Scrypt (Litecoin) ASIC Prototype
-
There is a thread posted on the Bitcointalk.org forum recently on the subject. To keep the story short, three guys have spent a year designing a scrypt ASIC prototype for personal use, but lost their project funds in a recent theft. They’re all broken and asking for donations or any other financial support. As expected, they didn’t receive a warm response there. Should we invite them to our forum for a more productive dialogue?
[url=https://bitcointalk.org/index.php?topic=231185.0]Scrypt (Litecoin) ASIC Prototype[/url]
-
I like it.But to get it off the ground it could cost hundreds of thousands!!!
But it would be a great project! -
-
“I had a million dollars doing something cool with it, but it got stolen, please donate a million to me so I can do it again” ~~~ smells like a pump and dump right into their wallets
-
[quote name=“RIPPEDDRAGON” post=“14750” timestamp=“1371215906”]
“I had a million dollars doing something cool with it, but it got stolen, please donate a million to me so I can do it again” ~~~ smells like a pump and dump right into their wallets
[/quote]EVen if it is not, why would we encourage ASIC scrypt mining? If anything we would put funds for code development to make it difficult for ASIC mining.
-
[quote name=“ghostlander” post=“14725” timestamp=“1371213488”]
To keep the story short, three guys have spent a year designing a scrypt ASIC prototype for personal use, but lost their project funds in a recent theft.
[/quote]Where are they from? Nigeria? ;)
Quote from the original post on Bitcointalk: “Lately he [i][Chinese student][/i] suddenly disappeared and took all our coin money which was around 100BTC and 7000LTC!!!”.
30,000 US$ was all they had and they developed an ASIC prototype? It takes at least 20 times this amount to even get started.
-
I have no doubt that Scrypt ASICs will one day hit us. However the post linked from BitcoinTalk is amusing and is a scam.
For our grass roots community we would want to stay away from dedicated mining hardware and stay in the remit of GPU miners. This would probably take the form of a move to Scrpyt-Jane. At that point Bitcoin and Litecoin could rule SHA-256 and Scrypt respectively while we could rule Scrypt-Jane :)
-
[quote name=“Magic8Ball” post=“14752” timestamp=“1371216003”]
[quote author=RIPPEDDRAGON link=topic=1829.msg14750#msg14750 date=1371215906]
“I had a million dollars doing something cool with it, but it got stolen, please donate a million to me so I can do it again” ~~~ smells like a pump and dump right into their wallets
[/quote]EVen if it is not, why would we encourage ASIC scrypt mining? If anything we would put funds for code development to make it difficult for ASIC mining.
[/quote]Scrypt ASICs will appear eventually whether you like them or not. It’s just a matter of time.
[quote author=Radacoin link=topic=1829.msg14754#msg14754 date=1371216218]
[quote author=ghostlander link=topic=1829.msg14725#msg14725 date=1371213488]
To keep the story short, three guys have spent a year designing a scrypt ASIC prototype for personal use, but lost their project funds in a recent theft.
[/quote]Where are they from? Nigeria? ;)
Quote from his post on Bitcointalk: “Lately he suddenly disappeared and took all our coin money which was around 100BTC and 7000LTC!!!”.
30,000 US$ was all they had and they developed an ASIC prototype? It takes at least 20 times this amount to even get started.
[/quote]That’s more than enough to design and debug a prototype. High expences start at the manufacturing stage when you order a full mask. Although BitFountain (ASICminer) has managed to reduce costs significantly by using multilevel masks.
-
[quote name=“ghostlander” post=“14758” timestamp=“1371216526”]
[quote author=Magic8Ball link=topic=1829.msg14752#msg14752 date=1371216003]
[quote author=RIPPEDDRAGON link=topic=1829.msg14750#msg14750 date=1371215906]
“I had a million dollars doing something cool with it, but it got stolen, please donate a million to me so I can do it again” ~~~ smells like a pump and dump right into their wallets
[/quote]Even if it is not, why would we encourage ASIC scrypt mining? If anything we would put funds for code development to make it difficult for ASIC mining.
[/quote]Scrypt ASICs will appear eventually whether you like them or not. It’s just a matter of time.
[/quote]Of course; doesn’t mean we will encourage or even fund them.
We should be focussing on updating our code to keep GPU miners happy.
-
[quote name=“Bushstar” post=“14757” timestamp=“1371216471”]
For our grass roots community we would want to stay away from dedicated mining hardware and stay in the remit of GPU miners. This would probably take the form of a move to Scrpyt-Jane. At that point Bitcoin and Litecoin could rule SHA-256 and Scrypt respectively while we could rule Scrypt-Jane :)
[/quote]That’s a bad idea. GPU performance degrades very much with high N. Considering a 32->1024 change, regular CPU performance decreases by about 10 times while GPU performance drops by about 20 times. Further N increase results in hash rate performance getting dependent on memory bandwidth nearly completely. A very inefficient approach.
-
[quote name=“ghostlander” post=“14758” timestamp=“1371216526”]
That’s more than enough to design and debug a [b]prototype[/b]. High expences start at the manufacturing stage when you order a full mask.
[/quote]Their posting starts with: “In a nutshell: we designed and built an [b]ASIC prototype[/b] …”.
It’s not like you take a breadboard and put some resistors, capacitors and ICs on it.
An ASIC prototype is always design [b]and[/b] production of a real piece of silicon. And that requires masks and lots of production stept. And is bloody expensive.
-
[quote name=“ghostlander” post=“14768” timestamp=“1371217641”]
[quote author=Bushstar link=topic=1829.msg14757#msg14757 date=1371216471]
For our grass roots community we would want to stay away from dedicated mining hardware and stay in the remit of GPU miners. This would probably take the form of a move to Scrpyt-Jane. At that point Bitcoin and Litecoin could rule SHA-256 and Scrypt respectively while we could rule Scrypt-Jane :)
[/quote]That’s a bad idea. GPU performance degrades very much with high N. Considering a 32->1024 change, regular CPU performance decreases by about 10 times while GPU performance drops by about 20 times. Further N increase results in hash rate performance getting dependent on memory bandwidth nearly completely. A very inefficient approach.
[/quote]Certainly we want an efficient hashing solution. We want something that will resist Scrypt ASICs and still be GPU friendly. I believe we can set the maximum of N. I will soon setup a testnet with four nodes and 5,000KH power to test out some different solutions. I’m currently involved in other high priority tasks so if anyone want to do some research on Scrypt-Jane and report what is possible in a new thread for the community that would be appreciated :)
-
Fixed that for you:
[quote name=“ghostlander” post=“14768” timestamp=“1371217641”]
[quote author=Bushstar link=topic=1829.msg14757#msg14757 date=1371216471]
For our grass roots community we would want to stay away from dedicated mining hardware and stay in the remit of GPU miners. This would probably take the form of a move to Scrpyt-Jane. At that point Bitcoin and Litecoin could rule SHA-256 and Scrypt respectively while we could rule Scrypt-Jane :)
[/quote][s]That’s a bad idea.[/s] [color=red]That’s a good idea.[/color] GPU performance degrades very much with high N. Considering a 32->1024 change, regular CPU performance decreases by about 10 times while GPU performance drops by about 20 times. Further N increase results in hash rate performance getting dependent on memory bandwidth nearly completely. A very inefficient approach.
[/quote]A block found with difficulty 10000 is as good as a block found with difficulty 1.
Why would we wanna have an arms race like that what’s currently going on in the Bitcoin world?
-
[quote name=“Radacoin” post=“14778” timestamp=“1371218536”]
[quote author=ghostlander link=topic=1829.msg14758#msg14758 date=1371216526]
That’s more than enough to design and debug a [b]prototype[/b]. High expences start at the manufacturing stage when you order a full mask.
[/quote]Their posting starts with: “In a nutshell: we designed and built an [b]ASIC prototype[/b] …”.
It’s not like you take a breadboard and put some resistors, capacitors and ICs on it.
An ASIC prototype is always design [b]and[/b] production of a real piece of silicon. And that requires masks and lots of production stept. And is bloody expensive.
[/quote]They had no money to order masks. What they had probably is some HDL design and basic hardware implementation using FPGAs and other industry standard logic. However, if it works as expected and they have a proof, the next stage is to prepare a business plan for the 1st round of venture capital gathering.
[quote]A block found with difficulty 10000 is as good as a block found with difficulty 1.[/quote]
I’m not sure if you understand the matters right, but N is not difficulty.
-
[quote name=“Radacoin” post=“14783” timestamp=“1371218676”]
Fixed that for you:[quote author=ghostlander link=topic=1829.msg14768#msg14768 date=1371217641]
[quote author=Bushstar link=topic=1829.msg14757#msg14757 date=1371216471]
For our grass roots community we would want to stay away from dedicated mining hardware and stay in the remit of GPU miners. This would probably take the form of a move to Scrpyt-Jane. At that point Bitcoin and Litecoin could rule SHA-256 and Scrypt respectively while we could rule Scrypt-Jane :)
[/quote][s]That’s a bad idea.[/s] [color=red]That’s a good idea.[/color] GPU performance degrades very much with high N. Considering a 32->1024 change, regular CPU performance decreases by about 10 times while GPU performance drops by about 20 times. Further N increase results in hash rate performance getting dependent on memory bandwidth nearly completely. A very inefficient approach.
[/quote]A block found with difficulty 10000 is as good as a block found with difficulty 1.
Why would we wanna have an arms race like that what’s currently going on in the Bitcoin world?
[/quote]Research thread here:
-
[quote name=“ghostlander” post=“14788” timestamp=“1371219195”]
[quote]A block found with difficulty 10000 is as good as a block found with difficulty 1.[/quote]
I’m not sure if you understand the matters right, but N is not difficulty.
[/quote]That’s right. Hence why I said difficulty and not N ::)
Why would you wanna have an arms race like in the Bitcoin world? Why would you wanna drive difficulty sky high?
Is - in your eyes - a block solved with difficulty 10000 better than a block solves with difficulty 1?
-
[quote name=“Radacoin” post=“14794” timestamp=“1371219792”]
Why would you wanna have an arms race like in the Bitcoin world? Why would you wanna drive difficulty sky high?[/quote]There won’t be any race if business is done in civilised way. If you want to stay away and hide in a cave, that’s your choice. The ASICs are coming. One day you may find out that scrypt ASICs are already here, owned by a limited group of people, and you have little to do against with your expensive power hungry GPUs.
[quote]Is - in your eyes - a block solved with difficulty 10000 better than a block solves with difficulty 1?[/quote]
Difficulty is not something to fear of. It’s just an instrument to control block generation rate.
-
[quote name=“ghostlander” post=“14820” timestamp=“1371221943”]
The ASICs are coming. One day you may find out that scrypt ASICs are already here, owned by a limited group of people, and you have little to do against with your expensive power hungry GPUs.
[/quote]You don’t get it, do you?
It will always end up being power hungry.
[b]People will always mine as long as the generated mining income will be more than the power costs.[/b]
Whether the network will be at 15 GHash/s with CPUs, 150 GHash/s with GPUs or 150,000 GHash/s with ASICS - or at 150,000,000 GHash/s with ASICS v2.0.
Why would you wanna have scrypt ASICS? Let me ask you again: Is a block solved with difficulty 10000 better than a block solves with difficulty 1?
-
[b]Radacoin[/b], you’re missing my point again and again. It doesn’t matter at all whether you, me or Satan Claus love scrypt ASICs or hate them. They’re profitable and they’re coming. Those who deliver them first get outstanding ROI. You can participate or ignore. You won’t stop them by putting a particular developer down. There are always others to come.
-
[quote name=“Radacoin” post=“14838” timestamp=“1371223530”]
[quote author=ghostlander link=topic=1829.msg14820#msg14820 date=1371221943]
The ASICs are coming. One day you may find out that scrypt ASICs are already here, owned by a limited group of people, and you have little to do against with your expensive power hungry GPUs.
[/quote]You don’t get it, do you?
It will always end up being power hungry.
[b]People will always mine as long as the generated mining income will be more than the power costs.[/b]
Whether the network will be at 15 GHash/s with CPUs, 150 GHash/s with GPUs or 150,000 GHash/s with ASICS - or at 150,000,000 GHash/s with ASICS v2.0.
Why would you wanna have scrypt ASICS? Let me ask you again: Is a block solved with difficulty 10000 better than a block solves with difficulty 1?
[/quote]I made a similar point in another thread about price v powercosts v !ASICS but I think I was missing the point also, they will come eventually, whether you or I like it is irrelevant and by not embracing it could be quite possibly naive :-\
