Bitcoin Miners enjoy 3nm Chips from Samsung

[348Judah]

Energy use in bitcoin mining has been a thing of consideration as the challenges is becoming contending on miners from the government, but all thanks to research and development from various advanced technological companies and individuals that has been providing support to help tackle the energy utilization and demand, this also is one of the good news from Samsung Tech company as it develops a 3nm chips technology to be use in bitcoin mining. A statement from its official webpage says.

Samsung said the first application of 3nm chips would be for high-performance and low-power computing, with plans to expand to mobile processors.
https://forkast.news/headlines/samsung-produces-3nm-chips-bitcoin-mining/

On this note, it is believed that the production of the 3 nanometer chip will cause a relieve on miners from the current situation of the bitcoin high cost for energy demand in mining and the dip in the present market price, according to the Samsung, it said this will bring a dropdown in the amount of energy consumption and increasing the overal efficiency of bitcoin mining from the 3nm chips.

[1714105040]

1714105040

[1714105040]

1714105040

[jackg]

The article itself mentioned nothing about bitcoin or these chips being used for that. I think 3nm chip technology is a fairly new space though and it'll probably take a while for bitcoin to start using 3nm chips in ASICs.

According to a quick search, I can find most chips are 7nm or 10nm (bitcoin currently mostly uses 7nm chips afaik - there are also 5nm chips on your article that haven't even been produced for crypto mining yet too). Realistically, all three should be coming at some point, it's just a case of when (and if these chips will be able to provide the same operations the larger chips can).

[DaveF]

3nm is years away from being used in miners.
No matter how important we like to think we are, mining chips are small runs at best compared to what is needed in for other applications.

Everything needs chips, the smaller and lower power the better. Even with an S19 using 340 chips Samsung probably uses more chips internally to make their consumer products in a day then Bitmain uses in a year (a bit of an exaggeration but you get the point)

But still good in general as progress marches on, lower power for faster is always better.

-Dave

[odolvlobo]

On this note, it is believed that the production of the 3 nanometer chip will cause a relieve on miners from the current situation of the bitcoin high cost for energy demand in mining and the dip in the present market price, according to the Samsung, it said this will bring a dropdown in the amount of energy consumption and increasing the overal efficiency of bitcoin mining from the 3nm chips.

Because of the economics of Bitcoin mining, improved efficiency does not reduce the overall energy consumption. Bitcoin energy consumption depends primarily on the cost of energy and the value of the block reward, and efficiency affects neither of those.

[dansus021]

as far as i know, is most miner using 7nm technology that already use by many company and I think this version is more stable and cheap than 3nm I know that less nm mean less energy but adapting 3 nm to miner I think is gonna drive the price too

[fiulpro]

The fact is to make sure this chip works and is distributed people would still have to wait, samsung itself said to wait for the second half of the 2022 they are targeting it because of TNMC a Taiwanese company therefore the one who is able to successfully complete this first would be able to take it in the market and complete further steps. I think it's not just about the technology it's also about the competition which can be very beneficial for people who are involved in mining since competition produces good cost price and also improves the overall technology.

[darkv0rt3x]

3nm is years away from being used in miners.
No matter how important we like to think we are, mining chips are small runs at best compared to what is needed in for other applications.

Everything needs chips, the smaller and lower power the better. Even with an S19 using 340 chips Samsung probably uses more chips internally to make their consumer products in a day then Bitmain uses in a year (a bit of an exaggeration but you get the point)

But still good in general as progress marches on, lower power for faster is always better.

-Dave

Regardless of what you say about Bitcoin mining being a niche in chips R&D (which is true), the fact is that the mining industry is getting more attention from major manufacturers which can only be good in terms of competition between brands and also in favour of future development.

We see graphic cards main goal is gaming and heavy processing applications but the market also has plans for mining shitcoins with them.

So, my point is that even being a niche, the chips industry community is paying more and more attention to the mining industry!

[stompix]

So, my point is that even being a niche, the chips industry community is paying more and more attention to the mining industry!

Samsung has been making chips for ASICs for years, it's not something new, it was a demand that has driven up prices for those and that's what caught the attention of foundries as mining gear manufacturers were able to pay a heft premium for each chip. Now that revenues from mining are 1/3 of what they were it means less income for miners, less expensive gear being sold, less revenue per gear, and less money to overpay for fabs.

