Topic: [Work in progess] Burnins Avalon Chip to mining board service - page 68. (Read 624200 times)

had a few drinks and this had me laughing out loud..

Meaning of the LEDs:

Green:
steady blinking - board is operational
off - in bootloader or something is wrong

Orange:
flashing - active data transfer from this board (this is a good indicator if CAN works or not)
off - no data transfer in progress from this board

Red:
steady blinking - board is in bootloader mode, it will jump to the firmware in a few seconds.
short red flash - found a nonce
solid red - something went horribly wrong

@SebastianJu
What  read from that cgminer output: 39.2ghash, perfect in line with 5x20x400mhz = 40ghash.
A few HW errors are to be expected when going past 400, but are not a problem.

The default avalon settings match a default BTB (although the '2' is wrong, it doesn't matter)
If you chain them, then you'll need to use --avalon-options

With 35:350, the 35 timeout is important to get correct to stop duplicates.
I'll look into adding the option to use a flag (e.g. d or * or ? or -1 or some such) so that it calculates it for you if you don't want to.
When you switch the frequency via the API it recalculates the timeout for you.
(so if you just set the frequency the same as it currently is, it will correct the timeout)

At the moment the firmware doesn't return information about the internal configuration, but that will be coming soon and I'll then get the code to use that to determine the correct "2x10" for the chips.

I think i know what you mean now. My setup is 10x10 chips. After i tested this i found that the hashrate was quite better now. I checked how far i can go and at 400mhz i got HW-Errors. I raised the voltage to 1.3V and zero HW-Errors again. And now i hash at 38GH/s without HW-Errors.

Unfortunately 420MHz lead to HW-Errors. Ill have to check out how to handle this. But i will let it run now over night.

Im not sure but it seems the avalon option is needed to get the full hashrate. So maybe the recognition isnt working fully right till now so that less chips are found or really used.

It looks like avalon-options was the source for the HW, regardless of what MHz or voltage i used it was always 50%. Even with:
I dont know if i have to change the number before the clockrate but i have 5 bitburner xx which should be 2x10 chips each.

Now that i took out avalon options the hashrate is lower again but zero HW-Errors again.

Ill let it run over night now this way...

I tried this now. I found that all boards work like they should when run per usb alone. When using 2 boards together the hashrate doubles but the third starts to not make a big difference. One miner makes 5GH, 2 makes 10GH but three, i tried all remaining three one after another, only leads to number around 9-10GH again. Like it doesnt make a difference. And i used the jumper correctly. The last board at the stacking cable and at the opposite site of the usb-board needs the jumpers. The usb-board itself seems not to need the jumper very much. But now i have the impression the HW-Errors are a bit less when the usb-board has jumpers too. Might be random.
When i put in the remaining 2 miners it only goes to 14.6GH or so and thats it.


This option seems to work. The hashrate jumped a bit from 14 to 17GH. I used your setting. Unfortunately the HW-Errors rise a lot. A/HW is only slightly better than 100/50. And when i set the voltage to 1300mv it doesnt help. Burnin's Tests showed that raising the voltage would help to lower the HW-Errors. It looks like i dont have luck with this.
And i noticed that the higher clockrate leads to often this line: BTB0: Idled 1 miners. I dont know what this means.
One time the miner crashed after a minute when i ran it with high clock but normal voltage.

I hope theres a solution for the hashrate and the HW-Errors. I updated the firmware of all boards before too.

When i look at the stats... since i use the higher clockrate the temperature isnt anymore around 36°C, its only 24°C now even though the fans arent louder so i doubt they work more.

I use this command in a batch:
And cgminer shows this after 5 minutes:

try
./cgminer-nogpu -o ... --avalon-options 115200:2:10:35:350

If you have different per device settings to 2x10 chips - set those numbers as appropriate

Also, I run mine at 400 without any problems: --avalon-options 115200:2:10:30:400
( ~1.4% errors and: --avalon-fan 99-100 )

 BTB  0: 48/ 48C 1225mV | 8.228G/7.766Gh/s | A:150253 R:930 HW:2107 WU:110.0/m
(approx 23 hours)

Maybe it's time for good-old problem solving technique: reduction to properly working minimum, then adding things one by one. Disconnect the boards, then power and test one by one. Make sure that each one is working as it should and that each one gives the same result. Then make a pair, and finally completely connect everything. Somewhere along this process things will start working properly, or a problem will be pinpointed.

I tried closing with q but it doesnt change. The flickering is a bit less, especially for the first board but it goes on somehow. And there is no process open anymore.

I already use 3.3.3 after i saw it some hours ago.

Will there be a miner-setting-menu at some point? So that cgminer dont have to be restarted for volt- and clockchanges?

And when i use it seems to not having an effect on the hashrate.

I've had trouble with various Class 6 SDs and no problems so far (touch wood) with class 10 SDs
The class 10s I got are SanDisk Ultra SDHC Cards

I've mentioned this here also: http://www.kano-kun.net/?p=87

Why class 10 SDHC ?
Can you recommend a class 10 SDHC that will work with the RasPi.
Thanks!

Not closing an old cgminer properly? (see task manager)
It is always best to quit with 'q' and see it exit - windows ...

Also try 3.3.3 and see if it has that windows problem you mentioned.
I'll try my BTB on windows again today with 3.3.3 and see if I have any problems.

I've not built 3.3.3 yet, it was only released last night while I was asleep - it's mainly CPU improvements, but there was also a problem with the first windows version of 3.3.2 in the downloads if you got the wrong copy of it.

Make sure it's a class 10 SD ...

The lights are a bit different form board to board. All green lights are the same. The board i put the usb in is blinking yellow very often, seldom red. The second board blinks half so often but most of the time yellow and red. Third similar but seldom at the same time. The next blinks more seldom and red any yellow are similar often. The fifth blinks seldom and sometimes more red than others.

It looks like the first miner with usb connected works more than any other.

What is the miner doing when i close cgminer and it still works on?

Didnt think about but you are right. Should be 28200MH/s at 282MHz.

What does the led's mean? Green, yellow and red?

And i miss a configuration menu like [G]PU for gpus. I cant change voltage on the fly. And it looks like the MHz i set in the batch is ignored too.

You can always power it off a powered hub...

I havent tried it myself, but it should work if u use any of the 5v leads from atx. The pi needs a 5 volt supply (same as any usb powered thing).

It was my intention to mine using my PC and Windows.  Has anyone else run their bitburners from Windows?

You have 5 boards, right? 15 GH/s seems to be a little low for 5 boards.

I plan to buy an sd-card tomorrow.

Regarding powering, you know a way how to power it with a atx-psu? I mean it has different voltages and is there anyway.

By the way... cgminer seems to run now. I use now cgminer-nogpu and it works. Showing 15GH/s with normal settings. 220W are drawn in average.