Jump to content


Photo

New MNC Algo


  • Please log in to reply
36 replies to this topic

#21 -Jt

-Jt

    MNC Dev Team

  • Moderators
  • 140 posts

Posted 10 October 2013 - 11:38 PM

I will just have to write an FPGA for whatever algorithm we choose then :)

 

But what is the point of changing it? do we just change the algorithm once we decide there are too many people mining it?

Thats a good question, I'm not sure.

 

Where's that guy RS when you need him? I know he likes MNC that bastard  :P

Where is he, he should be in here...


MNC Donations Accepted: MEfCq4U68vBaXP2B9ACAKaHAfRueYxDQhE

#22 sdf

sdf

    MNC Deserter

  • Validating
  • PipPipPipPipPipPipPipPip
  • 158 posts
  • LocationCurrently USA

Posted 11 October 2013 - 01:36 AM

Thats a good question, I'm not sure.

 

Where's that guy RS when you need him? I know he likes MNC that bastard   :P

Where is he, he should be in here...

 

I wonder how he sees MNC and MC coexisting.



#23 SuperTramp

SuperTramp

    The Boss

  • Administrators
  • 577 posts
  • Location: Planet MinCoin

Posted 11 October 2013 - 01:44 AM

I wonder how he sees MNC and MC coexisting.

 

I believe he sees this as MNC = Crypto-Commodity And MC = Crypto-Currency :)


MinCoin, The #1 Crypto-Commodity


#24 sdf

sdf

    MNC Deserter

  • Validating
  • PipPipPipPipPipPipPipPip
  • 158 posts
  • LocationCurrently USA

Posted 13 October 2013 - 01:43 PM

If we switch the algorithm, it MUST be memory hard. It MUST have a GPU miner that launches with it. Until you find those things, it should remain on scrypt.

 

This is because of two things: memory hard algorithms are extremely costly to put onto an ASIC. It is not impossible to find an FPGA with the required qdrii RAM, and it isn't very costly (10-15k USD total). But an ASIC remains many years away for scrypt. We don't want to have another Sha256 arms race if we choose a similarly "non-memory-hard" algorithm.

 

Secondly, people must be able to easily mine your coin, whatever it is. Eventually we will have many transactions occurring on our block chain. We will need miners to power the network. If we have to sit around and wait for transactions to process, it will likely be the death of whatever reputation MNC has.



#25 necom

necom

    Resident Hacker

  • Moderators
  • 222 posts
  • LocationAround the corner

Posted 14 October 2013 - 09:34 PM

I've been really looking forward to a change to this algo recently.  I am really tiring of the list of all the coins blending together into a mess and I'm on board with the notion that an algo change will keep MNC on the top of lists of innovative coins.

 

I also am fairly sure that Reaper is just flat-out more efficient than other software.  I know temperatures are noticeably lower and it seems that the speeds are solid compared to other software.  Reaper v13 is pretty spiffy and hopefully a v14+ with even more tweaks could come out with all this. It would be cool to have more options for tweaking, but overall, it's a working app that is pretty straight-forward to configure.

 

There are plenty of GPU miners out there who will love to have a good coin to keep mining.



#26 sdf

sdf

    MNC Deserter

  • Validating
  • PipPipPipPipPipPipPipPip
  • 158 posts
  • LocationCurrently USA

Posted 14 October 2013 - 10:39 PM

I've been really looking forward to a change to this algo recently.  I am really tiring of the list of all the coins blending together into a mess and I'm on board with the notion that an algo change will keep MNC on the top of lists of innovative coins.

 

I also am fairly sure that Reaper is just flat-out more efficient than other software.  I know temperatures are noticeably lower and it seems that the speeds are solid compared to other software.  Reaper v13 is pretty spiffy and hopefully a v14+ with even more tweaks could come out with all this. It would be cool to have more options for tweaking, but overall, it's a working app that is pretty straight-forward to configure.

 

There are plenty of GPU miners out there who will love to have a good coin to keep mining.

 

What algo do you suggest? I would be comfortable if we decided on one and then made a GUI, guide, one-click type launch for it all. We cannot expect people to migrate over randomly.

 

EDIT: I would also suggest making the new algo have a purpose, like XPM's does. Rendering videos, audio, global warming models- something other than arbitrarily hard algos that weren't meant to have hundreds of people mining them ;)



#27 5Dzz

5Dzz

    MinMember

  • Members
  • PipPip
  • 15 posts

Posted 23 October 2013 - 04:39 AM

Not sure about what specific enhancements to suggest being added to Mincoin. If they are valuable and people like them I'm all for it.  :)



#28 raspi

raspi

    MinMember

  • Members
  • PipPip
  • 27 posts

Posted 25 October 2013 - 02:19 PM

I will just have to write an FPGA for whatever algorithm we choose then :)

 

But what is the point of changing it? do we just change the algorithm once we decide there are too many people mining it?

 

The point is to avoid multi-pool mining that just switch in and out creating a high difficulties for loyal miners and cutting away the profits from the loyal miners when the difficulties drops.

