Author

Topic: To cryptonight-xmr and progpow developers how about a new algo mash up? (Read 234 times)

legendary
Activity: 4116
Merit: 7849
'The right to privacy matters'
One flip flop switch is probably not enough to stop ASIC makers. They will make new optimized chip for both algos combined. The new algo has to change itself based on random inputs from blockchain data. It has to be even tough to run on GPUs. In this case ASIC makers will have to recreate GPU functionalities in their ASICs. The new algo has to be highly dynamic but at the same time provide stable hashing speeds and security. GPUs were originally made for highly parallel complex computations such as computer graphics.

Some current algos such as ethash rely on limitations of memory bandwidth. It looks like it is still not enough to fight ASICs.

Well since tx hash is hex coded and there are 16 symbols 4 algo do the trick.

1 can be the current eth algo with a tweak (bigger dag say 8gb)
1 can be progpow
1 can be CNv2
1 can be the new ebwf algo with a tweak (more ram say 8gb)

that asic would be pretty wastefull

it would not be 1 chip  it would be 4 chips with a switch  so only ¼ is used at a time

And the first of each month a gpu miner simply switches.


the game is race to the bottom with out gpu's making some money.

and gpus make for millions of adapters of coins.  since millions of gpus  are used for other shit
jr. member
Activity: 47
Merit: 1
One flip flop switch is probably not enough to stop ASIC makers. They will make new optimized chip for both algos combined. The new algo has to change itself based on random inputs from blockchain data. It has to be even tough to run on GPUs. In this case ASIC makers will have to recreate GPU functionalities in their ASICs. The new algo has to be highly dynamic but at the same time provide stable hashing speeds and security. GPUs were originally made for highly parallel complex computations such as computer graphics.

Some current algos such as ethash rely on limitations of memory bandwidth. It looks like it is still not enough to fight ASICs.
legendary
Activity: 4116
Merit: 7849
'The right to privacy matters'

man pure genius, I was thinking about this but had no good way to articulate it

forking every much is too much but i would like to see cn7 implement some of the progpow calulations to
reduce the power of fpgas


https://github.com/monero-project/monero/pull/4218#issuecomment-425971822

thats the link on the cryptonite fork discussion

Seems to me  that  a monthly  flip flop or not flip flop  based on the hash of last block of the month.  Would be a P.I.T.A.  for all miners

but 
GPU--- Bad
FPGA-- Worse
Asic---- Worst


You need 2 asic chips  one for progpow and one for CNv2    so it would hurt asics  2x
You need to flash your fpga  and would hurt  it somewhat as fpga flashing is not super easy
You need to switch  from  one program to another on your gpus takes under 1 minute per gpu

My developing skills are small  but I am a good idea guy. 

Anything to hurt asics should be good.
And  this idea could be expanded.

If you have 4 hard algos    and hash tx for a block is hexi- decimal   

   4 numbers for algo 1
  4 numbers for algo 2
 4 numbers for algo 3
4 numbers for algo 4

that  would really hurt asic mining big time  as you need 4 separate chips  one for each algo
full member
Activity: 1148
Merit: 132

man pure genius, I was thinking about this but had no good way to articulate it

forking every much is too much but i wouls line to see cn7 implement some of the progpow calulations to
reuce the power of fpgas


https://github.com/monero-project/monero/pull/4218#issuecomment-425971822

thats the link on the cryptonite fork discussion
legendary
Activity: 4116
Merit: 7849
'The right to privacy matters'
Okay progpow seems to work okay to prevent or slow asics.

And CNv2 will be coming out.

So how about a mashup algo.

Use both algos and on the last day of every month  check to last block made for it tx hash and flip back and forth the two algos.

So a gpu only needs to switch from one software to the next.

Every asic needs two machine one for cn2 algo the other for progpow algo.


And every FPGA would need a flashing to switch.


How hard would it be to do this?

I realize it could cause a stability issue in the chain but it seems doable.
Jump to: