The internal oscillator has a frequency that's voltage dependent. So, increasing the core voltage is all you have to do to make the chips run faster. At least, that's the theory. In practice, the board is powered by an integrated regulator with 30A capacity, and fixed voltage by SMT chip resistor. If you increase resistor, voltage goes up, but so does the current, and you'll risk overloading the regulator. You need to carefully monitor the current.

The 2.7GH/sec result I got with a single ASIC test board required 2.5A, and 0.835V. If you were to attempt that on the whole board, the total current of 40A would exceed the capabilities of the regulator.

There's a limit to the output power that can be provided by the on-board regulator, and the voltage is not easily adjustable. However, if you feel a bit adventurous, you could heat up the soldering iron, remove a component to disconnect the on-board regulator, and solder down a wire to provide your own power.  The bottom side of the board is completely flat (at least in the part where the chips are), and can accomodate a heat sink. There are even mounting holes in the board that could be used for that purpose.

At the voltage these boards work, a small fan blowing air on a board, or between two boards is good enough. No extra heatsink is required.

The board in the pic from the web-shop is working. It only needs to be plugged in and fed some data.

230 MHz and 0.84 V requires quite aggressive cooling. The (small) test board had a small heatsink soldered on the bottom (originally for D2PAK), and a small fan aimed at it. Board temperature was about 50 deg C.

The chip is running at 0.84V and 2.5A, with a clock of 230MHz.

The disadvantage with an external clock is that there's no practical way of tuning the clock frequency to individual ASIC performance. Given variations in process and circumstances, it may be better to optimize each ASIC's internal clock, rather than using a single external clock for all of them.

I don't know if a respin is planned, but maybe a separate VDD for the oscillator circuit would help improve stability.

We haven't tried external oscillator.  I don't have a generator that can do 200MHz+ outputs. The internal oscillator is fast enough, but it does look jittery on the scope, so internal setup/hold timing may be improved with stable external oscillator, and a higher clock may be used.

2.6 GH/sec, with single 2W ASIC.

The hardest part will be a good estimate of the rise in difficulty. It is possible that difficulty will skyrocket with all the ASICs hitting the market. Personally, I wouldn't invest in anything with a break-even point more than 1-2 months away at current difficulty.

hi everyone. I've been lurking for a while, especially on the Bitfury threads, where I've been working with intron behind the scenes to create hardware and software for the Bitfury ASIC.

Poster intron has been relaying my comments, but I thought it would be easier if I registered and made comments myself.