Of course. However when everyone uses systems like viabit you're back to square one. It's a classic Queen's race, where everyone runs faster and faster to stay in place.

What is the temps on the cube that throws the errors. If the CPU temps are shooting up quickly past 40-45c, you might have an issue with the thermal interface between die and heat sink. Removing the heat sink, cleaning off all the old compound, putting a thin layer of new HS compound on, then putting back on the sink could fix things.

Another possibility is the chip got hot enough underneath to de-flow the solder a bit. Solution there would be a reflow but that is hard to do.

Another possibility: Have you tried running that die at a slightly higher voltage and lower (like 100mhz) clock speed? You might be able to pick up a few MH that way without fixing anything else, probably the most economical option.

Meantime run your cubes at 60-70mh and not full bore at 80. Helps them to last longer.

C

Of course the next problem is going to be the BTC core nodes. I run one of those as a service to the network just to relay transactions, and it's requiring more and more disk space, memory, and cpu time to keep running. At some point people are going to get annoyed at the thought of storing every transaction forever for no pay.

Bitcoin is a perfect example of a pure perfect market. Even if you fix one section (double or quadruple the block size) it causes problems in other sections (ok, what about nodes who now need more disk and CPU)? Segwit has its own problems as well, as does inertia.

So what is the solution....

I love that picture Maxumark. I've also had a titan shoot fire out the vents when I plugged it in for testing (the capacitors on one of the DC-DC's shorted, which ignited all the filth and dust and crap in the miner which the fan happily blew though the box). I take all miners apart now and clean and test before powering. Did wake me up though.

Remember that heat in connectors increases resistance. Which results in more heat which results in the normal solution, melted connectors. Get a thermal IR temp thingie and check temps on your miners.

This is an interesting issue. As a person who repairs bitcoin and litecoin miners for payment (in bitcoin), boosting the fees is annoying. You can say it will draw more miners, but that will simply increase the difficulty which will make each miner less effective which means no difference to the confirm rate.

You can raise fees, and that will make more money for miners, but that will simply bring in more miners which will increase the difficulty and make each miner less effective. Queen's race.

You can play the game of paying mining consortia to include your transactions in blocks, but that will just result in more consortia doing this, less confirmation of normal blocks, and more miners coming in which raises the difficulty which makes each mining consortia less effective. Queen's race again.

About the only people who profit are power companies. Invest in electric company stocks.

Another option: I'm going to start accepting payment in litecoin and see if I can convert that to bitcoin in large chunks, or buy some of my materials with ltc-cash. No problem confirming transactions there.

Another thing to watch out for is power factor: A lower PF supply is going to put more heat on that adapter and probably burn it out. PF problems will not trip breakers, they will burn plugs first. Ask me how I know....

Agree with the other posters: Just run your supplies at 240v. Half the amps, less heat load.

C

Back on topic, can someone update that stupid Wiley Coyote gif to go from 300 to 1300?

Because this is important.


C

Looks like an expired cert on their side. Oops.

Meantime question: Would it be worth my looking into these things to see if they can be fixed stateside?

Note to all: If you're sending me equipment please drop me a PM with the tracking number so I can know what to expect and when. I use the tracking numbers to keep track of the various work I'm doing here.

Never dull....

I did detailed TA on the entrails of a chicken before lunch. TO THE MOOOOON!

Hell, I was just amazed a coin shop would take bitcoins. Which was ok, I needed to level up a bit on my gold holdings so it was the proper decision at the time.

Kind of in a wait and accumulate mode. Problem is I have to keep lowering my repair prices in BTC because it is so stupid high.

C

This is just another push to 1300. But in the meantime you all realize that one bitcoin can buy an ounce of gold...

CURSE ME FOR SPENDING 2 BTC for an OUNCE OF GOLD! AARRGGHH!

C

Yeah it's been a good time this month, and thanks for banging on me to make it work.

Here is what I needed to do:

The key was in talking to a developer who said "I don't think you're using the right spimux.rbf". Well, I was, I had spimux.rbf in /etc linked to the spimux-titan.rbf over in home/pi/knc-asic/  However digging through the files I found that program-fpga was picking up from /home/pi, NOT /etc. Thus it was picking up the neptune code. So I deleted it, linked the titan code to spimux.rbf in /home/pi/knc-asic and WAAS finally worked.

Still no cube, went to bed. Next morning I fired up putty, turned on full logging and found this in the error logs.

[2017-02-28 16:41:37] KnC transport: Can not open SPI device /dev/spidev0.0: No such file or directory
 [2017-02-28 16:41:37] KnC transport: Can not open SPI device /dev/spidev0.0: No such file or directory
 [2017-02-28 16:41:37] KnC transport: Can not open SPI device /dev/spidev0.0: No such file or directory
 [2017-02-28 16:41:37] KnC transport: Can not open SPI device /dev/spidev0.0: No such file or directory

Of course. Once again the code is hard wired to use the Rpi numbering scheme which goes from spidev0 even though the bbb starts at spidev1.0. So another ln....

ln -s /dev/spidev1.0 /dev/spidev0.0

Fire up bfgminer and up it comes.....

 bfgminer 5.4.2-titan-2.02 - Start: [2017-02-28 22:27:44] - [  0 days 00:26:21]
 [M]anage devices [P]ool management Settings [D]isplay options  [H]elp [Q]uit
 Pool 0: us.litecoinpool.or  Diff:125m  +Strtm  LU:[22:54:04]  User: 111.1
 Block: ...73be070473a1fe38  Diff:99k (708.9G)  Started: [22:54:04]  I:?
 ST:21  F:0  NB:17  AS:0  BW:[ 26/ 14 B/s]  E:0.67  BS:104
 1/4          | 50.44/54.90/54.80Mh/s | A:168 R:1+0(.59%) HW:20/.39%
--------------------------------------------------------------------------------
 KNC 0:       | 48.93/55.01/54.90Mh/s | A:170 R:1+0(.58%) HW:21/.40%
--------------------------------------------------------------------------------
 [2017-02-28 22:52:02] Accepted 0007e16f KNC 0b Diff 126m/125m
 [2017-02-28 22:52:03] Pool 0 stale share detected, submitting as user requested

 [2017-02-28 22:52:03] Accepted 00030374 KNC 0d Diff 331m/125m
 [2017-02-28 22:52:05] Accepted 0007f539 KNC 0a Diff 125m/125m
 [2017-02-28 22:52:08] Accepted 0003abfc KNC 0b Diff 272m/125m
 [2017-02-28 22:52:13] Accepted 0007a089 KNC 0b Diff 131m/12

So that's how you do it. Do the steps I outlined in the first post, make sure spimux.rbf in /home/pi/knc-asic is titan code, fix the links and you are off to the races. Note the /dev/ links and the /var/run links have to be recreated every reboot so stick them in i2c-loop or something.

Cost for this? If you find this valuable, drop me a tip at: 16spS7JmUUKuuQxK84tA4NRWbVaG877Ci9  . It was a good bit of work to figure this out, but I'm glad Titan owners now have an alternative when the Pi's start to blow up.

C

Yes, I am a LIVING GOD! Titan works on BB.....

Found it. The problem with Qberty's boards was that they had crystals mounted on them instead of oscillators. Using the Digi-Key 300-8253-1-nd 24.576mhz oscillator I'm able to fix a Qberty board, see Titans, and hash happily.

This was one of the harder ways to make .1btc, took a good bit to get it figured out. Wish Qberty had just said what was wrong, but that wouldn't be as much fun.

So if you have a 2.0 berty board and need it working drop me a line.

Still no reply to questions on getting the BeagleBones working with Titans, but this will help keep em running.

C

Yeah, just takes some prodding. BBB set up again code doesn't work, mailed the guy who might be able to help. We'll see.

C

Course. I had to shut down my miners 3 days ago because it's 80 degrees in February. That always brings a block.

Been too busy working on other things to work on the BBB code. Part of the reason I posted my notes here, someone else can work on it in parallel.

This is a lot of work to do here to try and figure this stuff out.

Oh for fuck's sake, it just hit me: It's not a crystal alone, it's a crystal oscillator! That's why the 3.3v supply goes into it and the ground is there on both pins, it's a whole oscillator circuit.

Now to buy some stuff at Digikey. I got it now.

C

Meantime I've dragged out the scope from the basement and have verified that the Qberty boards with the trace cut so they don't short still do not put out any signal regardless of crystal frequency. Getting closer though, will figure this out.

C