Yeah because a network secured by gpu is going to go places right! cmon scrypt n is just for the miners chasing the dragon still...how about we start really innovating and learning to use the ledger..we already have enough currency type coins! you guys will go and go and go diluting the markets with you gpu cpu coins untill one day they are all worth 0 because thats all everyone is doing is chasing 100% profit all day long and not really contributing to any innovation
It is clear that people don't care about securing networks and creating something that works and is efficient and uses little to no power. People want money now and are expiditing the time where it drys up and they will move on. Leaving people who contributed to asic and proper innovation to prosper long into the future
Topic: is N-Scrypt.. really worth considering? looks full of flaws logically... - page 3. (Read 5677 times)
March 24, 2014, 06:44:18 AM
March 24, 2014, 06:40:23 AM
Quote
Terry L :
i haven't got into the scrypt N coins yet but to put 2GB ram (i would guess you mean sram, not external dram) in one 55nm ASIC is impossible (QFN) at this stage (unless using BGA with a humongous sized chip like the CPU or GPU). it's because ram itself takes huge amount of space. however if you are opting for 16nm and BGA, it might be a good bet but considering the kind of money you have to put into design and development as well as the yield, it's way less economical comparing to gpu.
using fpga and dram is a fine alternative for now.
i haven't got into the scrypt N coins yet but to put 2GB ram (i would guess you mean sram, not external dram) in one 55nm ASIC is impossible (QFN) at this stage (unless using BGA with a humongous sized chip like the CPU or GPU). it's because ram itself takes huge amount of space. however if you are opting for 16nm and BGA, it might be a good bet but considering the kind of money you have to put into design and development as well as the yield, it's way less economical comparing to gpu.
using fpga and dram is a fine alternative for now.
Yeah, that guy just say you can easily add DRAM.
Who will add SRAM into it anyway? The OP also means adding DRAM is really easy and I second that.
N-Scrypt is USELESS, as the above Terry L already mentioned by saying, it is possible to add DRAM.
N-scrypt is not worth it at all, there's no such thing as ASIC resistance, people should jsut get use to that already.
Look at Bitcoin and where it got it. (skyrocketing)
when ASIC comes out, suddenly, N-scrypt coins will have the lowest hashrate of the entire world.
and unless a coin comes out with a really good network, it will just be thrown away completely.
Indeed, the keyword there is YIELD. As of right now, there's no reason to add it, but it will be added if ever there are more coins with N-scrypt.
also : using fpga and dram is a fine alternative for now.
Indeed DRAM is possible and practical, and makes N-scrypt kinda pointless.
March 24, 2014, 06:21:05 AM
Quote
Terry L :
i haven't got into the scrypt N coins yet but to put 2GB ram (i would guess you mean sram, not external dram) in one 55nm ASIC is impossible (QFN) at this stage (unless using BGA with a humongous sized chip like the CPU or GPU). it's because ram itself takes huge amount of space. however if you are opting for 16nm and BGA, it might be a good bet but considering the kind of money you have to put into design and development as well as the yield, it's way less economical comparing to gpu.
using fpga and dram is a fine alternative for now.
i haven't got into the scrypt N coins yet but to put 2GB ram (i would guess you mean sram, not external dram) in one 55nm ASIC is impossible (QFN) at this stage (unless using BGA with a humongous sized chip like the CPU or GPU). it's because ram itself takes huge amount of space. however if you are opting for 16nm and BGA, it might be a good bet but considering the kind of money you have to put into design and development as well as the yield, it's way less economical comparing to gpu.
using fpga and dram is a fine alternative for now.
Yeah, that guy just say you can easily add DRAM.
Who will add SRAM into it anyway? The OP also means adding DRAM is really easy and I second that.
N-Scrypt is USELESS, as the above Terry L already mentioned by saying, it is possible to add DRAM.
N-scrypt is not worth it at all, there's no such thing as ASIC resistance, people should jsut get use to that already.
Look at Bitcoin and where it got it. (skyrocketing)
when ASIC comes out, suddenly, N-scrypt coins will have the lowest hashrate of the entire world.
and unless a coin comes out with a really good network, it will just be thrown away completely.
March 24, 2014, 06:17:08 AM
There's no point to it other than 'marketing', it's just another scam.
March 24, 2014, 05:47:23 AM
Quote
Terry L :
i haven't got into the scrypt N coins yet but to put 2GB ram (i would guess you mean sram, not external dram) in one 55nm ASIC is impossible (QFN) at this stage (unless using BGA with a humongous sized chip like the CPU or GPU). it's because ram itself takes huge amount of space. however if you are opting for 16nm and BGA, it might be a good bet but considering the kind of money you have to put into design and development as well as the yield, it's way less economical comparing to gpu.
using fpga and dram is a fine alternative for now.
i haven't got into the scrypt N coins yet but to put 2GB ram (i would guess you mean sram, not external dram) in one 55nm ASIC is impossible (QFN) at this stage (unless using BGA with a humongous sized chip like the CPU or GPU). it's because ram itself takes huge amount of space. however if you are opting for 16nm and BGA, it might be a good bet but considering the kind of money you have to put into design and development as well as the yield, it's way less economical comparing to gpu.
using fpga and dram is a fine alternative for now.
Yeah, that guy just say you can easily add DRAM.
Who will add SRAM into it anyway? The OP also means adding DRAM is really easy and I second that.
N-Scrypt is USELESS, as the above Terry L already mentioned by saying, it is possible to add DRAM.
N-scrypt is not worth it at all, there's no such thing as ASIC resistance, people should jsut get use to that already.
