The Psychology Behind Never Feeling Good Enough

I recall vividly the first job interview I had after graduating from the University. I was 21 years old. I had only applied for the job because I was trying to improve my skills in job application…


独家优惠奖金 100% 高达 1 BTC + 180 免费旋转

Bitcoin Mining Efficiency Set to Double by 2020

Bitfury’s new 14nm ASIC Clarke which claims an efficiency of around 18 GH/J.

After a distinct lull in innovation within the crypto mining sector since 2016, this year has seen a flurry of new developments largely driven by new chip technologies. In addition several new manufacturers have entered the industry. How will this effect the network and where is the trend going?

In the first half of the year Japanese tech company GMO Internet announced their entry into Bitcoin mining as well as the retail launch of their B-series miners. These devices, now sold out, feature 7nm ASIC chips making them potentially much more efficient than the previous 16nm state of the art.

But how does chip size make a difference? Well, essentially the chip sizes stated here refer to the size of the smallest component etched onto the chip (probably a transistor). Some of you may remember from physics class that transistors are made of positively and negatively doped semiconductor material and effectively act as switches allowing electricity to pass or blocking it depending on whether a certain voltage is applied. This works because the semiconductor material allows positive and negative charges to gather at the p/n interface and form a barrier and this is overcome, i.e. the transistor is switched, by applying a voltage. When we make smaller chips, we mean the transistors are smaller. In other words the interface is smaller and therefore is made up of fewer charges (electrons and holes). And of course this barrier requires less voltage (energy) to overcome. As an added bonus it also switches faster meaning more computations are possible for less energy.

In addition to the GMO news, Innosilicon and Ebang also released their new T and Ebit series miners based on 10nm technology. So are these new ASICs actually more efficient? The honest answer is marginally. However they are promising. Although their efficiency isn’t much better than Bitmain’s Antminer S9 from 2016, these companies are relatively new to the sector. We should expect that the quality of their rigs will improve over the next few models. And let’s not forget that Bitmain also announced a new, still unreleased 7nm chip in 2018, which on paper at least is by far and a way the most efficient ASIC to date.

What we see from the above chart is that the energy efficiency trend is growing exponentially. In fact, if it holds Bitcoin mining efficiency should double by 2020. So does that means we will cut our electricity consumption in half? Probably not.

Bitcoin price (yellow) and hash rate (blue). Note the very rough correlation between the two.

As the logarithmic chart on the left shows, hash rate continues to grow exponentially. There seems to be a real arms race among miners, which has continued unabated despite the market correction. And we can expect this to continue into the future. In fact this year’s new miners also push the bar higher in terms of computational power and seem to be making 20 TH/s the de facto standard.

From the above chart it is clear that the power of mining units is also growing exponentially to match demand. Again, if this trend continues, we can expect standard miners to offer 70 TH/s by 2020. This also corresponds to a doubling of our most powerful, contemporaneous miners.

So it seems that the current trend will only stave off the inevitable. Or will it? What about even smaller chips? The next gen chips currently being researched are TSMC’s 5nm and Samsung’s 4nm chips. (Btw Taiwan Semiconductor also make Bitmain’s ASIC chips — that’s food for thought now.) Well, making any chips smaller than the current 7nm requires overcoming quantum tunneling effects, whereby electrons effectively move through potential barriers (insulation) due to statistical effects leading to short circuits. However both the above companies report that planned production should commence in 2020.

And 2nm? Well it should be obvious that, at some point, a different solution will be needed. But I guess the community will have to cross that bridge when we get to it.

Add a comment

Related posts:

How to Fix Your Customer Service Escalation Process

One of the major frustrations faced by customer service teams is that the most serious tickets — the ones that can mean the difference between a retained client and a lawsuit — are often the most…


Escrevo o tempo inteiro. Meu bloco de notas, é abastecido constantemente com os pensamentos e sentimentos que tanto suplicam para serem escritos. A cada novo texto, consigo escutar atentamente o…