Pages:
Author

Topic: [GUIDE] BitFury Miner Support/Tuning - page 28. (Read 148079 times)

hero member
Activity: 574
Merit: 501
October 08, 2013, 09:18:05 PM
Nice Spotts version 1 case there....  Thanks for the photo!

Might try turning down the humidity some - there's stuff that looks kind of like like mold growing out of the tops of all your chips....   Cheesy

Here's one of my plain-jane Bitfury rigs.  No heatsinks, nothing much but a little USB fan blowing on the h-boards.  Been running for weeks now, like a Duracell Bunny.


legendary
Activity: 974
Merit: 1000
October 08, 2013, 03:43:43 PM
poser!  Cheesy
sr. member
Activity: 434
Merit: 265
October 08, 2013, 03:02:13 PM



legendary
Activity: 1513
Merit: 1040
October 08, 2013, 11:05:09 AM
is there a way to get longer sampls than 5 minutes? would be fine for tuning.
sr. member
Activity: 658
Merit: 250
October 08, 2013, 10:24:30 AM
I've been running the bitfury branch of BFGMiner for about a week now, with no major problems. Performance is identical to chainminer (30G + 30G + 35G accepted hashrate from 3 boards), but sometimes the hashrate (and power consumption) randomly drops 1%. This is fixed by restarting. I have no reason to use chainminer anymore, except to maybe fine-tune new boards as they come in.
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
October 08, 2013, 07:06:23 AM
Had to hard reset.

These things require so much babysitting it aint funny anymore/

I thought that, but then i backed off a bit on the tickering and let it do its thing. After about 6hrs to fully warm up and reach peak voltage, Ive been stable for 48hrs at 39.9GH/s average (+/- 2GH on the hourly reading)
hero member
Activity: 826
Merit: 500
Crypto Investor ;) @ Farmed Account Hunter
October 08, 2013, 06:10:08 AM
Great, woke up to 0Gh/s....

Drove to work (when I was off), yeah my luck.

My Pi was showing:



Had to hard reset.

These things require so much babysitting it aint funny anymore/
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
October 07, 2013, 12:58:56 PM
can someone explain the "coreOK" stat better for me? I sometimes see chips with this as 0 or <10, and other times they are all in the hundreds.

When i start the miner, all chips report 756. within 5 minutes they are all under 700

Each chip has 756 cores. The software tracks which cores have found good nonces. You can see the full details in /tmp/.core.log

The numbers you're looking at are actually how many cores have *not* yet found a good nonce. Over time, if all cores are good, that number will approach 0.

oh terrific, thanks. I was getting concerned to see some chips hitting '0' for this, and letting the crazy in my head associate it with temporary hashrate drops at the pool or chips switching off to 0 noncerates.

->  The chips turning off was fixed by a few more voltage tweaks - 0.825V across the caps seems to be the best, with tuning at 54 on 15 chips, and chip 1 at 53, I have a stable 38-42GH/s hourly rate (24hr average of ~39.85 GH/s).

At 0.840V i was getting about 41.5GH/s, but a chip turning off now and then bringing me down to 39GH. Not sure if this was a temp or voltage issue, but ive got some bigger 35x35x6mm heatsinks coming that i plan to strap 4 of on each board
hero member
Activity: 681
Merit: 500
October 07, 2013, 02:32:28 AM
can someone explain the "coreOK" stat better for me? I sometimes see chips with this as 0 or <10, and other times they are all in the hundreds.

When i start the miner, all chips report 756. within 5 minutes they are all under 700

Each chip has 756 cores. The software tracks which cores have found good nonces. You can see the full details in /tmp/.core.log

The numbers you're looking at are actually how many cores have *not* yet found a good nonce. Over time, if all cores are good, that number will approach 0.
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
October 02, 2013, 05:27:25 PM
can someone explain the "coreOK" stat better for me? I sometimes see chips with this as 0 or <10, and other times they are all in the hundreds.

