Author Topic: Main discussion thread (similar to BCT)  (Read 12099 times)

georgem

  • Tech Admin
  • ******
  • Posts: 842
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #105 on: January 14, 2017, 08:33:15 pm »
@georgem - When you say you are going to work on a new miner for Spreadcoin - do you mean the algorithm will change or you will make the miner better than what is available now?  I am curious because I use the custom one from SP however he hasn't updated it in quite a long time.  Just curious is all.  Thanks - pokeytex

Still working on the specifications for the spreadminer.
If you want to get a general idea about what it will consist of, just look at the spreadwallet, and what it tries to achieve:

1) customizability
2) capable to work with multiple different coins
3) serves as testbed for servicenode research

etc...

Basically the same goals will apply for the spreadminer, but adjusted for miner terminology (customizable hash algorithms, handling of multiple GPUs (with different architecture), setup-help (auto best parameter find), shareble configuration-files, etc..).  8)

Spreadcoin algorithm will 100% NOT CHANGE in the foreseable future, unless protocol itself changes because of something that we will add that is needed for servicenodes to work.

... or you will make the miner better than what is available now?

Oh god yes 1000% better, it will be a much better experience overall than what people have to put up with these days when they run a miner.  8)
Plus I get to have my research testbed, because GPUs will be at the center of everything, and very important for both Spreadcoin and Servicenodes.

georgem

  • Tech Admin
  • ******
  • Posts: 842
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #106 on: January 14, 2017, 08:38:17 pm »
Here's a quick screenshot/appetizer about one of the few current explorer functions that I will demonstrate in the next update video:



New video will be ready soon, I just need to finish a few functions and test them.

Thanks for your patience, dear patrons!  8)

minerpage

  • Admin
  • *****
  • Posts: 112
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #107 on: January 15, 2017, 01:49:23 am »
2) capable to work with multiple different coins

Multiple "different" coins as in merge mining SpreadX11 coins?

Different coins as in how different? What will be different?

Love to hear  8)

georgem

  • Tech Admin
  • ******
  • Posts: 842
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #108 on: January 15, 2017, 02:36:25 am »
Multiple "different" coins as in merge mining SpreadX11 coins?

Well... every AUX-POW enabled coin that is merge-mineable with any other coin (not just SpreadX11) will work within the spreadminer...
Only condition is that they both use the same hash-algo (and that one is a child-coin and the other the parent-coin)
... but that's not what I am aiming at here.
It goes a little bit farther than just merge-mining.  ;)

Different coins as in how different? What will be different?

Love to hear  8)

Well their hashing algorithm is different.
Or more specific, how they combine and order different hash functions (like DSHA-256, Keccak, Blake, Groestl, Skein, etc..) to search for a hash solution is different.

For example:

  • Bitcoin uses simple DSHA-256 on the blockheader data
  • Dash uses X11, which consists of 11 different such hashing functions in a specific order.
  • SpreadX11 uses that same hash functions in the same order, but adds some additional processing on top of everything (padding, hashing whole block, signing etc..)

So when I say "customizable" I mean that the spreadminer will be able to mine any coin (not just SpreadX11 or its children) that can be specified in a clear determined way, using an internal representative "language".
So there will be a graphical (and textual) representation of the hash-algo-CIRCUIT (if you will) that people (or devs rather) will be able to adjust.
You will be able to put together your own hashing algos like you would put together lego-blocks so to speak.

What this will do is help us create a tool that we and others can use, where we can change/update a hashing-mechanism by merely adjusting config and settings-files.
So there will be no need to recreate a miner from scratch (recompiling, building, etc.) each time there is a change in a coins algorithm.

Except maybe compiling the actual hash function kernels that are used by GPU... (I'm still researching and experimenting with this)

I will explain in more detail when I am finished with the specs,....

njs811

  • Jr. Member
  • **
  • Posts: 88
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #109 on: January 15, 2017, 10:07:57 am »
Though this is still something a bit way off, I'm curious as to your position on building dapps.  This is something I have become more interested in recently and would love to look into.

georgem

  • Tech Admin
  • ******
  • Posts: 842
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #110 on: January 15, 2017, 12:55:35 pm »
Though this is still something a bit way off, I'm curious as to your position on building dapps.  This is something I have become more interested in recently and would love to look into.

That is not way off at all, but rather spot on.

How to efficiently research decentralized/distributed apps is the third field of interest right after spreadwallet and spreadminer.

Creating P2P software is hard on many levels:

It's difficult to test, since you would need many nodes.
It's difficult to debug, since you would need to record all communication between nodes (and track the internal states of all nodes) to figure out where something went wrong.

I have done most of my P2P Experiments with just 5-6 computers and older laptops I have lying around.
This just doesn't cut it anymore, so I am working on a cheap but (much) more large-scale solution.