 

 

I am actually more supportive of CPU mining than GPU or even FPGA mining, as CPU mining would means every single person in the world will be able to mine it.

 

Not everyone has a GPU and not everyone can afford a $8K-$10K FPGA just to mine some coins for a beer gathering or for fun or support.


  • alttin likes this

#29 ropyu1978

ropyu1978

    Advanced MinCoiner

  • Members
  • PipPipPip
  • 32 posts

Posted 29 October 2013 - 10:58 PM

i like it



#30 Scrat

Scrat

    MinMember

  • Members
  • PipPip
  • 28 posts

Posted 30 October 2013 - 06:40 PM

Interesting, MNC perspectives is growing up!



#31 Razzler

Razzler

    MinMember

  • Members
  • PipPip
  • 27 posts

Posted 30 October 2013 - 10:41 PM

RS is cryptopresent.



#32 alttin

alttin

    MinMember

  • Members
  • PipPip
  • 29 posts

Posted 31 October 2013 - 04:26 PM

RS is cryptopresent.

 

I'm scared...



#33 Lishoplays

Lishoplays

    MinMember

  • Members
  • PipPip
  • 27 posts

Posted 01 November 2013 - 08:48 AM

MNC is bombing!


#34 alttin

alttin

    MinMember

  • Members
  • PipPip
  • 29 posts

Posted 01 November 2013 - 07:00 PM

The point is to avoid multi-pool mining that just switch in and out creating a high difficulties for loyal miners and cutting away the profits from the loyal miners when the difficulties drops.

I agree this could be a major ongoing issue for mincoin as well as most other altcoins.

 

I am actually more supportive of CPU mining than GPU or even FPGA mining, as CPU mining would means every single person in the world will be able to mine it.

 

Not everyone has a GPU and not everyone can afford a $8K-$10K FPGA just to mine some coins for a beer gathering or for fun or support.

Hmm, is it even theoretically possible to have an algorithm that would be optimized for CPU mining, but resistant to GPU, FPGA and ASIC mining? My guess is no. That is, for any computable algorithm, I expect that CPUs will underperform against specialized hardware that is developed to be optimized for nothing else but processing that algorithm.



#35 ssj4mo

ssj4mo

    Advanced MinCoiner

  • Members
  • PipPipPip
  • 31 posts
  • LocationNew York

Posted 01 November 2013 - 09:56 PM

Well, yea.  Nothing will perform better than an ASIC...because it's application specific.  But it is entirely possible to write an algorithm that is optimized for CPU; that is, you could easily write an algorithm with which a $300 CPU would drastically out-perform a $1,000 GPU.  It depends entirely on the math behind the algorithm.  Processors all do the same things essentially, so why do CPUs and GPUs perform differently for different tasks?  It's because of their design.  CPUs are designed to perform maths, which is what allows software to run on computers.  GPUs perform maths as well, but with a focus on graphics aka vectors and rendering.

 

 

I would be interested, however, to learn more about this "proposed" algorithm.


  • alttin likes this

#36 794259332

794259332

    MinMember

  • Members
  • PipPip
  • 28 posts

Posted 01 November 2013 - 10:39 PM

very cool,just do it



#37 Chicago

Chicago

    MinCoin Enthusiast

  • Members
  • PipPipPip
  • 35 posts
  • LocationFEMA Region V

Posted 20 June 2017 - 08:03 PM

Hello,

 

    Since the Mincoin distribution curve is flat and there are as many rewards per minute now as there were in the beginning and will be near the end, there are benefits for non-early adopters who want to start mining and still obtain a decent amount of MNC.  (ignoring the first 4,320 blocks for argument's sake)

 

    I don't necessarily have a particular favorite to propose in regards to a CPU algorithm with characteristics to make it GPU / ASIC resistant but I am inclined to think there must be some particular hardware extensions on the various newer AMD / Intel chips which could be featured in a proof-of-work algorithm where plain chips see a hash rate of x +/- 35% (due to clock frequency) and the good chips see a hash rate of somewhere in the ballpark of 3x.

 

    For comparison, think processors from the late 90s encoding an MP3 before MMX extensions (with integer math optimizations) or video encoding before the various floating-point SSE extensions improved the throughput dramatically.

 

    Since I haven't looked too deeply at potential candidate algorithms, I am open to discussion and listening to the thoughts from others about algorithm change related suggestions.

    More than likely a "good" candidate will already be included in pooler's or tpruvot's cpuminer and have seen some action in prime time on a relatively secure chain.

 

    At the same time -- people have their Scrypt ASICs from 2014 and might want to have a chance to find profitability with them in the not so distant future on the Mincoin chain.

 

    Is the impetus to change the proof-of-work algorithm to secure the network or to level the playing field with one cpu one vote?

    Or for the sake of argument, does one cpu one vote actually achieve the goal of better securing the network? (I think it does very much.)

 

Best Regards,

-Chicago






0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users