I think they are paying less and less attention since they already know energy prices are pushing miners out of the game and that in order to sell thems you need to make stuff that is both profitable for you, for the manufacturer, and for the miner. When selling a phone, you don't have to go through all this math, and phone sales won't drop by 50% cause some idiot Ponzi schemer ran a company into bankruptcy.
Also, you don't change your ASIC because this year's new model has a better camera and more battery life or a better screen!

We see graphic cards main goal is gaming and heavy processing applications but the market also has plans for mining shitcoins with them.

And we see Nviaida limiting their cards to be less attractive to miners....

3nm is years away from being used in miners.
No matter how important we like to think we are, mining chips are small runs at best compared to what is needed in for other applications.

With TSMC alone selling in one year 7 years of bitcoin mining revenue worth of chips it's a tiny drop, maybe not in the ocean but still in quite a large beer stein.

[BlackHatCoiner]

Bitcoin energy consumption depends primarily on the cost of energy and the value of the block reward, and efficiency affects neither of those.

Not sure if I understand this sentence. Isn't efficiency strictly correlated with the cost? If you find a more efficient way to calculate the hashes, it means you've found a cheaper way. If using the same amount of energy, all the miners upgraded to more efficient devices, difficulty would increase and, therefore, so would the energy required per bitcoin (AKA, energy consumption).

[stompix]

Bitcoin energy consumption depends primarily on the cost of energy and the value of the block reward, and efficiency affects neither of those.

Not sure if I understand this sentence. Isn't efficiency strictly correlated with the cost? If you find a more efficient way to calculate the hashes, it means you've found a cheaper way.

What he is saying is that the reward dictates how much money miners could theoretically spend in order to keep being profitable, the reward is the cap on energy consumption.  Efficiency matters only in terms of hashrate generated with a $x reward per block, or for personal finances not global.

If using the same amount of energy, all the miners upgraded to more efficient devices, difficulty would increase and, therefore, so would the energy required per bitcoin (AKA, energy consumption).

No, it doesn't work that way, if everyone would replace a gear rated 300W and 100th/s with one rated 3000w and 200th/s the consumption will not grow, the margin rate will neither since you get the same share for the same cost, the hash rate will go up but without increased energy needs.
The income per th/s vs cost per th/s generated is affecting global energy consumption, if the revenue is higher more miners will put gear to mine, if it goes negative so does the hashrate, as it's happening now.

[franky1]

3nm is years away from being used in miners.

as far as i know, is most miner using 7nm technology that already use by many company

TSMC started doing 7nm in 2016. but bitmain waited a couple years before renting foundry fabricator time and started only making 7nm batches of chips in 2017-8(the s15 range BM1391 chips)

why did they wait and not just buy some fabricator time slots in 2016.. obvious answer.. its cheaper to wait it out a bit.
(let the fabricator rental costs come down a bit)
yes the efficiency gain of watts/hash is 45% better in the 3nm compared to current nm chips
but if the chip itself(due to production cost) is more. then the efficiency:economics does not equate to be any more rewarding, and instead more costly to mine on new equipment if the equipment cost is too high

take this years batch
they have 2 variants of 19 generation ill use as an example

https://shop.bitmain.com/product/detail?pid=00020220711185109085o99L6nF90644
Bitcoin Miner T19 Hydro - 145thash - 5438watt  hardware cost $2,175

https://shop.bitmain.com/product/detail?pid=00020220210152002039CHAR8f2i0636
Bitcoin Miner S19 XP - 140thash - 3010watt hardware cost $11,620

as you can see the wattages are different by far but the hashrate are about the same (3%+-margin of error)
but the hardware costs are more noticable.

now guess which one is cheaper to mine with if based on a 4cent/kw electric cost and spreading the hardware across 2 years
answer below(scroll over to see white text)
[answer:

answer. T19 wins even though it uses 5kw instead of the s19 at 3kw

math:
hashrate 210000000thash               
T19 -145thash 5.4kw:               
1448275.862asics running to produce 210exahash
7875724.138kw used per hour to do 210exa
$315028.97 spent per hour at $0.04 electric rate
$8400.77electric cost per btc
$3150000000 hardware cost of 1.448m asics
$4800 hardware cost per btc, where costs are spread over 2 years (total cost/105k blocks/6.25coin)
so cost per btc of electric+hardware=
$13.2k


s19XP - 140thash 3kw      
1500000 asics running to produce 210exahash
4515000kw used per hour to do 210exa
$180600 spent per hour at $0.04 electric rate
$4816electric cost per btc
17430000000 hardware cost of 1.5m asics
$26560 hardware cost per btc, where costs are spread over 2 years (total cost/105k blocks/6.25coin)
so cost per btc of electric+hardware=
31376

]

[odolvlobo]

Bitcoin energy consumption depends primarily on the cost of energy and the value of the block reward, and efficiency affects neither of those.

Not sure if I understand this sentence. Isn't efficiency strictly correlated with the cost? If you find a more efficient way to calculate the hashes, it means you've found a cheaper way. If using the same amount of energy, all the miners upgraded to more efficient devices, difficulty would increase and, therefore, so would the energy required per bitcoin (AKA, energy consumption).

By "cost of energy", I mean the cost to purchase a joule of energy. In short, the energy consumption (in joules) to produce a bitcoin, E, approaches

E ⟶  P  / C

where

P is the average price of a bitcoin in $
C is the average cost of electricity in $/J

The reason that hash rate and efficiency are irrelevant is that the difficulty adjustments keep production constant, regardless of the hash rate, the amount of energy used, the efficiency of the equipment, and the amount miners are willing to spend.

However, assuming that miners (in the aggregate) are striving to be profitable, they will spend as much as they can on energy up to the value of the bitcoins they produce. So, the value of the energy consumed producing a bitcoin (E x C) approaches the value of the bitcoin (P). Thus, E ⟶  P  / C. Note that there are other variable costs, but energy is the biggest and it's what we are concerned about here.

If using the same amount of energy, all the miners upgraded to more efficient devices, difficulty would increase and, therefore, so would the energy required per bitcoin (AKA, energy consumption).

Small contradiction there. If using the same amount of energy, the energy required per bitcoin would be the same and would not increase. Note that I am assuming a time frame that includes difficulty adjustments and time for miners to add or remove equipment. An increase in efficiency would reduce energy consumption at least until the next difficulty adjustment, but consumption would return as miners compete for a bigger share.

[BlackHatCoiner]

No, it doesn't work that way, if everyone would replace a gear rated 300W and 100th/s with one rated 3000w and 200th/s the consumption will not grow, the margin rate will neither since you get the same share for the same cost, the hash rate will go up but without increased energy needs.

Right, the energy required per bitcoin does not change based on difficulty. What does change, based on it, is the hashes required per bitcoin. But there isn't a strict correlation between the cost of joule and the cost of calculating a hash.

Small contradiction there. If using the same amount of energy, the energy required per bitcoin would be the same and would not increase.

Let me rephrase: If using the same amount of energy, all miners upgraded to more efficient devices, difficulty would increase and, therefore, so would the hashes per bitcoin. So, what does change (specifically decreases) when mining becomes more efficient is the energy required to calculating 1 hash.

[darkv0rt3x]

So, my point is that even being a niche, the chips industry community is paying more and more attention to the mining industry!

Samsung has been making chips for ASICs for years, it's not something new, it was a demand that has driven up prices for those and that's what caught the attention of foundries as mining gear manufacturers were able to pay a heft premium for each chip. Now that revenues from mining are 1/3 of what they were it means less income for miners, less expensive gear being sold, less revenue per gear, and less money to overpay for fabs.

I think they are paying less and less attention since they already know energy prices are pushing miners out of the game and that in order to sell thems you need to make stuff that is both profitable for you, for the manufacturer, and for the miner. When selling a phone, you don't have to go through all this math, and phone sales won't drop by 50% cause some idiot Ponzi schemer ran a company into bankruptcy.
Also, you don't change your ASIC because this year's new model has a better camera and more battery life or a better screen!

We see graphic cards main goal is gaming and heavy processing applications but the market also has plans for mining shitcoins with them.

And we see Nviaida limiting their cards to be less attractive to miners....

3nm is years away from being used in miners.
No matter how important we like to think we are, mining chips are small runs at best compared to what is needed in for other applications.

With TSMC alone selling in one year 7 years of bitcoin mining revenue worth of chips it's a tiny drop, maybe not in the ocean but still in quite a large beer stein.

Yes, but the point is also that big brands keep their I&D teams working on these chips, means the there is market for them.

The fact that you think they are paying less attention to mining industry seems to be more of a misunderstanding about Bitcoin. But I believetheu unserstand, at keats part of it.