avastar
commented
5 years ago Just because a device has 128gb of memory, does not mean a program has access to even 2gb. Do botnets have access to 2gb of memory? I've heard that browser based botnets do not.
On Fri, Oct 4, 2019, 4:57 AM DevEidem notifications@github.com wrote:
So as we could see in the last past days, community voted with slight lead on the implementation of current version of the RandomX algorithm, however, lets analyze the decision
https://github.com/tevador/RandomX Does RandomX facilitate botnets/malware mining or web mining?
RandomX is profitable to mine only on CPUs, which open doors to big botnets, even if the author says, it needs 2 GiB of memory, which majority of average consumer devices around the globe have, even if we mention mobile phones (but that's not the issue). So we can easily say, for example a botnet with ONLY just 2000 laptops/DPC, would easily get majority of the network and open doors easily for a 51% attack. If that happens, the mining for average single CPU miners/or multiple machines and their impact will become so negligible that they will stop mining, so the only mining entity could become the botnet (who would mine, if the avg cpu gives only $0.05 a day (with botnets present on the network) with power consumption around 80 watt/h). Botnets will also push very big downward pressure on the coin price, because they don't simply care. They mine and whatever they mined out, they will DUMP it immediately, there are no thing as for example "hodl" botnet, as opposed to standard GPU miners. They will also have to sell for much higher price, because the electricity price, as opposed to botnet, which doesn't care about the price that will the botnet entity (computer of not knowing person) pay.
For all the cryptonight algos we implemented in the past, the botnets were still allowed into the network, sure, single botnet entity had small, but still considerable hashrate to give very high speeds at big numbers (which botnet does have).
There are also known other issues, which facilitates the devs of RandomX and other projects to modify it and produce a better version.
I would also like to point out the current suggestions that somebody sent to the graft team, to modify RandomX cache size or RAM needed, or what it was exactly, these steps would facilitate more botnet devices to have access for the mining and is a dangerous step in my opinion
The CN-GPU algo, on the other side, is currently dettering all of 3 dangerous centralized entities from entering the network 1) botnets , 2) ASICs, 3) FPGAs . The botnet issue is solved permanently, because almost no botnet has access to many GPUs, if any, and mining on these devices would be very hard to implement with GPU. There are currently no ASICs available for the algorithm, neither nicehash supports it. Building ASIC for this algorithm would not be easy and for the time being, for such low number and low volume coins that have it, it won't get implemented, same applies for the FPGAs
The only "issue" with CN-GPU which somebody already had pointed out, there is a very easy step to fix - adjust power settings and core clock. Everything will be balanced. Who wants to reduce the temperature and power input, will do it, giving slightly lower hashrate. Who can cool these cards, will even overclock the card, giving slightly more hashrate, as it always been before, there is nothing bad on this issue. Majority of amateur and experienced miners already did and will do modifications to their cards, so the issue is solved, we can say
There are also big number of people, who are waiting with their GPUs to enter the mining process to be balanced, and not be killed by high hashrate and free electricity botnets or some ASICS/FPGAs, so the network would be quite decentralized again
Feel free to join the discussion
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/graft-project/GraftNetwork/issues/345?email_source=notifications&email_token=AAKSHINYXGYUY72PRMGHBKTQM4OTDA5CNFSM4I5OJ25KYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HPUTWSQ, or mute the thread https://github.com/notifications/unsubscribe-auth/AAKSHILSELV7PHLYI3BF4ETQM4OTDANCNFSM4I5OJ25A .
DevEidem
fireice-uk
So as we could see in the last past days, community voted with slight lead on the implementation of current version of the RandomX algorithm, however, lets analyze the decision
https://github.com/tevador/RandomX Does RandomX facilitate botnets/malware mining or web mining?
RandomX is profitable to mine only on CPUs, which open doors to big botnets, even if the author says, it needs 2 GiB of memory, which majority of average consumer devices around the globe have, even if we mention mobile phones (but that's not the issue). So we can easily say, for example a botnet with ONLY just 2000 laptops/DPC, would easily get majority of the network and open doors easily for a 51% attack. If that happens, the mining for average single CPU miners/or multiple machines and their impact will become so negligible that they will stop mining, so the only mining entity could become the botnet (who would mine, if the avg cpu gives only $0.05 a day (with botnets present on the network) with power consumption around 80 watt/h). Botnets will also push very big downward pressure on the coin price, because they don't simply care. They mine and whatever they mined out, they will DUMP it immediately, there are no thing as for example "hodl" botnet, as opposed to standard GPU miners. They will also have to sell for much higher price, because the electricity price, as opposed to botnet, which doesn't care about the price that will the botnet entity (computer of not knowing person) pay.
For all the cryptonight algos we implemented in the past, the botnets were still allowed into the network, sure, single botnet entity had small, but still considerable hashrate to give very high speeds at big numbers (which botnet does have).
There are also known other issues, which facilitates the devs of RandomX and other projects to modify it and produce a better version.
I would also like to point out the current suggestions that somebody sent to the graft team, to modify RandomX cache size or RAM needed, or what it was exactly, these steps would facilitate more botnet devices to have access for the mining and is a dangerous step in my opinion
The CN-GPU algo, on the other side, is currently dettering all of 3 dangerous centralized entities from entering the network 1) botnets , 2) ASICs, 3) FPGAs . The botnet issue is solved permanently, because almost no botnet has access to many GPUs, if any, and mining on these devices would be very hard to implement with GPU. There are currently no ASICs available for the algorithm, neither nicehash supports it. Building ASIC for this algorithm would not be easy and for the time being, for such low number and low volume coins that have it, it won't get implemented, same applies for the FPGAs
The only "issue" with CN-GPU which somebody already had pointed out, there is a very easy step to fix - adjust power settings and core clock. Everything will be balanced. Who wants to reduce the temperature and power input, will do it, giving slightly lower hashrate. Who can cool these cards, will even overclock the card, giving slightly more hashrate, as it always been before, there is nothing bad on this issue. Majority of amateur and experienced miners already did and will do modifications to their cards, so the issue is solved, we can say
There are also big number of people, who are waiting with their GPUs to enter the mining process to be balanced, and not be killed by high hashrate and free electricity botnets or some ASICS/FPGAs, so the network would be quite decentralized again
RandomX would make sense, if there couldn't be any pool, and any miner would be FORCED to run full node, as for example on Nerva network
Feel free to join the discussion