search

pRizz / iota-transaction-spammer-webapp

https://prizz.github.io/iota-transaction-spammer-webapp/
MIT License
38 stars 13 forks source link

not using all six GPU's on my rig... #13

Open NateInSav opened 6 years ago

NateInSav commented 6 years ago

Great job and project to support the IOTA initiative!

I want to use this software on what had been a dedicated mining rig. I have downloaded and run the standalone installation. Not sure why, but it is only using one of the six GPU's for POW.

The power to run the rig is about 1300W so I would like the IOTA network to get the most out of the expenditure on my part.

How can we fix this?

pRizz commented 6 years ago

Does it still only use one GPU if you open 6 windows of this webapp?

It is possible that this could be an issue with the PoW dependecy, https://github.com/iotaledger/curl.lib.js Might need some more investigation or implement parallelism in that library.

NateInSav commented 6 years ago

First off great job on this project.

The software is using only one GPU of six available.  It appears that only one window can be active at a time and must be the top browser window (Chrome).  I tried multiple windows - did not work.

Also, the single GPU that is running is running at about 40% power which means it could probably use a higher intensity setting (more concurrent threads).  I typically run other miners at about 85% power level...

Would be GREAT for Iota if you could get this working with multiple GPU based systems!

Also  over on IOTA slack - I introduced the following idea:

Nate Nivens [5:52 PM] On transaction spamming as a way to increase tangle network confirmation speed... as spam transactions are added to the tangle - is it not a waste of PoW to have processing time used to confirm the spam transactions? If a tag or special message was embedded in the spam transaction which indicates no confirmation required, then couldn't the PoW be better applied to real world transactions instead? The thought is - what if my spam PoW is used to confirm other spam transactions - seems like wasted resources to me. Thoughts? Could this resource saving strategy somehow be implemented in the tangle? Is it already?

Malte Wedel [5:55 PM] Sounds like an interesting idea.... a spammer which does not create zero value transactions, but instead reattaches random valid transactions which have not been confirmed yet.

Anyway this could be implemented in your app?

Thanks,

Nate

On Sep 17, 2017, at 11:51 AM, Peter Ryszkiewicz notifications@github.com wrote:

Does it still only use one GPU if you open 6 windows of this webapp? It is possible that this could be an issue with the PoW dependecy, https://github.com/iotaledger/curl.lib.js Might need some more investigation or implement parallelism in that library. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

pRizz commented 6 years ago

Thanks for your compliment!

I believe confirming spam transactions is actually helpful to the network. The way the tangle works is that the more "descendents" a transaction has, the faster it will be confirmed. For example, let's say you have (valuable transaction) -> (spam transaction) -> (spam transaction), then the valuable transaction is actually better off, as it has more "weight"after it, thus confirming it faster. Of course, the valuable transaction could directly have more confirmations, but I think this doesn't work well in the tangle data structure and selection algorithms.

Also, this spammer has no say in what transactions to confirm; that is decided by the IRI, which the spammer then goes and performs the confirmations on; so that change would have to be done in the IRI.