Hacker News Reader

Comment by guardiangod | Hacker News Reader

From the article-

Bitmain, the largest Bitcoin miner on the planet, is now buying a whopping 20,000 16nm wafers a month from TSMC! That's double as much as in the previous quarter and a higher volume than what NVIDIA orders from TSMC.

A competitor to Bitmain just finished their prototype 12nm Bitcoin mining chip and is moving forward to their target 7nm mining chip.

View on HN

> A competitor to Bitmain just finished their prototype 12nm Bitcoin mining chip and is moving forward to their target 7nm mining chip

Huh, maybe the dark fibre of Bitcoin will be cheap silicon and subsidized fab research.

Apply that to energy efficiency research as well. The monetary incentives are too strong for it not to happen.

I don't think it's possible to be energy efficient in a proof-of-work blockchain? Whatever energy saving you come up with will be negated by having to match competing miners' increasing throughput. If you are not running on 100% you are losing out?

There are projects like gridcoin, uscore dot net, that use mining to compute scientific research. It is considering using a Proof Of Research approach.

It would be game changing if blockchain mining efforts could be applied like this to other uses as well, like someone needing huge datasets analyzed and formatted into a presentable state on an ad-hoc basis, sort of like AWS Athena.

Just found out about Filecoin which is an interesting concept, you "mine" by storing and serving files people save to the network.

Stuff like this is a much more interesting use of cryptocurrency and blockchain technology in my opinion, instead of burning electricity for PoW you can do meaningful work while collecting a reward.

FileCoin and Upfiring both claim to have this PoW, but their Github repos are empty and their Reddit channels are drowning in shilling.

Unfortunate, I like the concept - I wouldn’t mind giving up a couple TB of space on my FreeNAS rig that’s currently going unused.

And the point of blockchain is to set a minimum difficulty level of work, which translates to a minimum energy use.

It translates to a maximum cost per hash calculation, which becomes a minimum level of energy efficiency.

Does the efficiencies discovered not get distributed to the rest of the economy? My point was that it could lead to developments in energy use beneficial in separate domains.

The more efficient your hardware is the more profit you make per block (because you buy less electricity per block reward).

If all hardware was equivalent and an unlimited amount were available efficiency wouldn't matter, but neither of those things are true.

Wouldn't the difficult per block rise, making it the same current usage as before?

Eventually. But newer hardware will tend to be the most efficient component of the hashrate and push out older hardware by generating more hashes per unit of electricity (hence the CPU->GPU->ASIC progression).

That doesn't mean anything.

Bitcoin network bandwidth remains the same, and energy usage has continued to increase in a race to the bottom in terms of work and waste for bitcoin mining - this is by design of how Satoshi's difficulty algorithm reduces efficiency with increase in hash rates.

The difficulty rises with the total hashrate across all miners, not on the individual hashrate/efficiency of any particular miner - if you can eke out a temporary efficiency advantage over everybody else, you can profit nicely while they're trying to catch up.

You're thinking from the perspective of the miners.

From the perspective of the hardware manufacturers that's an advantage. If you're making the hardware and you come up with a 10% energy savings, everyone must now replace their old hardware or be left in the dust. Hardware manufacturers will make a killing. The prisoner's dilemma you are thinking of only applies to the miners, it's a benefit for the parties selling them the hardware.

> Hardware manufacturers will make a killing. The prisoner's dilemma you are thinking of only applies to the miners

Something something shovels in a gold rush...

I think the only energy efficiency PoW can hope for is the demise of traditional banks and what energy they consume.

There would definitely be a first mover advantage. On the other hand, for most other VLSI applications, improved energy usage is a PERMANENT advantage: If e.g. Apple improves energy efficiency of Annn chips, it enables them to build better devices, even if the rest of the industry catches up.

For Bitcoin, my understanding is that this is not the case, because any cost improvement in the industry will attract more miners until difficulty adjusts to the previous equilibrium where (cost of mining a bitcoin) = (value of a bitcoin)-(some risk premium)

Yeah - that "first mover advantage" can be lucrative though.

Simplistically they're betting on:

(cost to you of mining a bitcoin)-(cost to everybody else of mining a bitcoin) * (time it takes to reach equilibrium)

being a very big number...

It's first mover advantage in the same way surviving is a first mover advantage. You're fine if you're profitable, but with constant improvements that won't be the case for long, unless you too start moving to new technology.

>Apply that to energy efficiency research as well. The monetary incentives are too strong for it not to happen.

now i know where the fusion and cheap space access are to come from. Finally we've got real economic incentives for both.

>> Huh, maybe the dark fibre of Bitcoin will be cheap silicon and subsidized fab research.

EUV lithography subsidized by a pyramid scheme based on a bubble in virtual currency!

>Huh, maybe the dark fibre of Bitcoin will be cheap silicon and subsidized fab research.

If you think that the world needs ever more efficient ASICs for computing sha256 hashes, then sure.

Where is the source for how many wafers nvidia buys? That's usually confidential information.

Well, it's bounded above by how many GPUs are sold, no?

I suppose could estimate if I knew the number of GPUs sold but I've only seen estimates of that. As far as I know nVidia doesn't give their shipment numbers.

Keep in mind, though, that the constraints for Bitcoin mining are a lot different. The Bitmain chips are far smaller than the NVIDIA ones to increase yield, so the absolute chip counts will be higher but NVIDIA is probably still ahead in terms of area shipped (latency is totally unimportant so no reason to have gigantic chips).

For those that don't know, BitMain's miners are compact servers with an ARM processor and almost 190 SHA256-specialized ASIC chips:

> the S9, which had 189 of the company’s BM1387 chips [1]

[1] https://news.bitcoin.com/chinese-bitcoin-miner-might-be-tryi...

Pretty sure quantum tunneling starts to come into play under ~6 nm (hard limit of usual consumer processor) Do these do anything different to be able to aim for 7nm so fast ?

I thought it was 4 or 5?

How long do you think it takes them to amortize them?

At current prices and difficulty probably on the order of 6-12 months.

>How long do you think it takes them to amortize them?

About 6 months at current prices with cheap electricity (<$0.10/kWh)

https://shop.bitmain.com/productDetail.htm?pid=0002018010910...

https://www.cryptocompare.com/mining/calculator/btc?HashingP...

Reply on news.ycombinator.com