Ultra Under-overclock image for A2 Innosilicon by Emdje - V5.0

[emdje]

Hi, looking at stat page I have a chip with a higher temperature, all the chips are at about 40° but one is at 62°, How can i solve the problem?

Thanks W_M

Your chips might need better cooling paste, sometimes it is not applied properly.
And get better cooling in general (cooler environment, better airflow).

[emdje]

Put the overvolted board on 1400 MHz. Has been running for over 22 hours and produced only 34 HW errors.
The other board at 1300 MHz produced 591 HW errors, and if I put that board on 1400 I effectively cripple it.

Wow, that's great! Might have to do the same mod, I have 1 88MH unit and 1 60MH unit. Good job!

How fat is the blade hashing at 1400mhz?

Thnx.
Hashing at about 17.35 Mhash/s with 1% HW error. Compared to 16% HW errors on the other blade at 1300 MHz.

[Prelude]

Cool, so an increase of about 1.4MH/s per blade can be expected. Have you measure what the increase in power consumption is?

[emdje]

Cool, so an increase of about 1.4MH/s per blade can be expected. Have you measure what the increase in power consumption is?

I have not yet measured the power consumption. When I have time (probably in the weekend) I will resolder the other board and then measure powerconsumption.

EDIT: Just saw my miner has been offline for the past 5 hours. Computer at home is not on so can't check anything until I get of from work, hopefully nothing wrong  

[Prelude]

Hope your rig is OK!

[emdje]

Hope your rig is OK!

Thnx. Just got home, luckily nothing wrong with the miner, was hashing normally again. From the logs I can see that both the main pool (miningrigrentals) and nicehash were not responding. So I assume it was an Internet issue. No overheating issues so that is good (overclocked board is 33 degrees Celsius)

Code:

[2015-02-05 17:36:06] Pool 0 stratum+tcp://eu-01.miningrigrentals.com:3333 not responding!                    
 [2015-02-05 17:36:06] Switching to pool 1 stratum+tcp://stratum.nicehash.com:3333                    
(5s):629.6K (avg):33.45Mh/s (pool):32.76Mh/s | A:87109888  R:0  HW:1307  WU:29996.0/m
(5s):370.6K (avg):33.45Mh/s (pool):32.76Mh/s | A:87109888  R:0  HW:1307  WU:29994.9/m
(5s):218.1K (avg):33.45Mh/s (pool):32.76Mh/s | A:87109888  R:0  HW:1307  WU:29993.9/m
(5s):128.4K (avg):33.45Mh/s (pool):32.76Mh/s | A:87109888  R:0  HW:1307  WU:29992.9/m
(5s):75.58K (avg):33.44Mh/s (pool):32.76Mh/s | A:87109888  R:0  HW:1307  WU:29991.8/m
(5s):44.49K (avg):33.44Mh/s (pool):32.76Mh/s | A:87109888  R:0  HW:1307  WU:29990.8/m
 [2015-02-05 17:36:46] Pool 1 stratum+tcp://stratum.nicehash.com:3333 not responding!                    
(5s):26.19K (avg):33.44Mh/s (pool):32.76Mh/s | A:87109888  R:0  HW:1307  WU:29989.8/m
(5s):15.41K (avg):33.44Mh/s (pool):32.76Mh/s | A:87109888  R:0  HW:1307  WU:29988.7/m
(5s):9.073K (avg):33.44Mh/s (pool):32.75Mh/s | A:87109888  R:0  HW:1307  WU:29987.7/m
(5s):5.340K (avg):33.44Mh/s (pool):32.75Mh/s | A:87109888  R:0  HW:1307  WU:29986.7/m
(5s):3.143K (avg):33.44Mh/s (pool):32.75Mh/s | A:87109888  R:0  HW:1307  WU:29985.6/m
(5s):1.850K (avg):33.44Mh/s (pool):32.75Mh/s | A:87109888  R:0  HW:1307  WU:29984.6/m
(5s):1.089K (avg):33.44Mh/s (pool):32.75Mh/s | A:87109888  R:0  HW:1307  WU:29983.6/m
(5s):641.0 (avg):33.43Mh/s (pool):32.75Mh/s | A:87109888  R:0  HW:1307  WU:29982.5/m
(5s):377.0 (avg):33.43Mh/s (pool):32.75Mh/s | A:87109888  R:0  HW:1307  WU:29981.5/m
(5s):222.0 (avg):33.43Mh/s (pool):32.75Mh/s | A:87109888  R:0  HW:1307  WU:29980.5/m
(5s):130.0 (avg):33.43Mh/s (pool):32.74Mh/s | A:87109888  R:0  HW:1307  WU:29979.4/m
(5s):76.00 (avg):33.43Mh/s (pool):32.74Mh/s | A:87109888  R:0  HW:1307  WU:29978.4/m
(5s):45.00 (avg):33.43Mh/s (pool):32.74Mh/s | A:87109888  R:0  HW:1307  WU:29977.4/m
(5s):26.00 (avg):33.43Mh/s (pool):32.74Mh/s | A:87109888  R:0  HW:1307  WU:29976.4/m
(5s):15.00 (avg):33.43Mh/s (pool):32.74Mh/s | A:87109888  R:0  HW:1307  WU:29975.3/m
(5s):9.000 (avg):33.42Mh/s (pool):32.74Mh/s | A:87109888  R:0  HW:1307  WU:29974.3/m
(5s):5.000 (avg):33.42Mh/s (pool):32.74Mh/s | A:87109888  R:0  HW:1307  WU:29973.3/m
(5s):3.000 (avg):33.42Mh/s (pool):32.74Mh/s | A:87109888  R:0  HW:1307  WU:29972.2/m
(5s):1.000 (avg):33.42Mh/s (pool):32.74Mh/s | A:87109888  R:0  HW:1307  WU:29971.2/m
(5s):1.000 (avg):33.42Mh/s (pool):32.73Mh/s | A:87109888  R:0  HW:1307  WU:29970.2/m
(5s):0.000 (avg):33.42Mh/s (pool):32.73Mh/s | A:87109888  R:0  HW:1307  WU:29969.1/m
................................................
................................................
(5s):0.000 (avg):29.43Mh/s (pool):28.82Mh/s | A:87109888  R:0  HW:1307  WU:26387.7/m
(5s):0.000 (avg):29.43Mh/s (pool):28.82Mh/s | A:87109888  R:0  HW:1307  WU:26386.9/m
(5s):0.000 (avg):29.42Mh/s (pool):28.82Mh/s | A:87109888  R:0  HW:1307  WU:26386.1/m
 [2015-02-06 00:13:46] Pool 1 difficulty changed to 20                                       
 [2015-02-06 00:13:46] Accepted untracked stratum share from pool 1                    
 [2015-02-06 00:13:51] Network diff set to 16.1K                                                             
 [2015-02-06 00:13:51] New block detected on network before longpoll                                       
 [2015-02-06 00:13:51] Work available from pools, resuming.                    
(5s):14.63K (avg):29.42Mh/s (pool):28.82Mh/s | A:87109908  R:0  HW:1307  WU:26385.4/m
 [2015-02-06 00:13:51] Accepted 59751ef4 Diff 732/20 BA1 0 pool 1                                       
 [2015-02-06 00:13:51] Stratum from pool 1 requested work restart                    
 [2015-02-06 00:13:52] Accepted 01d0143d Diff 141/20 BA1 0 pool 1                                       
 [2015-02-06 00:13:52] Accepted 01699234 Diff 181/20 BA1 0 pool 1                                       
 [2015-02-06 00:13:52] Accepted 01e19c07 Diff 136/20 BA1 1 pool 1                                       
 [2015-02-06 00:13:52] chip(cs0) 5: invalid nonce 0xf5ce1300                                                     
 [2015-02-06 00:13:52] Accepted fad8b4c1 Diff 261/20 BA1 1 pool 1                                       
 [2015-02-06 00:13:53] Accepted 01cbd5fd Diff 142/20 BA1 0 pool 1                                       
 [2015-02-06 00:13:54] Accepted 01f904fd Diff 129/20 BA1 0 pool 1     

[Prelude]

Good to hear! Seriously considering doing this with my 10 blades. Could you list the materials required by any chance? I'm decent with a soldering iron.

[emdje]

HOW TO VOLTAGEMOD YOUR BOARD:

The requirements are:
- Cooling paste (you need to take of the heatsinks otherwise you can't desolder the SMD resistor)
- Tweezers
- for each board; 4 x 12.7k resistor (high precision)
- soldering flux
- small solder
- thin tip soldering iron
- optional: anti static mat



First thing you do is take the blade out of the box it is in. Then you screw of the heatsinks, otherwise the heat from your soldering iron is transfered away from the resistor immediately and you won't be able to solder is off. Clean the board and the heatsinks from old cooling paste. If you have it use an anti static mat (I didn't).
Locate the voltage regulators:


Now locate the resistor you need to desolder:


The configuration of the circuit differs a little, but it is always the resistor next to the empty space and capacitor, marked '05c'.

Put some flux on the resistor, heat up both sides and use the soldering iron to push it off. It will most likely stick to the soldering iron so shake it of onto a peace of toilet paper or something that it does not bounce away. Resolder the two contact points with some new solder and put some flux on it.

Take a 12.7k resistor, fold it so that the two legs are about 1.5 mm apart. Cut them into the right length and rough up the legs so the solder adheres better. Solder the legs of the resistor.
Take the tweezers and hold the resistor on the two contact points. Use the thin tipped soldering iron to one by one secure the legs to the board.




After you have repeated this four times on each board, put new cooling paste on the chips and screw the heatsinks back on.


DISCLAMER: I am of course not liable if anything happens to you board, it is entirely at your own risk. Do first try it on one board and test it you before you do the rest of you boards.

If you want to use other values:
03c   0,828   volt   10,5   kiloOhm
05c   0,84   volt   11   kiloOhm
09c   0,88   volt   12,1   kiloOhm
   0,9   volt   12,375   
10c   0,902   volt   12,4   kiloOhm
11c   0,924 volt   12,7   kiloOhm
   0,93   volt   12,79   kiloOhm

[BakSAj]

Good job emdje! This would be useful 6-9 months ago, but now Im afraid it would actually lower the net profit of the miner due to increased power consumption. Did you measure watts already? Thanks.

[emdje]

Second board did not go as planned. I knew the risks when I started.
Board is not recognized anymore. Tomorrow I will troubleshoot the issue 

Code:

 [2015-02-10 06:46:24] Started cgminer 3.9.0
 [2015-02-10 06:46:24] Run Reset=1
 [2015-02-10 06:46:24] ST MCU hardware reset start
 [2015-02-10 06:46:28] SPI Speed 4000 kHz
 [2015-02-10 06:46:28] ST MCU - Enable (Pre-header)
 [2015-02-10 06:46:28] A1 = 1300,5
 [2015-02-10 06:46:28] A1 PLL Clock = 1300MHz
 [2015-02-10 06:46:28] A1 = 1400,0
 [2015-02-10 06:46:28] A1 PLL Clock = 1400MHz
 [2015-02-10 06:46:28] A1 = 1000,17
 [2015-02-10 06:46:28] A1 PLL Clock = 1000MHz
 [2015-02-10 06:46:28] A1 = 1000,17
 [2015-02-10 06:46:28] A1 PLL Clock = 1000MHz
 [2015-02-10 06:46:28] A1 = 1000,17
 [2015-02-10 06:46:28] A1 PLL Clock = 1000MHz
 [2015-02-10 06:46:28] A1 = 1000,17
 [2015-02-10 06:46:28] A1 PLL Clock = 1000MHz
 [2015-02-10 06:46:28] AUTO GPIO CS
--->[2015-02-10 06:46:29] Failure(cs0)(20): missing ACK for cmd 0x04
--->[2015-02-10 06:46:29] ACK(cs0) timeout:cmd_RESET_BCAST-1.2444s
 [2015-02-10 06:46:29] spidev0.0(cs1): Found 8 A1 chips
 [2015-02-10 06:46:29] Found chip 1 with 54 active cores
 [2015-02-10 06:46:29] Found chip 2 with 54 active cores
 [2015-02-10 06:46:29] Found chip 3 with 54 active cores
 [2015-02-10 06:46:29] Found chip 4 with 54 active cores
 [2015-02-10 06:46:29] Found chip 5 with 54 active cores
 [2015-02-10 06:46:29] Found chip 6 with 54 active cores
 [2015-02-10 06:46:29] Found chip 7 with 54 active cores
 [2015-02-10 06:46:29] Found chip 8 with 54 active cores
 [2015-02-10 06:46:29] Found 8 chips with total 432 active cores
 [2015-02-10 06:46:29] SPI(cs2) no device
 [2015-02-10 06:46:29] ACK(cs2) timeout:cmd_RESET_BCAST-0.0421s
 [2015-02-10 06:46:29] SPI(cs3) no device
 [2015-02-10 06:46:29] ACK(cs3) timeout:cmd_RESET_BCAST-0.0415s
 [2015-02-10 06:46:29] SPI(cs4) no device
 [2015-02-10 06:46:29] ACK(cs4) timeout:cmd_RESET_BCAST-0.0403s
 [2015-02-10 06:46:29] SPI(cs5) no device
 [2015-02-10 06:46:29] ACK(cs5) timeout:cmd_RESET_BCAST-0.0392s