This is what my other announcement will be about.
 8)

Can you guess what the solution will be about?
Lots and lots of raspberries or lots and lots of hamsters?

pokeytex

  • Jr. Member
  • **
  • Posts: 77
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #111 on: January 15, 2017, 05:34:29 pm »
Though this is still something a bit way off, I'm curious as to your position on building dapps.  This is something I have become more interested in recently and would love to look into.

That is not way off at all, but rather spot on.

How to efficiently research decentralized/distributed apps is the third field of interest right after spreadwallet and spreadminer.

Creating P2P software is hard on many levels:

It's difficult to test, since you would need many nodes.
It's difficult to debug, since you would need to record all communication between nodes (and track the internal states of all nodes) to figure out where something went wrong.

I have done most of my P2P Experiments with just 5-6 computers and older laptops I have lying around.
This just doesn't cut it anymore, so I am working on a cheap but (much) more large-scale solution.



This is what my other announcement will be about.
 8)

Can you guess what the solution will be about?
Lots and lots of raspberries or lots and lots of hamsters?

Hrmmm- methinks - "clusters"!  ;D

pokeytex

  • Jr. Member
  • **
  • Posts: 77
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #112 on: January 16, 2017, 01:31:19 am »
Though this is still something a bit way off, I'm curious as to your position on building dapps.  This is something I have become more interested in recently and would love to look into.

That is not way off at all, but rather spot on.

How to efficiently research decentralized/distributed apps is the third field of interest right after spreadwallet and spreadminer.

Creating P2P software is hard on many levels:

It's difficult to test, since you would need many nodes.
It's difficult to debug, since you would need to record all communication between nodes (and track the internal states of all nodes) to figure out where something went wrong.

I have done most of my P2P Experiments with just 5-6 computers and older laptops I have lying around.
This just doesn't cut it anymore, so I am working on a cheap but (much) more large-scale solution.



This is what my other announcement will be about.
 8)

Can you guess what the solution will be about?
Lots and lots of raspberries or lots and lots of hamsters?

Hrmmm- methinks - "clusters"!  ;D

I keep toying with the idea of building a (10) RPi 3 Cluster but don't know what the hell I will do with it when it is done! LOL - I am not a programmer.  I see all of the tutorials online about building them and honestly the only thing I can think of is if there was a way to mine with it.  Other than that I don't know.  I just think to build it would be really cool!  I am a geek for hardware!

georgem

  • Tech Admin
  • ******
  • Posts: 842
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #113 on: January 16, 2017, 11:58:33 am »
Yes, I'm gonna build a Raspberry Pi 3 Cluster to bring my overlay and P2P research to the next level.
I'll probably start with a 15-16 node Cluster, to make a few tests.
And then move to much higher numbers (64, then 100?).  8)

It would be fantastic if I'm not the only one building and running such a cluster,
so I will keep you updated with any cluster-software I create so you can test it yourself.
Especially if you are a patron.

I keep toying with the idea of building a (10) RPi 3 Cluster but don't know what the hell I will do with it when it is done! LOL - I am not a programmer.  I see all of the tutorials online about building them and honestly the only thing I can think of is if there was a way to mine with it.  Other than that I don't know.  I just think to build it would be really cool!  I am a geek for hardware!

Well, one of the things you could always do with your RPi 3 Cluster is use it with the spreadwallet, and manage your 10 favorite coins with it. (1 full node per RPi 3).  8)

BTW, using RPi 3s for full nodes is just one of the tasks they could fullfill, but there are other ideas I have of how to put those "tiny 24/7 workers" to good use:

1) Exchange-Tick-Data Aggregator, so that you recreate a coinmarketcap-style site within the spreadwallet (and better yet: have access to historical tick-data if you let your worker run long enough)
2) Trading-Bot so you can be alerted or trade directly within the spreadwallet
3) Address-Observer, get alerted when there is movement in a coin address...

etc

Maybe you could explore a few ideas of your own? What jobs would you give your oompa loompas?

I'm currently not spending any time on this apect, but I'm just saying I design the spreadwallet's architecture as an open system, where other devs can add functionality in the form of plug-ins/dlls (each representing a worker-functionality and its overview screen)

minerpage

  • Admin
  • *****
  • Posts: 112
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #114 on: January 16, 2017, 04:22:38 pm »
It would be fantastic if I'm not the only one building and running such a cluster,
so I will keep you updated with any cluster-software I create so you can test it yourself.

Would love to setup something like this

use it with the spreadwallet

Excellent setup, let me know if you need testers to help  8)

pokeytex

  • Jr. Member
  • **
  • Posts: 77
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #115 on: January 17, 2017, 12:13:24 am »
Yes, I'm gonna build a Raspberry Pi 3 Cluster to bring my overlay and P2P research to the next level.
I'll probably start with a 15-16 node Cluster, to make a few tests.
And then move to much higher numbers (64, then 100?).  8)

It would be fantastic if I'm not the only one building and running such a cluster,
so I will keep you updated with any cluster-software I create so you can test it yourself.
Especially if you are a patron.

