Show Posts
Edwin [new.capital]:
The CPU issue has not been fixed and new version for TWINS testnet is coming shortly with another attempt to fix it. @kemi can share more details about the implementation.
kemi:
Now, to shed some light on what has been done and how this is different from the previous attempts:
1. Muti-tiered masternode reward ratios are already correct and the only remaining issue is high CPU usage
2. PIVX has solved the issue of high CPU usage by increasing minimum masternode requirements to 2GB of RAM and 50GB of storage space, however, most TWINS and FIX masternode service providers offer only 1GB of RAM and 25GB of storage space, what makes it more complicated. This update exchange some computation heavy processes for storage space, meaning that instead of making a lot of computations and, thus, using a lot of CPU power, masternode will save more data in their memory. This should significantly decrease the CPU power usage and even speed up the masternodes
Now, what are those computation-heavy processes:
For every block, there are 10 masternodes selected randomly to choose a winner of a block. I will refer to these masternodes as deciders from now on. At first, these blocks have to form a queue, according to masternode payment times (long story short, those masternodes which have been rewarded a long time ago are put in the front of the queue and those which have just been rewarded - in the end of the queue [in reality, this is a more complicated process, but it isn't the main focus now]). In order to do this, for every single masternode in the network, the deciders have to find the latest block where particular masternodes have been selected as winners. This can mean iterating through more than thousand of blocks again, again and again. This process requires a lot of operations and lead to increased CPU usage.
Now, instead of iterating through the blocks, the deciders will simply have to read their memory and that's it. This should solve the high CPU usage issue
https://discord.gg/qfuNu2w