Look at Bitcoin and where it got it. (skyrocketing)
when ASIC comes out, suddenly, N-scrypt coins will have the lowest hashrate of the entire world.
and unless a coin comes out with a really good network, it will just be thrown away completely.
March 24, 2014, 05:31:39 AM
I believed a member here once asked about developing an ASIC miner which will take into account scrypt-N by making an ASIC with lotsof RAM.
I've also seen posts that scrypt-N is useless and an ASIC maker may design their miner to mitigate adaptive-N with ease.
I quote a post from litecointalk from Terry L who is developing a seemingly reliable ASIC scrypt ZeusMiner. In summary he stated that it isn't as easy as it sounds mainly being cost prohibitive for now...
I've also seen posts that scrypt-N is useless and an ASIC maker may design their miner to mitigate adaptive-N with ease.
I quote a post from litecointalk from Terry L who is developing a seemingly reliable ASIC scrypt ZeusMiner. In summary he stated that it isn't as easy as it sounds mainly being cost prohibitive for now...
Quote
from: Xardas on Today at 11:31:08 AM
*I had several sha256 miners(which someone bought from me for a ridiculous price) and they would also do scrypt. They weren't very efficient at it because the guys had done some kind of coding that made it work. I can only assume they somehow used external memory(off-chip) that was on the units.
Quote
from: Xardas on Today at 11:21:00 AM
Terry I asked a question of jasinlee in another thread, but maybe you are still online and can answer. How difficult/expensive would it be to put 2GB of memory on an ASIC chip?
The reason I ask is to make a miner that can handle Scrypt N which needs a lot memory. If an ASIC had a sufficient amount of memory on it or coded to use external memory*, it would prevent the devs of coins like Vertcoin from raising the memory requirement of N too high. Most vid card have 2-4GB of memory and my guess the majority have 2GB, so that would limit how high they can set N without losing vast amounts of miners.Terry I asked a question of jasinlee in another thread, but maybe you are still online and can answer. How difficult/expensive would it be to put 2GB of memory on an ASIC chip?
*I had several sha256 miners(which someone bought from me for a ridiculous price) and they would also do scrypt. They weren't very efficient at it because the guys had done some kind of coding that made it work. I can only assume they somehow used external memory(off-chip) that was on the units.
Quote
Terry L : i haven't got into the scrypt N coins yet but to put 2GB ram (i would guess you mean sram, not external dram) in one 55nm ASIC is impossible (QFN) at this stage (unless using BGA with a humongous sized chip like the CPU or GPU). it's because ram itself takes huge amount of space. however if you are opting for 16nm and BGA, it might be a good bet but considering the kind of money you have to put into design and development as well as the yield, it's way less economical comparing to gpu.
using fpga and dram is a fine alternative for now.
using fpga and dram is a fine alternative for now.
March 24, 2014, 05:27:41 AM
Long term Scrypt-Jane and Adaptive-N do not make sense, each step up in N factor we see a halving of hashpower and the difficulty in Vertcoin can not even go low enough to support the current hash rate at 2^32, i.e 4GB.
Please take a look at Hirocoin, it uses the X11 hashing solution which sees triple hash power over Scrypt with a lot less energy and heat used. Long term it will win out over coins that have variable Scrypt N factor.
Please take a look at Hirocoin, it uses the X11 hashing solution which sees triple hash power over Scrypt with a lot less energy and heat used. Long term it will win out over coins that have variable Scrypt N factor.
heavycoin algo is even better than x11
2° less compared to x11
March 24, 2014, 05:25:49 AM
the problem is not the RAM size, but the access time to RAM (latency) and bandwith, these are actually are bottle neck, that's why adding more RAM to ASIC will not help
March 24, 2014, 05:24:53 AM
VTC has a policy of changing the alg to kill ASICs, making any Scrypt-N ASIC project financial suicide.
you can't just fork a coin that easy. What if certain people don't switch , then you have 2 vertcoins.
March 24, 2014, 05:19:30 AM
Long term Scrypt-Jane and Adaptive-N do not make sense, each step up in N factor we see a halving of hashpower and the difficulty in Vertcoin can not even go low enough to support the current hash rate at 2^32, i.e 4GB.
Please take a look at Hirocoin, it uses the X11 hashing solution which sees triple hash power over Scrypt with a lot less energy and heat used. Long term it will win out over coins that have variable Scrypt N factor.
Please take a look at Hirocoin, it uses the X11 hashing solution which sees triple hash power over Scrypt with a lot less energy and heat used. Long term it will win out over coins that have variable Scrypt N factor.
March 24, 2014, 05:07:33 AM
VTC has a policy of changing the alg to kill ASICs, making any Scrypt-N ASIC project financial suicide.
March 24, 2014, 04:56:42 AM
Look this way.
RAM is cheap, way cheaper than most components in an ASIC device, the only reason they are not there, is caused people didn't add more than required.
Example, if more coins use N-scrypt, then it will just be another year before ASIC includes them too.
heck, if I am a scrypt ASIC developer, (which I am not, I do something else related), I will just add that to my device and call it a perk.
I have looked at many angles.. but I cannot see how N-scrypt is even going to be any different from Scrypt.
How is that .. ASIC resistant? I mean, Litecoin was 'suppose' to be the ASIC resistance to SHA-256, look now.
and N-Scrypt is Waaaaaaaay easier to add comparing to the switch from Sha256 to Scrypt..
can anyone give me a point in the right direction?
but right now, it looks like just a marketing theme only.. am i right?
Jump to: