Just got my KNC today, seems like its running at 522 GH/s or so (with 0.96 firmware) according the web interface. It hasn't been running long enough for me to get accurate stats from the pool I'm using. I'll measure power consumption when I bring it home tonight.

I'm surprised on how quiet it is. It is so much quieter than my Avalon and BFL singles.

The web interface says its hot (52, 56.5, 58, and 49 are the temps) but the heatsinks are not even warm.

It arrived with three fans off and sliding around inside the case. Taking off the case was a PITA because I didn't the special screwdriver needed.

What number were you?

The ironic thing is I had a WeMo attached to it so I could reboot it remotely. But then I noticed my raspberry pi was crashing from time to time (I'm not sure why, my other pis never crash) so I moved the WeMo to the pi. Within an hour the BitBurner had crashed! I didn't have the time to run home yet again to reboot it so it's going to have to wait till after work now. I'll probably lose a quarter of a bitcoin because of it . Fortunately I have another WeMo, so WeMos are going on both the pi and the miner now.

Ugh, I had an extra M board on order but I canceled it. The M board kit is $1300 with just 1 H board so it really isn't worth it.

I'm on the verge of canceling my order. Bitfury from megabigbpower is showing the worst ROI right now, when compared to BitBurner, HashFast, and KNC, especially if a 16-board kit can only do 500 GH/s or so. It needs to be able to do 600 GH/s and arrive in the next week to justify it's $8k cost.

I know that the chips are actually capable of doing 800 GH/s as my BitBurner with half the chips can do 400GH/s. The power regulator and cooling issues will limit it though, as for the fact there are only 2 6pin connectors. I'm guessing those two connectors limits it to 600w maybe?

If the power supply is a single rail design, why not? If not, maybe I can figure out a way to add more connectors. Or do we risk blowing out the M-board?

What is the max speed that could reasonably be overclocked to with a 16-board setup?

I'm going by goxed's post where he said he could get 41 GH/s drawing maybe 70w. It makes sense since I have Bitburner Fury setup that also uses Bitfury chips. I'm able to overclock each 16-chip board to about 49-50GH/s which draws about 100w. So 70w at 40GH/s seems accurate.

Bookmark this thread!

Here is what mining will be a year from now:

- All miners will use 28nm technology.
- Miner power consumption will be roughly 0.5w per GH/s
- The speed of a medium-sized mining machine will be 1 TH/s, but there will be 2+ TH/s machines available. There also will be plenty of smaller machines in the 200 GH/s range.
- The medium-sized 1 TH/s machine will cost in the hundred of dollars range. Probably $400 to $750.
- Difficulty will be around 30-35 billion.
- The vast majority of mining will take place in areas of the world with cheap electricity.
- The average Bitcoin enthusiast will fit into one of two categories:
    - Miners who have very cheap or free electricity (perhaps in an apt or dorm room where electricity is include). They will pay off their device in roughly 6 month to a year and make a very small profit.
    - Miners who do not have cheap electricity but choose to mine for fun or simply to secure the bitcoin network.

40GH/s and 70w per board!? Wow.

That means a full 16-board set should be able to do 640 GH/s at 1120w.

I have a 1250w power supply. I hope thats good enough

Maybe I'll aim for 37.5 GH/s per board and hopefully around 1000w

Is there a way to recover from the dreaded "usb_write error on avalon_write" error without power cycling the miner?

I'm getting it now and I'm not at home to power cycle it. No matter how many times I reboot my pi or restart cgminer seems to fix it.

How overclockable are the boards that are being shipped now? Is the power regulator still the limiting factor? Has anyone tried to replace it with one that can handle more power?

The price difference is December vs January shipping.

I'm so tempted to order from them. 2 TH/s for that price seems amazing. But then I think about what the difficulty will be in January and I realize that 2 TH/s in January isn't any better than 500 GH/s now. Sigh.

My electricy is $0.16198/kwh.

Here are the difficulties where the various miners (for me) will make only as much as the electricity costs to run to them:

BFL FPGA                             156,496,295
82GH/s Avalon                  1,770,608,552
BFL 7GH/s Jalapeno           3,067,327,400
BFL 60GH/s Singe             3,572,198,060


The FPGAs have been obsolete for 2 weeks now. The BFLs still have quite some time before they are obsolete. Avalon has less time but still a bit.

The newer machines will be obsolete in the 15-25 billion range depending (KNC is around 18 billion, HashFast is around 24). I don't know the exact numbers as I don't have those machines yet.

ok, makes sense. Has anyone tried using just heatsinks then? Is there some off-the-shelf heatsinks that would work or is it better to go with something custom made. I can get custom made metal parts made pretty cheaply so that is an option if necessary.

Any look into building a water cooling system for this. As I understand there are no heat sinks on the board.

What is the spacing between the boards? Is there enough room for a custom water cooling solution?

Yes, I'm aware a lot of the software already exists. The more that I think about it, the more I feel like colored bitcoins is the way to go. But a real exchange is a lot more than that. You need everything else that goes along with it. Namely, a p2p way of distributing information regarding the available securities and finding people to trade with.

The thing is, no one runs a decentralized exchange. That's the point. Who runs Bitcoin? The only thing you need is a website or someplace for users to download the client software and there you go.

Now, the better question is why isn't there incentive for people (both traders and issuers) to use a decentralized exchange. Up until now there hasn't been. But BTCT is gone. Bitfunder can't be used by US residents so it might as well be gone. And Havelock Invesments is too small and will likely suffer the same fate as Bitfunder and BTCT if it gets bigger. There is now a HUGE incentive to a decentralized exchange that can't get shut down because we are almost out of options.

Here is what we need:

- Colored coins as the backend to deal with marking transactions as belonging to a specific security, etc. We need an API based on top of colored coins that let's a trader or issuer perform everything they'd need to do with a security and have it happen using colored coins. For example, create a security, trade shares from on address to another for a given price, get the last trade price of a security, etc. I'm working on specifying this API right now.

- A client that runs as a p2p system for distributing security data and facilitating trading. This can work like any other p2p system (gnutella, bit torrent, bitcoin, etc). It will be used to distribute information and updates about each security by the issuers and allow the traders to post orders for buying and selling which can be automatically filled by the network. This is the part that doesn't exist yet, but it is nothing novel, technology-wise. It is also the part that doesn't require the bitcoin network or the same level of security. All the security and anonymity are provided by the bitcoin network. Because this will be a p2p network with no central server, it can not get shut down.

oh thanks! It was in "devs". I thought I had tried every command. Yeah, it would make sense if it was in "summary".

I've been using the cgminer API to get statistics about my mining devices.


One thing I can't seem to find is the current hashrate (or 5sec average hashrate). I see an average hashrate that appears to be since the machines was started but I don't see the current hashrate.

If this statistic is not available, how can I calculate it from what is available?

mgio   23xx   19 June   1   0   0   USA   None

Tracking number received this morning. Expected delivery date Oct 17

I can't quite put my finger on why but my feeling is that HashFast is going to end up being the most professional and reliable of the bitcoin miner hardware companies. That is, if they make their deadline.

Of course, they don't have that tough of a competition.

If they don't ship in the next week, it makes sense to cancel your order. That is, assuming bitfury and hashfast are going to be shipping in the next week. They both are cheaper and match the performance of KNC so I see no reason in holding on to your KNC order. That is what I plan on doing, at least. I'm not angry at all. It's just the way that mining has to work these days. The downside of accepting customers for pre-orders is that they may cancel (and contest their credit card charges) at any moment if you miss your deadline. With BFL and Avalon the risk has been born 100% on the consumers. It's about time that mining companies step up and absorb some of the risk themselves. So far there has not been one company that has shipped on time. I'm not saying that KNC has failed, but they have 3 days to ship roughly 1,000 miners to keep their promise to their customers. If there is no guarantee, if there is no downside for a mining company if they break their promises, what is to stop them from just promising anything?