It came from weeks of statistical measurement. In theory, you could leave timing out and BFGMiner will autodetect it, but setting it correctly gets better performance, at least at the start.
I did try leaving it out and got the same result as having a, for lack of better term, bad value plugged in - start about 10GH/s and decline toward 0. I do not have the means to make statistical measurements, and from what I've been told and experienced with 3 U3s myself is that each one is different and will react to volt/freq settings uniquely, so I wouldn't be surprised if each machine had a slightly different timing value that it preferred. I do have the brute force method, so I'll keep playing with the timing on one unit and see if I can keep it alive for more than a few minutes. 0.021, for example, would error out and only 2 of the 4 chips would be seen by bfg. I'll start at 0.020001 and work my way up, hopefully there's a sweet spot in there somewhere. |
|
|
|
I am willing to pay someone to repair/replace the USB and power connectors on these boards. I'm just too lazy to do it myself  Contact ncsupanda: https://bitcointalk.org/index.php?topic=1009705.0, he fixed 2 of my U3s, one he replaced the power connector actually. He is in USA. |
|
|
|
i saw in antminer u3 pdf file that there are column for preferred timing
so i take in my case 0.54 and divided it with 100 ms
0.54/100 = 0.0054
That makes sense. But I wonder, then, where the value in the README "--set antminer:clock=x1286 --set antminer:timing=0.022421" came from. According the the U3 documentation the timeout for 1286 should be 0.0057. Unfortunately for me, either value ends in the same result: my U3 starts, but rapidly the hash drops to 0 and the unit goes SICK. By trying several combos I've gotten it to start hashing as high as 40GH/s, but it only lasts seconds. BTW, this is a completely separate computer and location from my stick miners. EDIT Just started messing around with the timeout, using a voltage of x785 and freq of x1286. When I entered 0.02 the U3 stayed alive! It's hashing away at 48GH/s, which is about what I would expect for the volt/freq combo. Could it be that timeout in the user guide is not the same as timing in bfg? I need more insight into how to determine the timing value... EDIT 2 It's sick again. Well, getting somewhere at least. |
|
|
|
|
The one thing gold has going for it that bitcoin doesn't is thousands of years of human history as a valuable commodity. I have a feeling it will always be one of the kings of the hill.
|
|
|
|
why do people buy bronze psu's you lose money with them.
Yup, the extra money you spend on a gold or platinum is quickly made back in the efficiency gain between psu's. I had a crappy bronze unit on my S3+, switched it to a gold and saw a 5% wattage reduction at the wall. |
|
|
|
|
Gotta keep the wheels of the military industrial machine greased and rolling, and can't do that unless the boys have someone to play toys with. Makes it much easier when you supply the toys.
|
|
|
|
If I invest in financing halfway through the month, will I earn half a month's interest, a full month's interest, or no interest?
No interest. So, if I invest today, I will not earn any interest for June. Therefore, I should leave my money in the "Banking" wallet earning 8% and wait to "Buy Financing" until the last day of the month. Is that the right way to look at it? No, you'd earn 1 day's worth of the higher interest if you did that, which would be silly. If you don't need the funds for 4/5/6/7/etc. weeks, put it in the financing. If you want liquidity, use the banking. |
|
|
|
You have to actually use it to expect any results... skipping is just a waste of time.
Test each build presented (they are all different) and determine if it's good or bad.
Use a new session name to start over.
It said to skip if I was unsure between the other two choices, so I did. I obviously didn't understand the process.  So it gives me a new build at each step, that's what those 32-bit 64-bit links are? I get it, will try again. When I am presented the working and non-working entries at the beginning I am changing the version numbers at the end to -5.1.0 and -5.2.0, is that step correct? |
|
|
|
Contrary to what some have posted, you do need 8333 open outgoing to sync the blockchain.
Another bullshit claim. As of this moment getaddr.bitnodes.io shows 59 full nodes listening on a non-default port, different than 8333. Ok, if someone changed the default port to something other than the default they would then need to open/forward said port in lieu of 8333. You got me and my "bullshit".  |
|
|
|
Here's the end of the bisect, not sure if I did it correctly or no, please let me know if this is what you were looking for (by the way the "16 revisions left... 4 steps" displayed on every screen of the bisect, I expected the numbers to change but they didn't): |
|
|
|
There's been quite a bit of discussion on this topic throughout the forum. Some feel empty blocks are valuable. Others feel they are of no value.
I thought it might be fun to write a small program to see just how many of these blocks exist in the blockchain. For those who don't know, an empty block is a block that is mined containing only the coinbase transaction that awards the new coins. To be honest, I was quite frankly surprised by the number there were, especially since people seemed to think they were rare occurrences. Sure, I expected there to be quite a few at the beginning of the blockchain when BTC was pretty much an unknown and only a select few were mining it. However, I was not expecting there to be so many empty blocks as there are.
As of block 360189 there are 85295 empty blocks on the chain. For the math challenged, 23.68% of all blocks are empty. 8 of the last 189 blocks are empty. 25 of the last 1000 are empty.
Maybe I'll get ambitious and write the output to a spreadsheet so I can get some good data points out of this to see trends.
Where do you stand on this debate?
Thank you for the research, the numbers are enlightening. I tend to fall on the side of the discussion that holds that blocks should contain transactions, since that is their ultimate purpose. |
|
|
|
Wow, that was a lot of to-do about nothing... Look at any block explorer and tell me how many "empty" blocks you find. I've never seen one, though I know they can theoretically exist. Just because of this conversation, I think I'll pay more attention to it. but seriously?
Over 23% of all blocks are empty, quite a significant figure and worthy of discussion: https://bitcointalk.org/index.php?topic=1085800.0So yes, seriously. |
|
|
|
Everyone should have own mining pool  If everyone who mined had their own pool, who would mine to all the pools? derp |
|
|
|
Looks like you're on to something here, 0 incoming connections and 8 outgoing connections. That is typical if you do not have port 8333 open and forwarded. You only get more than 8 connections (i.e. incoming) when your firewall is configured properly. You don't need to do this to sync the blockchain, though, evidenced by the fact you were able to sync up to a point; you should have just deleted the installation and run it again at the beginning. Contrary to what some have posted, you do need 8333 open outgoing to sync the blockchain. For example my personal firewall blocks all traffic, in or out, so I specifically had to open 8333 outgoing for the sync and incoming to be a contributing full node. |
|
|
|
In that case, can you also try putting the libhidapi-0.dll from 5.1.0 into the 5.2.0 directory?
The hidapi update in 5.2.0 was supposed to only fix bugs, but maybe it introduced new problems too..
Running through the WeBisect now, never done it before so I'm just clicking the skip button each time and letting it crank through... I tried replacing libhidapi-0.dll from 5.1.0 and still no luck. I'm also trying both 32 and 64-bit 5.2.0, just in case. When I start both 5.1.0 and 5.2.0 each stick lights up when "[DATE TIME] Started bfgminer 5.X.0" appears in the command line window, then the lights go out on the sticks. In 5.1.0 a few seconds pass, then they light up again and bfg launches. In 5.2.0 they remain dark and the command line window sits at "Started" indefinitely (I've left it alone for over an hour, just to be sure). Will post up again when webisect is done. |
|
|
|
I'll give it a shot after lunch. I am using Cryptorig HitchHikers, they use the nanofury chip. |
|
|
|
"set-device" : [
"antminer:voltage=x830",
"antminer:clock=x982",
"antminer:timing=0.0054" So the clock and voltage can be taken from the U3 user guide, however, it seems the leading 0 from the user guide needs to be an x, is this the case (frequency 250 has a value of 0982, so you input x982)? If so, what about the frequency setting that begin with 1? e.g. Frequency 243.73 has a value of 1306, what would this be in the config file for bfg? How did you determine the timing setting? I know the readme says "the number of nanoseconds each hash takes at the given configuration", but how do you determine that number? 5.20 docs say this replaces the older "--no-opencl-binaries" command to ignore using CPU for mining.
Missed that in the readme, but I never used the old command, either, and have never had bfg try to mine with my cpu... |
|
|
|
|
My solo miners are on failover and my monitoring webpage has timed out, problems with the pool?
|
|
|
|
|
There's no god, the answer is something way way way bigger than our brains could ever hope to understand. There are infinite reasons why infinite universes exist, have existed, and will exist. Trying to look for a simple answer is human nature; there is no answer.
|
|
|
|
Mass adoption is happening now, slowly but happening for sure. There is more than 6,500 physical and internet based shops/services who accept bitcoin as payment option. And number raise every day http://coinmap.org/For comparison, there are over 425,000 ATMs in the USA alone. There are millions of businesses in the USA. Put that 6500 in perspective and you can hardly call it "mass adoption". |
|
|
|
|
Show Posts