[emdje]

Found the error, soldering error. But when I tested the board, I f*cked up the cooling and fried one of the chips
I did however fix the issue, but now it doesn't work any more.
So either I need to get a new chip to put on the board, or I take it off completely and solder some wires on it so the daisy chain is good again.

(IF ANYONE HAS A BROKEN BLADE AND WOULD LIKE TO DONATE THAT TO ME, OR LOOSE CHIP(S), PLEASE LET ME KNOW)


But powerconsumption of Overvolted board;

at 1400MHz:
Power using no blades: 17 Watts (Rasp pi and fans)
Total Power using 1 blade: 208 Watts
Power for 1 blade at 17.36 MH/s: 191 Watts
That is 11.00 Watts per MH/s

at 1200MHz:
Power using no blades: 17 Watts (Rasp pi and fans)
Total Power using 1 blade: 179 Watts
Power for 1 blade at 14.88 MH/s: 162 Watts
That is 10.89 Watts per MH/s

at 1000MHz:
Power using no blades: 17 Watts (Rasp pi and fans)
Total Power using 1 blade: 154 Watts
Power for 1 blade at 12.4 MH/s: 137 Watts
That is 11.05 Watts per MH/s

[BakSAj]

Compared to my consumption on 1200Mhz its +25% increase.

[wolfbg]

Any chance of just getting the cgminer executable w/ Extranonce support so i can upload to my image.  I don't want a bunch of down time just to swap the cgminer executable.  Thanks

[emdje]

Sorry, but as soon as I do that I will get a lot more questions about stuff not working. The image works. You would just need one downtime and than copy the executable on a USB stick for the rest of your miners.

[wolfbg]

FYI,

stratum+tcp://us-central01.miningrigrentals.com:3333/#xnsub

the above works for Extranonce support on miningrigrentals if you add it to the local config on the A2.  So far its been working good, thanks again emodje!

[emdje]

Thnx for the input wolfbg

[emdje]

Voltage mod is posted on crypto blog: http://fuk.io/how-to-voltagemod-your-a2-board-in-ten-steps/

[hrustix]

Hello, I recently bought A2 new mini, and in stats it says, that the board has 12 chips, not 10, here's the pic: https://i.imgur.com/Ow4TmxO.png?1
Also, if you still want to finish your job with A2 new mini image, I want to help you in any possible way.
Will try your 2.1 version within an hour, then reply.

[emdje]

Hello, I recently bought A2 new mini, and in stats it says, that the board has 12 chips, not 10, here's the pic:
Also, if you still want to finish your job with A2 new mini image, I want to help you in any possible way.
Will try your 2.1 version within an hour, then reply.

Yeah you are right. I recently found out the board seems to 'expect' 12 chips, but the logs only show 10 (at least the logs I received):

Code:

 [2015-02-25 03:58:52] Found chip 1 with 53 active cores
 [2015-02-25 03:58:52] Found chip 2 with 54 active cores
 [2015-02-25 03:58:52] Found chip 3 with 54 active cores
 [2015-02-25 03:58:52] Found chip 4 with 54 active cores
 [2015-02-25 03:58:52] Found chip 5 with 54 active cores
 [2015-02-25 03:58:52] Found chip 6 with 0 active cores
 [2015-02-25 03:58:52] Found chip 7 with 54 active cores
 [2015-02-25 03:58:52] Found chip 8 with 54 active cores
 [2015-02-25 03:58:52] Found chip 9 with 54 active cores
 [2015-02-25 03:58:52] Found chip 10 with 54 active cores
 [2015-02-25 03:58:52] Found chip 11 with 54 active cores
 [2015-02-25 03:58:52] Found chip 12 with 0 active cores

So I expect there is room for 12 chips but they only put 10 on it.

Can you check your logs to see if the same goes for your miner?

[hrustix]

I can for sure, just tell me how to do this, I'm not very familiar with linux.
PS: tried your 2.1 image, cgminer didn't start.