When i start the miner, all chips report 756. within 5 minutes they are all under 700
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
October 02, 2013, 04:33:13 PM
I'm still totally uncertain if I want to use these guys or not: http://www.amazon.de/gp/product/B004BU6ITM/

I'm scared of heat congestion between the boards getting worse. Just blowing the heat away softly through the naked gaps will help the best I think.

id make sure to use a heatsink less than 10mm tall. 6mm would be better
legendary
Activity: 974
Merit: 1000
October 02, 2013, 04:16:10 PM
I'm still totally uncertain if I want to use these guys or not: http://www.amazon.de/gp/product/B004BU6ITM/

I'm scared of heat congestion between the boards getting worse. Just blowing the heat away softly through the naked gaps will help the best I think.
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
October 02, 2013, 02:42:50 PM
what is the spacing on the heatsink mount holes of the h-board? It looks like they may be 37.5mm or so width&length
legendary
Activity: 2128
Merit: 1005
ASIC Wannabe
October 02, 2013, 09:22:40 AM
Guys, I want to know who has the highest clock rates.   Does anyone have over 40GH/s per card or close to it?
 I was able to get about 43GH/s for 1 card.  Most cards run ~35-36GH/s though.  one runs 31 GH/s
Please post your epic results and if you have 35+ please post your resistor settings, if you used heatsinks and what other cooling you have.

It seems to be really board dependent but with a general resistor setting of about 1.15k I see anywhere from 30Gh/s up to 40. I am not using any heatsinks right now just 3x120mm fans blowing at them.

Cooling seems to be the factor for stability at the higher voltages as I have had stable boards start to lose hashrate and go to 0Gh/s and back again when temps go up, and even when they don't.  Wink

Hi All, which components seem to need the most cooling (obviously the chips need cooling, but are there other hotspots?) Is this loss of hashrate related to the voltage regulator overheating and shutting down (I think I remember reading somewhere that this could protect itself in this way)? If so, is the voltage regulator designed to loose heat through its top surface or through the board (as this will affect where we put a heatsink)?

I haven't been able to tell yet if the hashrate decreases because of the regulator or due to increased errors forcing the autotune to downclock the chips. I need to turn off autotune and see but considering the hardware cost I really don't play with it much.  Just keep hashing....just keep hashing. Grin

The regulator is very hot on the surface and not really hot at all on the backside. (finger test)  the chips are hot on both sides.

Great, thanks, so it sounds like it's worth us putting a heatsink on the top side of the regulator then (in addition to chip heatsinks). That's good to know.


I put 2 small heatsinks on my regulator and they get quite warm/hot weven with good airflow. chip voltage is around 0.834V once the resistor is up to temp (starts up from cold state at about 0.815V)

however, ive recently seen it go from 2-4hrs at 40+GH down to 20 for 5-15 minutes, then climb back again. autotune is off (most chips at 53/54) so i dont know if the board or the chips are behind the issue. voltage seems constant as far as i can tell
hero member
Activity: 826
Merit: 500
Crypto Investor ;) @ Farmed Account Hunter
October 02, 2013, 08:39:26 AM
Ugh.. great, my EOL card just went to all 0s.
Took 2 power disconnects to get it back working.
This is a weekly occurrence.

Now stat.log is thinking I have 171 chips on 2 cards. lol
hero member
Activity: 493
Merit: 500
Hooray for non-equilibrium thermodynamics!
October 02, 2013, 08:36:07 AM
Guys, I want to know who has the highest clock rates.   Does anyone have over 40GH/s per card or close to it?
 I was able to get about 43GH/s for 1 card.  Most cards run ~35-36GH/s though.  one runs 31 GH/s
Please post your epic results and if you have 35+ please post your resistor settings, if you used heatsinks and what other cooling you have.

It seems to be really board dependent but with a general resistor setting of about 1.15k I see anywhere from 30Gh/s up to 40. I am not using any heatsinks right now just 3x120mm fans blowing at them.

Cooling seems to be the factor for stability at the higher voltages as I have had stable boards start to lose hashrate and go to 0Gh/s and back again when temps go up, and even when they don't.  Wink

Hi All, which components seem to need the most cooling (obviously the chips need cooling, but are there other hotspots?) Is this loss of hashrate related to the voltage regulator overheating and shutting down (I think I remember reading somewhere that this could protect itself in this way)? If so, is the voltage regulator designed to loose heat through its top surface or through the board (as this will affect where we put a heatsink)?

I haven't been able to tell yet if the hashrate decreases because of the regulator or due to increased errors forcing the autotune to downclock the chips. I need to turn off autotune and see but considering the hardware cost I really don't play with it much.  Just keep hashing....just keep hashing. Grin

The regulator is very hot on the surface and not really hot at all on the backside. (finger test)  the chips are hot on both sides.

