maths says this is not a true error rate.

2.33Gh less 0.81% =  2.311Gh    The pool shows 2.24Gh so corrected HW error rate must be 3.07%                 

The extra 56.16% is probably down to running a 5Gh chip at less than 2.5Gh.

So with 2.24Gh at the pool I am quite happy for now.


In short ignore the HW error rate, Its not true.  The actual figure reported at the pool is the one that pays you.

Sounds like a old version of bfgminer?
Try a git pull

Git pull was done saturday.     BFGMiner 3.4.0. Peon 0.2.4       I will do another later.     Version 3.6.0 is now out.

Unless it makes the device hash faster does it matter?    HW error is just a number.    Luke could program it to read 0.00 all the time if he was inclined.

[Beastlymac]

Thanks we will take that into consideration for the next batch. It is my understanding that the heatsinks where sourced at a rush because we didn't want to delay shipping any further.

I just remembered I have a temp probe.  Would you like me to measure the temps and report back?  I could do that later on this evening if you'd like.

I don't have the equipment to accurately measure temperature, sorry.

Read what I said again...

I'm offering to report temps to you.  So you can see what temps real world units are reaching when overclocking using the bent piece of foil heat sink.

Sorry my bad, yes if you can report temperatures to me that would be a great help!

Thanks

Thanks we will take that into consideration for the next batch. It is my understanding that the heatsinks where sourced at a rush because we didn't want to delay shipping any further.

I just remembered I have a temp probe.  Would you like me to measure the temps and report back?  I could do that later on this evening if you'd like.

I don't have the equipment to accurately measure temperature, sorry.

My Pi is giving an average of 2.33 on MinePeon, with an average of 2.24 at the pool. No special mods, no special drivers, just:-    -S BPM:all

Reject rate is 0.81%    HW Error rate is 59.23%

simple maths says this is not a true error rate.

2.33Gh less 0.81% =  2.311Gh    The pool shows 2.24Gh so corrected HW error rate must be 3.07%                 

The extra 56.16% is probably down to running a 5Gh chip at less than 2.5Gh.

So with 2.24Gh at the pool I am quite happy for now.


In short ignore the HW error rate, Its not true.  The actual figure reported at the pool is the one that pays you.

Sounds like a old version of bfgminer?
Try a git pull

Git pull was done saturday.     BFGMiner 3.4.0. Peon 0.2.4       I will do another later.     Version 3.6.0 is now out.

Unless it makes the device hash faster does it matter?    HW error is just a number.    Luke could program it to read 0.00 all the time if he was inclined.

It just gives an accurate hw error %
But if it doesn't worry you then it is ok.

[Mudbankkeith]

My Pi is giving an average of 2.33 on MinePeon, with an average of 2.24 at the pool. No special mods, no special drivers, just:-    -S BPM:all

Reject rate is 0.81%    HW Error rate is 59.23%     simple maths says this is not a true error rate.

Peon shows 2.33Gh less 0.81% =  2.311Gh    The pool shows 2.24Gh so corrected HW error rate must be 3.07%                 

The extra 56.16% is probably down to running a 5Gh chip at less than 2.5Gh.

Git pull was done 7th nov     BFGMiner 3.4.0. Peon 0.2.4       (I will do another later.     Version 3.6.0 is now out.)



Edit:-  New git pull                BFGMiner 3.5.1  Peon 0.2.4.2

         Reject rate is now  1.47%    HW Error rate is 4.39%

         Peon shows 2.31Gh less 1.47% =  2.27Gh    The pool shows 2.16Gh so corrected HW error rate must be 4.85% 

Reject rate is up, HW error rate has dropped,  but so has my pool rate   

The moral is:- do not mess to much

[Beastlymac]

It also shows how different mining software produces different results and you can never solve an issue for everyone without creating more problems in the process. You can always revert back the version by re flashing the sd card. It seems that people experience drastically different results based on what setup they have.

[Mudbankkeith]

It also shows how different mining software produces different results and you can never solve an issue for everyone without creating more problems in the process. You can always revert back the version by re flashing the sd card. It seems that people experience drastically different results based on what setup they have.

No problem, I always flash 2 sd cards, one with the original image and one for the git pull update that almost always arrives 2 days later, you can never be to careful.
I also keep the disc image on my NAS server.(I think there are currently about 10 versions there)

Latest working in the Pi, Clean spare of latest download and image backup.(for when Sh** happens). SD cards do fail.

[HellDiverUK]

OK, just run tests.  Miners were running for 5 minutes - the one I tested was running at 2.5GH.

Temp on top of the BitFury chip:  95C



Temp at middle of heat sink:  89C




That's burning hot - probably a safety issue.

The Block Erupter in the same hub was running at 69C, again probably a safety issue, but not warm enough to almost boil water.

[zac2013]

I have been running 9 blue furries from outcast since Saturday. They have been running ok with an average 20ghs for 9 off - 2.22 average). I found that one kept dieing on me - displaying "dead" in bfgminer.

Coincidently, i changed the psu and fans because my old setup was too loud. I ending up putting two Sharkoon Silent Eagle 1000 Fan's (120x120x25) directly on top of the blue furries (vertical airflow) (as opposed to standard desktop fan's pointing straight at them) and what may be of interest to peeps is that they became more stable (no "dead") error's and my hash rate went up a little bit.

So as helldiveruk has pointed out, it may be best to cool these bad boys, as best you can.


My setup:
bfgminer as per the setup instructions at the start of this thread.
win7
running block erupters at the same time

[tk1337]

Well... my 10k resistors came in today, switched out the 22k resistor for a 10k on one of my good units.

BFGMiner 3.6.0 - 2.82 GH/s - 0 HW Errors
CGMiner 3.8.0 - 2.51 GH/s

I only let it sit for about 10 minutes on each mining software, so going to let it run for awhile in BFGMiner now.

Before the installation of the resistor this unit was consistently at 2.2GH/s

EDIT:

Temperature at heat-sink: 27.8 C / 82 F
Temperature at chip: 37.7 C / 100 F

[btmtb]

It's my understanding that the hashrate meter is constantly being revised at ozcoin and it's been rather inaccurate of late. Check the actual number of shares match to confirm there's nothing going astray, or try it on another pool (for at least an hour).

Thanks for the tip, didn't realise that, swapped to Slush' pool now, will report back.

EARLIER RESULTS REPORTED BY POOL (Ozco.in)
BF1      1,294.84 MHash/s

RECENT REPORTED BY ALTERNATIVE POOL (Slush's)
BF1      2,249.389 MHash/s

CGMiner 3.8.1 summary output for the above session at Slush's:
BF1 0:                | 2.200G/2.184Gh/s | A:16170 R:100 HW:1111 WU:30.8/m

The number of shares reported by Slush's stacks up, so I think the earlier issue is Ozcoin's counters as you suggest ckolivas, cheers for the hint. Just to repeat my comments from my earlier post, this is with stock heatsink, no fan, and bus powered (for giggles). High error rate for my taste, but respectable throughput IMO considering the circumstances. I'm most keen to see if I can reduce the HW with sufficient cooling, will report back when I've had time to play properly this weekend.