[GUIDE] BitFury Miner Support/Tuning

[greaterninja]

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.

[klondike_bar]

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.

just tweaked mine today.

BEFORE: 1.130K at cold state = 0.788 V    ------> unknown hot state resistance = 0.808 V     ---> 39.5GH average
NOW:     1.100K at cold state = 0.815 V   ------> unknown hot state resistance = 0.834 V     ----> 42.0GH average (based on 2.5hours)

I am using small chip-sized heatsink on 14/16 chips (i need more lol - and its been suggested that backside cooling may be optimal) and have 5 1"x1" heatsinks on the back, centered over the 4 capacitor clusters and in the center of the board. a 120mm fan is aimed at the back and the front is cooled by a 60mm and an 80mm fan on opposite sides

I don't intend to push further until I know that:
a) the chips can handle a certain amount of heat (if they can handle 70% of what asicminer does, then ive got a lot of headroom)
b) what the 30A converter can handle and its failure mode(s) (intron suggested that they simply switch off at overvolt/overheat)

The heat issue seems minimal so long as the chips can handle some warth and voltage. The issue lies in the 30A limit. If my 'math' is right, the chips use around 0.8-0.9w/GH at 2.65GH/s * 16 chips * 0.835 V = 28.3-31.8 A. The other components may draw additional power on top of this...?

[goxed]

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.

Here. Resistor at 1.1KOhms, along with heatsinks for each hashing chip

Code:

0:	811	37.781	38.463	2639	81	0	0	16	0	0	(2.361/chip)	97%
1:	871	37.223	37.892	2600	83	0	0	16	0	0	(2.326/chip)	99%

[cscape]

When you reverse a board in the M-BOARD connector, it doesn't swap GND with 12V wires, but it swaps the GND/12V with the 1.8V SPI section on the other side, and it puts the 12V straight into the chips. Especially the first and last chip in the chain will get the full effect, and are the most likely to fry.  The power regulator may still be okay. The fuse is probably still good too.

[-Redacted-]

It's nice of the chips to sacrifice themselves to protect the fuse like that....  

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....

[zurg]

It's nice of the chips to sacrifice themselves to protect the fuse like that....  

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....

Can anyone assist me with figuring out what traces to cut for bypassing #25? U4B chip I assume.