Great, thanks, so it sounds like it's worth us putting a heatsink on the top side of the regulator then (in addition to chip heatsinks). That's good to know.
legendary
Activity: 1400
Merit: 1000
I owe my soul to the Bitcoin code...
October 02, 2013, 07:59:36 AM
Guys, I want to know who has the highest clock rates.   Does anyone have over 40GH/s per card or close to it?
 I was able to get about 43GH/s for 1 card.  Most cards run ~35-36GH/s though.  one runs 31 GH/s
Please post your epic results and if you have 35+ please post your resistor settings, if you used heatsinks and what other cooling you have.

It seems to be really board dependent but with a general resistor setting of about 1.15k I see anywhere from 30Gh/s up to 40. I am not using any heatsinks right now just 3x120mm fans blowing at them.

Cooling seems to be the factor for stability at the higher voltages as I have had stable boards start to lose hashrate and go to 0Gh/s and back again when temps go up, and even when they don't.  Wink

Hi All, which components seem to need the most cooling (obviously the chips need cooling, but are there other hotspots?) Is this loss of hashrate related to the voltage regulator overheating and shutting down (I think I remember reading somewhere that this could protect itself in this way)? If so, is the voltage regulator designed to loose heat through its top surface or through the board (as this will affect where we put a heatsink)?

I haven't been able to tell yet if the hashrate decreases because of the regulator or due to increased errors forcing the autotune to downclock the chips. I need to turn off autotune and see but considering the hardware cost I really don't play with it much.  Just keep hashing....just keep hashing. Grin

The regulator is very hot on the surface and not really hot at all on the backside. (finger test)  the chips are hot on both sides.
hero member
Activity: 493
Merit: 500
Hooray for non-equilibrium thermodynamics!
October 02, 2013, 07:55:07 AM
It's nice of the chips to sacrifice themselves to protect the fuse like that....  Wink

A little power-grinder work, and you're back in business in no time, eh?  I personally would spend the time to cut traces, but I guess if you are in a really big hurry and all, like some folks around here, the grinder might be faster....

Using a grinder and pulling it off gets you the most respect though Wink!
hero member
Activity: 493
Merit: 500
Hooray for non-equilibrium thermodynamics!
October 02, 2013, 07:51:48 AM
Guys, I want to know who has the highest clock rates.   Does anyone have over 40GH/s per card or close to it?
 I was able to get about 43GH/s for 1 card.  Most cards run ~35-36GH/s though.  one runs 31 GH/s
Please post your epic results and if you have 35+ please post your resistor settings, if you used heatsinks and what other cooling you have.

It seems to be really board dependent but with a general resistor setting of about 1.15k I see anywhere from 30Gh/s up to 40. I am not using any heatsinks right now just 3x120mm fans blowing at them.

Cooling seems to be the factor for stability at the higher voltages as I have had stable boards start to lose hashrate and go to 0Gh/s and back again when temps go up, and even when they don't.  Wink

Hi All, which components seem to need the most cooling (obviously the chips need cooling, but are there other hotspots?) Is this loss of hashrate related to the voltage regulator overheating and shutting down (I think I remember reading somewhere that this could protect itself in this way)? If so, is the voltage regulator designed to loose heat through its top surface or through the board (as this will affect where we put a heatsink)?
legendary
Activity: 1400
Merit: 1000
I owe my soul to the Bitcoin code...
October 02, 2013, 07:09:52 AM
Guys, I want to know who has the highest clock rates.   Does anyone have over 40GH/s per card or close to it?
 I was able to get about 43GH/s for 1 card.  Most cards run ~35-36GH/s though.  one runs 31 GH/s
Please post your epic results and if you have 35+ please post your resistor settings, if you used heatsinks and what other cooling you have.

It seems to be really board dependent but with a general resistor setting of about 1.15k I see anywhere from 30Gh/s up to 40. I am not using any heatsinks right now just 3x120mm fans blowing at them.

Cooling seems to be the factor for stability at the higher voltages as I have had stable boards start to lose hashrate and go to 0Gh/s and back again when temps go up, and even when they don't.  Wink
Pages:
Jump to: