Question New eth asics arrive!!!

[Fallen Kell]

Woo-hoo. Finally might get some relief on GPUs as ethereum asics are finally here! They mine as fast as 32x3080 GPUs and I have to believe won't cost anywhere near the $35,000 or more it takes to buy 32x3080 cards at this point.

New Ethereum ASIC Mines Like 32 GeForce RTX 3080 GPUs

The ultimate Ethereum miner

www.tomshardware.com

 

[VirtualLarry]

just about to post that.

 

[GodisanAtheist]

I doubt the supply of these things is going to be nearly enough to offset the current GPU situation. They are, after all, reliant on all the same parts and silicon that are currently underlying the GPU shortage.

Additionally, until Eth implodes or becomes resistant to GPUs, all this does is compliment existing GPU mining operations, not subvert them. Nothing stopping someone from buying these and GPUs so long as it stays profitable to do so...

 

[beginner99]

I doubt the supply of these things is going to be nearly enough to offset the current GPU situation. They are, after all, reliant on all the same parts and silicon that are currently underlying the GPU shortage.

I it takes far less silicon (eg. wafers) for the same amount of compute power (for ETH), then it certainly does help. Unless they miners still buy up everything...

 

[fleshconsumed]

I doubt the supply of these things is going to be nearly enough to offset the current GPU situation. They are, after all, reliant on all the same parts and silicon that are currently underlying the GPU shortage.

Additionally, until Eth implodes or becomes resistant to GPUs, all this does is compliment existing GPU mining operations, not subvert them. Nothing stopping someone from buying these and GPUs so long as it stays profitable to do so...

When BTC ASICs hit the market in numbers there was noticeable increase in difficulty which hit GPU profitability. Yes, any ETH ASIC will keep pressure on memory supply, but they won't need much silicon for GPU part. If the new ETH ASICs are as good as advertised I can see light at the end of the tunnel in about 6 months even if ETH prices stay level/go up.

 

[GodisanAtheist]

I it takes far less silicon (eg. wafers) for the same amount of compute power (for ETH), then it certainly does help. Unless they miners still buy up everything...

- Fair, but then the ASIC manufacturers likely don't have the same pull our clout to ensure their products get the supply as the bigger players do.

Less product required, but perhaps less access to the product as well...

Regardless, any news on this front is good news. Take coins to warp speed so GPUs become unprofitable faster for what its worth...

 

[Mopetar]

Part of the rise in popularity of Etherium was the result of BitCoin becoming dominated by ASICs. What's to stop history from repeating itself once again and GPU miners finding a new currency to mine?

 

[Fallen Kell]

Part of the rise in popularity of Etherium was the result of BitCoin becoming dominated by ASICs. What's to stop history from repeating itself once again and GPU miners finding a new currency to mine?

The thing that prevents it is that the house of cards that is the foundation of cryptocurrency value in the first place collapses. Once people start to realize that there is in fact infinite supply of cryptocurrency, it has no value due to laws of supply/demand. If there are an infinite number of different types of cryptocurrency (which is what happens as people switch to different ones), there can never be any value in it.

 

[Mopetar]

The thing that prevents it is that the house of cards that is the foundation of cryptocurrency value in the first place collapses. Once people start to realize that there is in fact infinite supply of cryptocurrency, it has no value due to laws of supply/demand. If there are an infinite number of different types of cryptocurrency (which is what happens as people switch to different ones), there can never be any value in it.

What you're saying isn't unique to cryptocurrencies. The same holds true for any fiat currency which could also exist in infinite varieties and or could be issued in infinite amounts, but it doesn't make anyone richer because the goods and services which the currency serves as a means of exchange for are themselves still finite. We could say the same of the stock market as well. There could be an infinite number of different stocks, but they still have meaning and value based on something physical. Even a cryptocurrency which seems completely nebulous and intangible still requires physical computational work to be done for the currency to exist and there's a finite amount of that available.

A currency or any kind only has real value because people are willing to accept it in exchange for something else. If there are a lot of GPU miners who are willing to use that new cryptocurrency as a means of exchange they can immediately create some of that value through their own willingness to use it. If they have a big enough mass of users, it could attract other people to use it as well which adds some additional value to the currency in the sense that it's useful for fulfilling it's purpose.

 

[ThatBuzzkiller]

This will massively improve the supply of GPUs once we think about it ...

A GA102 die has a very high logic circuit footprint and we're talking about Ethereum mining which isn't logic bound at all so Bitmain can probably get away with using far smaller specialized logic die. As for memory, Ethereum currently has a DAG file size of roughly ~4.2GB where it is projected to reach over 5GB in a little under a year from now so instead of wasting 2/3GB worth of memory modules for every GPU, we can equip *just the right amount* of memory for these ASICs for a reasonable time that they'll have to remain viable. We're saving a moderate amount of memory chips in this process ...

Ethereum mining on GPUs could already be made obsolete if all Ethereum ASIC designers could combine their ASICs with HBM2E memory ... (Ethereum mining boils down to having necessary amount of memory capacity to store the DAG file size and getting as much memory bandwidth as possible after that)