I keep toying with the idea of building a (10) RPi 3 Cluster but don't know what the hell I will do with it when it is done! LOL - I am not a programmer.  I see all of the tutorials online about building them and honestly the only thing I can think of is if there was a way to mine with it.  Other than that I don't know.  I just think to build it would be really cool!  I am a geek for hardware!

Well, one of the things you could always do with your RPi 3 Cluster is use it with the spreadwallet, and manage your 10 favorite coins with it. (1 full node per RPi 3).  8)

BTW, using RPi 3s for full nodes is just one of the tasks they could fullfill, but there are other ideas I have of how to put those "tiny 24/7 workers" to good use:

1) Exchange-Tick-Data Aggregator, so that you recreate a coinmarketcap-style site within the spreadwallet (and better yet: have access to historical tick-data if you let your worker run long enough)
2) Trading-Bot so you can be alerted or trade directly within the spreadwallet
3) Address-Observer, get alerted when there is movement in a coin address...

etc

Maybe you could explore a few ideas of your own? What jobs would you give your oompa loompas?

I'm currently not spending any time on this apect, but I'm just saying I design the spreadwallet's architecture as an open system, where other devs can add functionality in the form of plug-ins/dlls (each representing a worker-functionality and its overview screen)

@georgem - I will DEFINITELY support the Spreadcoin network with a cluster.  I am more than happy to do so.  I like the idea of running nodes on the Pi's and will do that when the time is right.  I would love to create a cluster that can almost act like "Siri" or "Alexa", on a better scale, and have a central AI in the house.  Ask all sorts of hard questions that can go to the web and answer me.  Or better yet create their own programs for tasks I want done.  I know that is some "Tony Stark" kinda tech but hey one can dream!  ;D

edit - let me know when to order the parts and I will.

pokeytex

  • Jr. Member
  • **
  • Posts: 77
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #116 on: January 21, 2017, 01:56:16 am »
Though this is still something a bit way off, I'm curious as to your position on building dapps.  This is something I have become more interested in recently and would love to look into.

That is not way off at all, but rather spot on.

How to efficiently research decentralized/distributed apps is the third field of interest right after spreadwallet and spreadminer.

Creating P2P software is hard on many levels:

It's difficult to test, since you would need many nodes.
It's difficult to debug, since you would need to record all communication between nodes (and track the internal states of all nodes) to figure out where something went wrong.

I have done most of my P2P Experiments with just 5-6 computers and older laptops I have lying around.
This just doesn't cut it anymore, so I am working on a cheap but (much) more large-scale solution.



This is what my other announcement will be about.
 8)

Can you guess what the solution will be about?
Lots and lots of raspberries or lots and lots of hamsters?

Hang tight on the Rpi3's - I read that we can expect the Rpi4 end of February - of course just gossip - but where there's smoke there's fire!

http://www.itpro.co.uk/desktop-hardware/27763/raspberry-pi-4-what-to-expect

georgem

  • Tech Admin
  • ******
  • Posts: 842
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #117 on: January 21, 2017, 02:48:16 pm »
Hang tight on the Rpi3's - I read that we can expect the Rpi4 end of February - of course just gossip - but where there's smoke there's fire!

http://www.itpro.co.uk/desktop-hardware/27763/raspberry-pi-4-what-to-expect

I heard such rumours a few weeks ago.
If the next Raspberry Pi comes with 2 GByte RAM, this will be awesome.

But I need to start somewhere.
What I will do is what I have always been doing: don't wait for the perfect product, instead buy the hardware now, treat it well, keep all the original packing and wrapping, so you can sell it for a nice enough price on ebay later, then use this money to subsidise the newer model.

Atleast the switches, power supplies, cables (USB + Patch) and the whole casing you will not change very often.
But the raspberry units should be exchangeable quickly.

minerpage

  • Admin
  • *****
  • Posts: 112
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #118 on: January 21, 2017, 04:40:27 pm »
Atleast the switches, power supplies, cables (USB + Patch) and the whole casing you will not change very often.

Pi cluster boards (eBay), USB hub, cable ducts and KVM.

Unplug the main boards when there's a new version and swap.

Easy setup for a dozen of these Raspberries 8)


georgem

  • Tech Admin
  • ******
  • Posts: 842
    • View Profile
Re: Main discussion thread (similar to BCT)
« Reply #119 on: January 23, 2017, 01:11:43 am »
Finally, here's a short new update video about the current state of spreadwallet development.

I demonstrate the general behaviour of the explorer tabs, and a few of the commands that you can currently use.
You can type the block_nr, with or without coin name. Currently it will just show basic block-header-data. (More to come soon)
You can do a getaddr to ask a full node to inform you about the peers it knows (or has seen in the past. It will list the 100 most recent ones)
You can do a ping on any IP.
You can do a portcheck on an IP that will tell you which of the coins in your coin-list is running on that IP.

Stay tuned for more.  :)



https://vimeo.com/200603212

Still working on the podcast, don't worry.  ;)