1       AIfDSo  55      1.990   2.040   139     6       0       0       193     [0:0]   0       7 9 9 9 9 9 9 8 7 9 10 9 9 9 9 8        2 0 0 $
2       AIfDSo  55      1.704   1.955   119     17      0       0       185     [0:1]   0       7 7 8 8 8 5 8 5 6 7 9 9 9 8 8 7         1 1 0 $
3       AIfDSo  55      1.718   1.966   120     4       0       0       186     [0:2]   0       8 8 8 8 8 8 8 7 7 7 7 7 5 8 8 8         0 0 0 $
4       AIfDSo  55      1.890   1.977   132     4       0       0       187     [0:3]   0       7 7 8 8 8 9 9 9 9 9 9 9 9 8 8 6         1 1 0 $
5       AIfDSo  55      1.875   1.934   131     7       0       0       183     [0:4]   0       9 9 8 9 9 9 9 9 9 6 7 7 7 7 8 9         0 0 1 $
6       AIfDSo  55      2.076   1.977   145     2       0       0       187     [0:5]   0       10 9 9 9 9 9 9 9 9 9 8 9 9 9 10 9       0 0 0 $
7       AIfDSo  55      2.047   2.008   143     8       0       0       190     [0:6]   0       7 8 9 9 9 9 10 9 9 10 9 9 9 9 9 9       2 1 0 $
8       AIfDSo  55      1.646   1.955   115     6       0       0       185     [0:7]   36      7 7 7 6 6 7 7 8 8 8 6 8 8 7 7 8         0 0 0 $
9       AIfDSo  55      1.947   2.114   136     15      0       0       200     [0:8]   0       9 8 10 10 9 9 9 9 9 8 8 8 8 6 7 9       0 1 0 $
10      AIfDSo  55      1.990   2.061   139     5       0       0       195     [0:9]   0       8 9 9 9 8 9 9 9 8 8 9 9 9 9 9 8         1 0 0 $
11      AIfDSo  55      1.747   2.029   122     8       0       0       192     [0:A]   0       8 8 7 7 7 8 8 7 8 7 9 9 8 6 7 8         0 0 1 $
12      AIfDSo  55      1.790   1.945   125     13      0       0       184     [0:B]   0       8 8 9 7 8 8 9 8 9 8 9 8 7 7 6 6         0 0 0 $
13      AIfDSo  55      1.976   1.955   138     9       0       0       185     [0:C]   0       7 9 8 9 9 8 8 9 9 8 10 10 9 8 9 8       2 0 1 $
14      AIfDSo  55      2.019   2.019   141     2       0       0       191     [0:D]   0       9 9 9 9 9 9 9 9 9 9 8 9 8 8 9 9         0 0 0 $
15      AIfDSo  55      1.918   2.029   134     5       0       0       192     [0:E]   0       8 9 8 9 9 8 8 8 8 8 8 9 8 9 8 9         1 0 1 $
16      AIfDSo  55      1.976   2.029   138     6       0       0       192     [0:F]   0       9 9 9 9 8 8 9 9 8 9 9 9 8 8 8 9         0 0 0 $
17      AIfDSo  55      2.219   2.146   155     4       0       0       203     [1:0]   0       10 10 9 9 10 10 10 10 10 10 9 10 10 10 8 10   $
18      AIfDSo  55      2.319   2.315   162     8       0       0       219     [1:1]   0       11 10 10 11 10 10 10 9 9 9 11 10 11 10 10 11  $
19      AIfDSo  55      1.704   2.082   119     12      0       0       197     [1:2]   0       9 9 7 8 8 8 6 6 7 7 8 8 8 6 7 7         0 0 1 $
20      AIfDSo  55      2.205   2.124   154     5       0       0       201     [1:3]   0       9 8 10 10 10 10 10 9 9 9 10 10 10 10 10 10    $
21      AIfDSo  55      2.477   2.188   173     5       0       0       207     [1:4]   36      11 10 11 11 11 11 11 11 10 12 12 11 11 10 11 9$
22      AIfDSo  55      1.833   1.860   128     15      0       0       176     [1:5]   1       9 8 9 8 7 9 8 8 7 9 9 8 7 7 9 6         0 1 0 $
23      AIfDSo  55      1.833   2.135   128     11      0       0       202     [1:6]   36      9 8 8 9 9 9 8 8 8 9 6 8 7 7 7 8         0 1 1 $
24      AIfDSo  55      1.890   2.135   132     7       0       0       202     [1:7]   0       8 9 9 9 9 8 7 9 8 8 7 8 8 8 8 9         1 0 0 $
25      Aifdso  55      0.000   0.000   0       0       0       0       0       [1:8]   756     0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0         0 0 0 $
speed:1320 noncerate[GH/s]:46.787 (1.871/chip) hashrate[GH/s]:48.980 good:3268 errors:184 spi-err:0 miso-err:0 jobs:376 cores:95% good:24 bad:$
0:      880     30.308  31.994  2117    117     0       0       16      0       0       (1.894/chip)    100%
1:      440     16.478  16.985  1151    67      0       0       8       0       1   

[Cablez]

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.  

[jlsminingcorp]

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.  

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)?

[jlsminingcorp]

It's nice of the chips to sacrifice themselves to protect the fuse like that....  

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 !

[Cablez]

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.  

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.

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.

[jlsminingcorp]

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.  

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.

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.

[zurg]

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

[klondike_bar]

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.  

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.

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

[klondike_bar]

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

[Micky25]

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.

[klondike_bar]

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

[klondike_bar]

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

[Keefe]

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.

[klondike_bar]

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

[zurg]

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/